what do you think of these bricks?
Reaching, Tue Aug 01 2006, 10:36AMa few days ago i came across a bunch of 9 igbt bricks from fuji 1MBI400F-060
datasheet:
i got them very cheap for around 5$ for one and the ratings are very good, 600volt 400A continious and 800A peak and relatively fast though, gate capacitance of around 22nF.. so now i wanted to build a real monster drsstc out of them and wanted to know what do you guys think ? what is the best way to drive these bricks, i wanted to use a fullbridge running of +30 - 15volts with a 1:1 gdt for each igbt, should work ? the secondary frequency of the coil will be around 60-80khz .so what do you think?
Re: what do you think of these bricks?
teravolt, Wed Aug 02 2006, 03:07AM
they will work. go to Steve Ward's web site and find the link for Richie Burrnet's link for the theory and any of the senior members links for more stuff. just read. thats how I figured it out
teravolt, Wed Aug 02 2006, 03:07AM
they will work. go to Steve Ward's web site and find the link for Richie Burrnet's link for the theory and any of the senior members links for more stuff. just read. thats how I figured it out
Re: what do you think of these bricks?
Marko, Wed Aug 02 2006, 12:30PM
I think he just wanted to pimp out some of his loved bricks
The datasheet you put there didn't work for me tough, this is from catalog
gate capacitance seems to be 38nf, and switching times
0.8
0.6
1.5
1.0
microseconds, probably could be used on lowest frequency you can manage to do (40, 50kHz).
Marko, Wed Aug 02 2006, 12:30PM
I think he just wanted to pimp out some of his loved bricks
The datasheet you put there didn't work for me tough, this is from catalog
gate capacitance seems to be 38nf, and switching times
0.8
0.6
1.5
1.0
microseconds, probably could be used on lowest frequency you can manage to do (40, 50kHz).
Re: what do you think of these bricks?
Reaching, Wed Aug 02 2006, 03:46PM
not micro seconds, y seconds ,, and 40khz seems very low though, i tested one of them okay up to 120khz, with 140khz the current draw increases and the squarewave is distorted but under 100khz everything is perfect , and that was with a single 9A gate driver ic on 12 volt, the quality of the signal will be much better on a drive fullbridge at -15 + 30volt (already build one for my old bricks)
Reaching, Wed Aug 02 2006, 03:46PM
not micro seconds, y seconds ,, and 40khz seems very low though, i tested one of them okay up to 120khz, with 140khz the current draw increases and the squarewave is distorted but under 100khz everything is perfect , and that was with a single 9A gate driver ic on 12 volt, the quality of the signal will be much better on a drive fullbridge at -15 + 30volt (already build one for my old bricks)Re: what do you think of these bricks?
Steve Ward, Wed Aug 02 2006, 06:31PM
Its hard to really know for sure what the switching rate is, it depends on the load characteristics. But, i think the *delay* times are the most crucial thing, and cant really be fixed. In your case, you will probably need a way to generate a little deadtime, so that you dont get shoot-through. I also would suggest operating at 40khz or even *less*. I run considerably faster devices at 40khz, and the delays are very significant, meaning i have to switch about 1/6th of the peak primary current, or about 250A in my case. With slower bricks, the problem just gets worse and worse. When you hard switch more current, you have to deal with larger, more energetic voltage spikes.
Steve Ward, Wed Aug 02 2006, 06:31PM
Its hard to really know for sure what the switching rate is, it depends on the load characteristics. But, i think the *delay* times are the most crucial thing, and cant really be fixed. In your case, you will probably need a way to generate a little deadtime, so that you dont get shoot-through. I also would suggest operating at 40khz or even *less*. I run considerably faster devices at 40khz, and the delays are very significant, meaning i have to switch about 1/6th of the peak primary current, or about 250A in my case. With slower bricks, the problem just gets worse and worse. When you hard switch more current, you have to deal with larger, more energetic voltage spikes.
Re: what do you think of these bricks?
Reaching, Wed Aug 02 2006, 08:25PM
mhh, why so pessimistic? you build a drsstc with bricks, many other people done it and i already build a sstc with some older bricks and with a much higher frequency than 40khz. okay, it´ll be a big drsstc but i definately cant get it down to 40khz, and to tune the primary circuit to 40khz when the secondary fres is around 60khz (which is in a possible level) is senseless.. and really, voltage spikes can be a problem, but with some surpressor diodes and big decoupling caps that shouldnt be a problem, delay is the problem, but can it be really dangerous to 800A peak Bricks,okay, when the delay is too long it can result in shoot through but i tested one igbt up to 140 khz and below 100khz i get a perfect signal even with a load, so i dont think it´ll be that tricky to build a good fullbridge out of them for frequencies of around 60-80khz .. i´ve build several drsstcs and never had any problems with the delay times even when i ran 100khz igbts with around 300khz (with my mini drsstc5)
okay, first tests will show what i can make with these bricks, i can make failures, they were cheap and i have 9 of them
Reaching, Wed Aug 02 2006, 08:25PM
mhh, why so pessimistic? you build a drsstc with bricks, many other people done it and i already build a sstc with some older bricks and with a much higher frequency than 40khz. okay, it´ll be a big drsstc but i definately cant get it down to 40khz, and to tune the primary circuit to 40khz when the secondary fres is around 60khz (which is in a possible level) is senseless.. and really, voltage spikes can be a problem, but with some surpressor diodes and big decoupling caps that shouldnt be a problem, delay is the problem, but can it be really dangerous to 800A peak Bricks,okay, when the delay is too long it can result in shoot through but i tested one igbt up to 140 khz and below 100khz i get a perfect signal even with a load, so i dont think it´ll be that tricky to build a good fullbridge out of them for frequencies of around 60-80khz .. i´ve build several drsstcs and never had any problems with the delay times even when i ran 100khz igbts with around 300khz (with my mini drsstc5)
okay, first tests will show what i can make with these bricks, i can make failures, they were cheap and i have 9 of them
Re: what do you think of these bricks?
Steve Ward, Wed Aug 02 2006, 08:42PM
I agree, i am being a bit pessimistic.
But if the inductance leading to the diodes and decoupling caps is too great (>50nH), the spikes will still be present. This is the headache ive been going through when working with these older bricks, they have too much internal inductance, which means they are meant to be switched sloooooowly to prevent huge voltage spikes from developing across the CE junction. Its a whole new beast compared to the new, very fast, very low inductance IGBTs, in which case, you can usually exceed the ratings with little worry.
Steve Ward, Wed Aug 02 2006, 08:42PM
I agree, i am being a bit pessimistic.
and really, voltage spikes can be a problem, but with some surpressor diodes and big decoupling caps that shouldnt be a problem,
But if the inductance leading to the diodes and decoupling caps is too great (>50nH), the spikes will still be present. This is the headache ive been going through when working with these older bricks, they have too much internal inductance, which means they are meant to be switched sloooooowly to prevent huge voltage spikes from developing across the CE junction. Its a whole new beast compared to the new, very fast, very low inductance IGBTs, in which case, you can usually exceed the ratings with little worry.
Re: what do you think of these bricks?
Reaching, Thu Aug 03 2006, 03:35PM
hehe, just got 2 other bricks from ebay dirt cheap for around 5$ for both, they are halfbridge modules rated at 1200 volts 210A continous and 420A pulsed, and are much faster than the other big fuji igbts, maybe good for a "smaller" drsstc with a voltage doubler,and with only 11nF gate capacitance, means you can run a fullbridge of them with normal gdts and 2 pairs of driver ics at around 100khz
. heres the datasheet
hehe, i like those cheap outgoing auctions
Reaching, Thu Aug 03 2006, 03:35PM
hehe, just got 2 other bricks from ebay dirt cheap for around 5$ for both, they are halfbridge modules rated at 1200 volts 210A continous and 420A pulsed, and are much faster than the other big fuji igbts, maybe good for a "smaller" drsstc with a voltage doubler,and with only 11nF gate capacitance, means you can run a fullbridge of them with normal gdts and 2 pairs of driver ics at around 100khz
. heres the datasheethehe, i like those cheap outgoing auctions
Re: what do you think of these bricks?
Steve Ward, Thu Aug 03 2006, 05:05PM
Those IGBTs are actually capable of processing slightly more power than the 600V 400A bricks. I bet they could do 10 foot sparks in a full bridge.
Steve Ward, Thu Aug 03 2006, 05:05PM
maybe good for a "smaller" drsstc with a voltage doubler
Those IGBTs are actually capable of processing slightly more power than the 600V 400A bricks. I bet they could do 10 foot sparks in a full bridge.
Re: what do you think of these bricks?
HV Enthusiast, Thu Aug 03 2006, 06:46PM
Just keep in mind, size doesn't mean everything. You don't need big bricks to get big arcs. Just for comparison, I have a DRSSTC running with a half-bridge of small D2Pak Surface Mount IGBTs (which cost about $4.00 each), putting out over 50" worth of arc.
HV Enthusiast, Thu Aug 03 2006, 06:46PM
Just keep in mind, size doesn't mean everything. You don't need big bricks to get big arcs. Just for comparison, I have a DRSSTC running with a half-bridge of small D2Pak Surface Mount IGBTs (which cost about $4.00 each), putting out over 50" worth of arc.
Re: what do you think of these bricks?
Reaching, Thu Aug 03 2006, 08:14PM
mhh, interesting, so the "smaller" igbts are better than the 800A peak monsters?? mhh, when my halfbridge modules arrive i can build a fullbridge for testing and compare the fujis to the siemens. i have a great dealer on ebay who sells them for a few bucks each, maybe i should write a mail to him to get more of these halfbridge modules
Reaching, Thu Aug 03 2006, 08:14PM
mhh, interesting, so the "smaller" igbts are better than the 800A peak monsters?? mhh, when my halfbridge modules arrive i can build a fullbridge for testing and compare the fujis to the siemens. i have a great dealer on ebay who sells them for a few bucks each, maybe i should write a mail to him to get more of these halfbridge modules
Re: what do you think of these bricks?
HV Enthusiast, Fri Aug 04 2006, 02:37AM
Not necessarily. To evaluate an IGBT, a lot of different things need to be taken into consideration - thermal interface, electrical parameters, etc...
Big thing with the small packages is cooling. If you can provide a coldplate that can maintain a decent temperature, along with a good thermal interface between the components and the coldplate, then you can definitely run them like i've been doing. My thermal interface between component and coldplate is quite good. During operation, i only measure a 5 degree difference between component and coldplate so the thermal impedance is quite good.
HV Enthusiast, Fri Aug 04 2006, 02:37AM
Not necessarily. To evaluate an IGBT, a lot of different things need to be taken into consideration - thermal interface, electrical parameters, etc...
Big thing with the small packages is cooling. If you can provide a coldplate that can maintain a decent temperature, along with a good thermal interface between the components and the coldplate, then you can definitely run them like i've been doing. My thermal interface between component and coldplate is quite good. During operation, i only measure a 5 degree difference between component and coldplate so the thermal impedance is quite good.
Re: what do you think of these bricks?
Steve Ward, Fri Aug 04 2006, 04:20AM
I take it you are referring to your new modules, Dan? Just out of curiousity, what is the thermal interface between the surface mount device, the heatspreading plate, and heatsink?
Of course every device has its thermal limit when the peak die temp starts to exceed 150*C. The transient is so fast, that the best you can do is keep the die cool before it hits. Of course, the more massive your IGBT die, the more energy it can soak up before hitting critical temperatures. This is why Dan's surface mount IGBT bridge probably wont beat out a CM300 bridge, but because its newer technology, it's still more "bang" for your silicon.
Newer bricks from powerex and semikron, etc, are looking really great now, offering greater frequency operation, lower gate charge, and lower package inductance. If you could get ahold of these, then you are set.
Steve Ward, Fri Aug 04 2006, 04:20AM
I take it you are referring to your new modules, Dan? Just out of curiousity, what is the thermal interface between the surface mount device, the heatspreading plate, and heatsink?
Of course every device has its thermal limit when the peak die temp starts to exceed 150*C. The transient is so fast, that the best you can do is keep the die cool before it hits. Of course, the more massive your IGBT die, the more energy it can soak up before hitting critical temperatures. This is why Dan's surface mount IGBT bridge probably wont beat out a CM300 bridge, but because its newer technology, it's still more "bang" for your silicon.
Newer bricks from powerex and semikron, etc, are looking really great now, offering greater frequency operation, lower gate charge, and lower package inductance. If you could get ahold of these, then you are set.
Re: what do you think of these bricks?
HV Enthusiast, Fri Aug 04 2006, 12:41PM
The interface is an aluminum substrate board and grafoil pad.
HV Enthusiast, Fri Aug 04 2006, 12:41PM
The interface is an aluminum substrate board and grafoil pad.
Re: what do you think of these bricks?
Reaching, Sat Aug 05 2006, 12:04PM
hehe, for cooling my igbts i got a very huge heatsink from my fathers company. it is 75cm long and 16x16cm in diameter, its a so called "kühlprofil" with rails on one side to screw 120mm fans on it. i want to cut this thing in 3 pieces, one piece for other projects and the remaining two pieces for the fullbridge, eeach heatsink with 2x 120mm fans, should stay cool
here is it .
Reaching, Sat Aug 05 2006, 12:04PM
hehe, for cooling my igbts i got a very huge heatsink from my fathers company. it is 75cm long and 16x16cm in diameter, its a so called "kühlprofil" with rails on one side to screw 120mm fans on it. i want to cut this thing in 3 pieces, one piece for other projects and the remaining two pieces for the fullbridge, eeach heatsink with 2x 120mm fans, should stay cool
here is it .
Re: what do you think of these bricks?
HV Enthusiast, Sat Aug 05 2006, 04:00PM
Keep in mind, that you could have a heatsink one mile wide, by 50 feet high, and it would be useless. The limiting factor is almost always the thermal impedance at the interface between heatsink and device. This thermal impedance will determine the temperature drop between device and heatsink, and if its too high, then doesn't matter what kind of heatsink you are using.
HV Enthusiast, Sat Aug 05 2006, 04:00PM
Keep in mind, that you could have a heatsink one mile wide, by 50 feet high, and it would be useless. The limiting factor is almost always the thermal impedance at the interface between heatsink and device. This thermal impedance will determine the temperature drop between device and heatsink, and if its too high, then doesn't matter what kind of heatsink you are using.
Re: what do you think of these bricks?
Reaching, Sat Aug 05 2006, 05:45PM
yeah sure, but the thermal impendance of the device is an unchangeable factor in the datasheet, the only thing i can do is to screw the devices strong onto the heatsink with a small amount of thermal compound "wärmeleitpaste" between igbt and heatsink in hope the thermal factor wont change much.
Reaching, Sat Aug 05 2006, 05:45PM
yeah sure, but the thermal impendance of the device is an unchangeable factor in the datasheet, the only thing i can do is to screw the devices strong onto the heatsink with a small amount of thermal compound "wärmeleitpaste" between igbt and heatsink in hope the thermal factor wont change much.
Re: what do you think of these bricks?
HV Enthusiast, Sat Aug 05 2006, 08:02PM
I'm referring to the interface between device and heatsink, not the thermal impedance between case and die. Depending on application, one might need an electrically isolated pad between device and heatsink, or use an inferior thermal substrate, and this just kills the thermal performance.
HV Enthusiast, Sat Aug 05 2006, 08:02PM
I'm referring to the interface between device and heatsink, not the thermal impedance between case and die. Depending on application, one might need an electrically isolated pad between device and heatsink, or use an inferior thermal substrate, and this just kills the thermal performance.
Re: what do you think of these bricks?
..., Sat Aug 05 2006, 08:18PM
But for our drsstc's isn't the average power way lower than it would be if the device was being used for a 'normal' application, so we don't really need to worry about that? I mean, a brick that sized would normally be mounted on a copper waterblock with a big radiator to cool it... Then you would be worried about a perfect thermal contact, but if we only have a few hundred watts (you are gonna need a pretty big fan cooled heat sink to handle that) we should be fine with mediocre thermal contact?
..., Sat Aug 05 2006, 08:18PM
But for our drsstc's isn't the average power way lower than it would be if the device was being used for a 'normal' application, so we don't really need to worry about that? I mean, a brick that sized would normally be mounted on a copper waterblock with a big radiator to cool it... Then you would be worried about a perfect thermal contact, but if we only have a few hundred watts (you are gonna need a pretty big fan cooled heat sink to handle that) we should be fine with mediocre thermal contact?
Re: what do you think of these bricks?
JimG, Sat Aug 05 2006, 09:30PM
When your device has only .13" square inches of surface area to dissipate heat out of how it is connected is pretty important no matter what the average power is. If you can't cool it down while its off then heat will just build up. Even with 100us pulses my surface mount bridge can get hot pretty fast.
JimG, Sat Aug 05 2006, 09:30PM
When your device has only .13" square inches of surface area to dissipate heat out of how it is connected is pretty important no matter what the average power is. If you can't cool it down while its off then heat will just build up. Even with 100us pulses my surface mount bridge can get hot pretty fast.
Re: what do you think of these bricks?
..., Sat Aug 05 2006, 09:38PM
Well considering that Reaching has about 10in^2 he shouldn't have any worries
..., Sat Aug 05 2006, 09:38PM
Well considering that Reaching has about 10in^2 he shouldn't have any worries
Re: what do you think of these bricks?
Steve Ward, Sat Aug 05 2006, 10:01PM
My CM300 bridge gets pretty stinking hot without forced air cooling. The dissipation is in the 300W range. I try to apply the thermal grease as evenly as possible, with the recomended thickness of 8 mils. The cooler you keep the IGBTs, the harder you can push them .
Steve Ward, Sat Aug 05 2006, 10:01PM
My CM300 bridge gets pretty stinking hot without forced air cooling. The dissipation is in the 300W range. I try to apply the thermal grease as evenly as possible, with the recomended thickness of 8 mils. The cooler you keep the IGBTs, the harder you can push them
.Re: what do you think of these bricks?
Reaching, Sun Aug 06 2006, 12:05AM
yeah steve is right, the brick sized igbts get very hot, i remember from my old brick sstc that a temparature of the big heatsink of 100 degrees celsius was normal for longer periods of time, okay without a fan.
lets take a look at my bridge or the pieces i already have. today i cutted the heatsink in 3 pieces, with a normal handsaw!!nobody wanted to cut it for me for a few bucks so i made it.
when my fans arrive every heatsink gets 2x 120mm fans with an air flow of 92m³ per hour, should be enough for extended runtimes
Reaching, Sun Aug 06 2006, 12:05AM
yeah steve is right, the brick sized igbts get very hot, i remember from my old brick sstc that a temparature of the big heatsink of 100 degrees celsius was normal for longer periods of time, okay without a fan.
lets take a look at my bridge or the pieces i already have. today i cutted the heatsink in 3 pieces, with a normal handsaw!!nobody wanted to cut it for me for a few bucks so i made it.
when my fans arrive every heatsink gets 2x 120mm fans with an air flow of 92m³ per hour, should be enough for extended runtimes
Re: what do you think of these bricks?
Steve Ward, Sun Aug 06 2006, 02:08AM
Those are some excellent heatsinks! You might want to check out my CM600 bridge for ideas on a low inductance layout:
You can ignore the first half of those pics, just waveforms of the gate rise and fall time (25nS measured at the terminals). This was of course slowed down to something more realistic (about 500nS rise time and 200nS fall time)
Steve Ward, Sun Aug 06 2006, 02:08AM
Those are some excellent heatsinks! You might want to check out my CM600 bridge for ideas on a low inductance layout:
You can ignore the first half of those pics, just waveforms of the gate rise and fall time (25nS measured at the terminals). This was of course slowed down to something more realistic (about 500nS rise time and 200nS fall time)
Re: what do you think of these bricks?
Reaching, Sun Aug 06 2006, 12:39PM
thanks steve, good idea, need some copper sheets .i want to use similar capacitors 12000yf 350volt, 2 in series 2 paralelled for a total of 12000yf 700volt dc
so, now the igbts are screwed on the heatsinks, and now i have to build a good power driver to drive these bricks, i thought of 2 fullbridges with 30v + and -15v input, each fullbridge driving 2 bricks, every brick has its own 1:1gdt, okay never needed such a high power to drive a brick but i think it should work, or will the waveform look like crap cause of the many gdts?
Reaching, Sun Aug 06 2006, 12:39PM
thanks steve, good idea, need some copper sheets
.i want to use similar capacitors 12000yf 350volt, 2 in series 2 paralelled for a total of 12000yf 700volt dcso, now the igbts are screwed on the heatsinks, and now i have to build a good power driver to drive these bricks, i thought of 2 fullbridges with 30v + and -15v input, each fullbridge driving 2 bricks, every brick has its own 1:1gdt, okay never needed such a high power to drive a brick but i think it should work, or will the waveform look like crap cause of the many gdts?
Re: what do you think of these bricks?
Steve Ward, Sun Aug 06 2006, 07:55PM
The GDTs will not transfer the -15V +30V driver properly. It will come out as +/- 22.5V on the gates. You could try the "standard" gate drive scheme with just +/- 30V from a GDT, but the GDT will limit the peak current due to its stray inductance. For really big bricks, you simply must use high side drivers that connect straight to the IGBT. It might seem like a lot of extra work, but its really the best way. Those 1200V 210A IGBTs you have might work OK from GDT drive, with a healthy gate driver behind it (maybe 4 parallel UCC 9A drivers for a 36A gate driver). Try to minimize the GDT inductance. It might be worth going "coaxial" where your primary is simply a foil shield around the several secondary windings inside (imagine coax, but with 4 conductors inside instead of 1).
Steve Ward, Sun Aug 06 2006, 07:55PM
The GDTs will not transfer the -15V +30V driver properly. It will come out as +/- 22.5V on the gates. You could try the "standard" gate drive scheme with just +/- 30V from a GDT, but the GDT will limit the peak current due to its stray inductance. For really big bricks, you simply must use high side drivers that connect straight to the IGBT. It might seem like a lot of extra work, but its really the best way. Those 1200V 210A IGBTs you have might work OK from GDT drive, with a healthy gate driver behind it (maybe 4 parallel UCC 9A drivers for a 36A gate driver). Try to minimize the GDT inductance. It might be worth going "coaxial" where your primary is simply a foil shield around the several secondary windings inside (imagine coax, but with 4 conductors inside instead of 1).
Re: what do you think of these bricks?
..., Sun Aug 06 2006, 11:10PM
Might I suggest Blackplasma's uberdrivers? Seem like an easy solution to a complex problem
..., Sun Aug 06 2006, 11:10PM
Might I suggest Blackplasma's uberdrivers? Seem like an easy solution to a complex problem
Re: what do you think of these bricks?
Marko, Sun Aug 06 2006, 11:59PM
Uberdrivers® are unable to charge above 15 volts so IGBT's would die when overstressed.
I think Ward's solution is simplest and best at time.
Another way for making UberGDT® could be using a bunch of thin wires (0,1..0,2mm) wound multifilary, and braided together tightly. It is a bit pain to find your leads after but coupling is reallly high and leakage inductance low. Altough I'm not sure if it is really better than shielded networking cable GDT. (maybe you can get little less ohmic resistance?).
Marko, Sun Aug 06 2006, 11:59PM
Uberdrivers® are unable to charge above 15 volts so IGBT's would die when overstressed.
I think Ward's solution is simplest and best at time.
Another way for making UberGDT® could be using a bunch of thin wires (0,1..0,2mm) wound multifilary, and braided together tightly. It is a bit pain to find your leads after but coupling is reallly high and leakage inductance low. Altough I'm not sure if it is really better than shielded networking cable GDT. (maybe you can get little less ohmic resistance?).
Re: what do you think of these bricks?
Steve Conner, Mon Aug 07 2006, 09:40AM
Meh. I looked at all these schemes for driving IGBT bricks and in the end I decided to make a driver similar to Ward's (there's also a similar design by Dan Strother)
However, I seem to remember Jimmy Hynes was successful driving bricks off a GDT with a 1:2 stepup ratio to give +/-30V drive. But I think he did have a lot of trouble getting good looking waveforms.
Reaching your heatsink is "Supergeil"
Steve Conner, Mon Aug 07 2006, 09:40AM
Meh. I looked at all these schemes for driving IGBT bricks and in the end I decided to make a driver similar to Ward's (there's also a similar design by Dan Strother)
However, I seem to remember Jimmy Hynes was successful driving bricks off a GDT with a 1:2 stepup ratio to give +/-30V drive. But I think he did have a lot of trouble getting good looking waveforms.
Reaching your heatsink is "Supergeil"
Re: what do you think of these bricks?
Reaching, Mon Aug 07 2006, 10:30AM
hehe, yeah , supergeil is the right word for these uberbig heatsinks
i started to build the low inductance wiring and it looks nice. i left enough space to change the driving pcbs on each brick so i can experiment with some different driving shemes,. the problem is i dont understand exactly how wards brick driver works, first it looks like some sort of halfbridge driven by a ucc driver ic, but from expierience the ucc driver ic is very sensible for high side driving igbts or mosfets, but i wonder about the smps is it really needed, or can i simply build a transformer based powersupply for the drivers?. firstly i want to try the normal gdt sheme, the igbts only have a gate capacitance of 38nF , i already wired a 50nF brick with gdts and a resonant frequency of around 100khz and it worked well for me, .
Reaching, Mon Aug 07 2006, 10:30AM
hehe, yeah , supergeil is the right word for these uberbig heatsinks
i started to build the low inductance wiring and it looks nice. i left enough space to change the driving pcbs on each brick so i can experiment with some different driving shemes,. the problem is i dont understand exactly how wards brick driver works, first it looks like some sort of halfbridge driven by a ucc driver ic, but from expierience the ucc driver ic is very sensible for high side driving igbts or mosfets, but i wonder about the smps is it really needed, or can i simply build a transformer based powersupply for the drivers?. firstly i want to try the normal gdt sheme, the igbts only have a gate capacitance of 38nF , i already wired a 50nF brick with gdts and a resonant frequency of around 100khz and it worked well for me, .
Re: what do you think of these bricks?
Marko, Mon Aug 07 2006, 02:01PM
Ward's brick driver uses a GDT, two mosfet's in halfbridge and capacitive divider, arranged 2/3+1/3 on isolated 45V supply.
ICL7667 (ward) or another mosfet (Conner) are solely used to keep dorsal mosfet on and gate at -15V.
I also tought if we could connect the GDT to '+' (drain of dorsal mosfet) so it is defaultly on while GDT is 'off'. Steve conner pointed out there will be trouble with such a drive, probably because voltage spike at end of cycle would open the mosfet and cause shoot-trough.
Two mosfets directly on GDT could probably be done using a P-channel mosfet, but it's usually simpler to throw a '7667 in.
Marko, Mon Aug 07 2006, 02:01PM
Ward's brick driver uses a GDT, two mosfet's in halfbridge and capacitive divider, arranged 2/3+1/3 on isolated 45V supply.
ICL7667 (ward) or another mosfet (Conner) are solely used to keep dorsal mosfet on and gate at -15V.
I also tought if we could connect the GDT to '+' (drain of dorsal mosfet) so it is defaultly on while GDT is 'off'. Steve conner pointed out there will be trouble with such a drive, probably because voltage spike at end of cycle would open the mosfet and cause shoot-trough.
Two mosfets directly on GDT could probably be done using a P-channel mosfet, but it's usually simpler to throw a '7667 in.
Re: what do you think of these bricks?
Reaching, Mon Aug 07 2006, 03:42PM
mhh, okay, for the brick driver i thought about using this sheme
if i use this shematic should i look for some complementary mosfets or can i use any suitable mosfet? and for the supply voltage, what voltage should i choose, from the shematic i think less than 30volts, cause i dont want to go into trouble building a switch mode powersupply, simple transformer should do the same.
sorry, but im a complete noob in high side driving, .i understand the basics of this shematic but there are some things i dont understand .For example , why is the igbt emitter connected to pin 1 of the ucc gate driver? why is there a 30volt zener connected to pin 8 of the ucc driver, and i wonder about the 9,4volt zener on the 2200yf cap..some people out there who can explain it a bit?
Reaching, Mon Aug 07 2006, 03:42PM
mhh, okay, for the brick driver i thought about using this sheme
if i use this shematic should i look for some complementary mosfets or can i use any suitable mosfet? and for the supply voltage, what voltage should i choose, from the shematic i think less than 30volts, cause i dont want to go into trouble building a switch mode powersupply, simple transformer should do the same.
sorry, but im a complete noob in high side driving, .i understand the basics of this shematic but there are some things i dont understand .For example , why is the igbt emitter connected to pin 1 of the ucc gate driver? why is there a 30volt zener connected to pin 8 of the ucc driver, and i wonder about the 9,4volt zener on the 2200yf cap..some people out there who can explain it a bit?
Re: what do you think of these bricks?
williamn, Mon Aug 07 2006, 04:34PM
I have used Steves newest highside Gate drive circuit now a few times, I added a few things like undervoltage lockout but basically its a bastard of Steves circuit. The Zeners act as voltage regulators, the capacitive dividers form the grnd reference point and inturn provide + and - voltage rails. You can use a plain old 60/50hz transformer to power the circuits, for my big coil I did exactly this, I just made 4 isolated windings on a large laminant core, it was 120VAC:36VAC(4) resulting in 50volts on each circuit after their own fullwave rectifiers. I also have used a simple 20watt SMPS using a 50% dutycycle squarewave source (any type works, I used a 555) running at about 50Khz, TC44xx gatedrivers, gdt and simple halfbridge of cheapo mosfets. It works well and is much lighter than the heavy core transformers. This is a really good and simple gate drive circuit.
BTW here is a picture of the transformer I built, apparently my cat decided at the last second to make his opinion of my workmanship known.
williamn, Mon Aug 07 2006, 04:34PM
I have used Steves newest highside Gate drive circuit now a few times, I added a few things like undervoltage lockout but basically its a bastard of Steves circuit. The Zeners act as voltage regulators, the capacitive dividers form the grnd reference point and inturn provide + and - voltage rails. You can use a plain old 60/50hz transformer to power the circuits, for my big coil I did exactly this, I just made 4 isolated windings on a large laminant core, it was 120VAC:36VAC(4) resulting in 50volts on each circuit after their own fullwave rectifiers. I also have used a simple 20watt SMPS using a 50% dutycycle squarewave source (any type works, I used a 555) running at about 50Khz, TC44xx gatedrivers, gdt and simple halfbridge of cheapo mosfets. It works well and is much lighter than the heavy core transformers. This is a really good and simple gate drive circuit.
BTW here is a picture of the transformer I built, apparently my cat decided at the last second to make his opinion of my workmanship known.
Re: what do you think of these bricks?
Reaching, Mon Aug 07 2006, 04:56PM
okay, seems like the circuit works reliably, i made a board layout in eagle which fits perfectly on the bricks with all parts on it, except the 1:1 signal transformer.. but 50volts dc? i thought about 30volts dc or so, the 30volt zener will conduct all the time on 50volt input , what voltage is on the ucc driver ic? i hope nothing more than 15 volts.mhh, i have to give it a try, . . what type of p channel and n channel mosfets did you use?, i have no idea but i made my pcb with to220 devices, so any help is welcome.
Reaching, Mon Aug 07 2006, 04:56PM
okay, seems like the circuit works reliably, i made a board layout in eagle which fits perfectly on the bricks with all parts on it, except the 1:1 signal transformer.. but 50volts dc? i thought about 30volts dc or so, the 30volt zener will conduct all the time on 50volt input , what voltage is on the ucc driver ic? i hope nothing more than 15 volts.mhh, i have to give it a try, . . what type of p channel and n channel mosfets did you use?, i have no idea but i made my pcb with to220 devices, so any help is welcome.
Re: what do you think of these bricks?
williamn, Mon Aug 07 2006, 05:18PM
With load the voltage dropped to approx 45VDC which meant +30VDC and -15VDC. I use a 15V zener for the negative rail regulation. The mosfets where sampled from On and where
MTB30P06VT4G
NTB75N06T4G
williamn, Mon Aug 07 2006, 05:18PM
With load the voltage dropped to approx 45VDC which meant +30VDC and -15VDC. I use a 15V zener for the negative rail regulation. The mosfets where sampled from On and where
MTB30P06VT4G
NTB75N06T4G
Re: what do you think of these bricks?
Steve Ward, Mon Aug 07 2006, 06:35PM
Ok, some key points with the brick driver. To get the ultimate, super fast drive, use small mosfets. I used some small fets from fairchild, the Pch is rated 11A at 60V, and the N-ch is 13A at 60V. Look for fets with the smallest gate charge, but are still in the ~10A range!! This will result in a much faster driver than if you put in big monster 60A fets or something (using such large fets does not have any merit). The fets dont need to be a matched pair to work, but try to find a pair that have relatively similar ON/OFF times, gate charge, and current ratings.
Power supply issues. If you want the +30V, -15V drive, you will need a total of 45V, and then use a resistive divider (or you *could* rely on zeners to divide it, i actually use both) to get a -15V supply and +30V supply. The gate driver chip is powered from the -15V supply, so the "ground" pin of the chip is tied to -15V while the Vcc is tied to "0V". This puts a positive 15V on the chip, as required. The 15V and 30V zeners limit the -15V and +30V supplies as necessary. I suggest using 5W devices.
You must use an INVERTING driver so that when the input signal is low, the bottom fet gets 15V on it, turning it on, and pulling the IGBT gate to -15V when the coil is off between bangs.
The p-channel fet is driven through the .1uF capacitor, i got this clever trick from Jimmy. This inherently limits the maximum ON time of the IGBT due to the 1k resistor eventually draining the charge from the .1uF cap (and the Pch gate capacitance). But typically this time is considerably longer than you would ever operate at normally.
Steve Ward, Mon Aug 07 2006, 06:35PM
Ok, some key points with the brick driver. To get the ultimate, super fast drive, use small mosfets. I used some small fets from fairchild, the Pch is rated 11A at 60V, and the N-ch is 13A at 60V. Look for fets with the smallest gate charge, but are still in the ~10A range!! This will result in a much faster driver than if you put in big monster 60A fets or something (using such large fets does not have any merit). The fets dont need to be a matched pair to work, but try to find a pair that have relatively similar ON/OFF times, gate charge, and current ratings.
Power supply issues. If you want the +30V, -15V drive, you will need a total of 45V, and then use a resistive divider (or you *could* rely on zeners to divide it, i actually use both) to get a -15V supply and +30V supply. The gate driver chip is powered from the -15V supply, so the "ground" pin of the chip is tied to -15V while the Vcc is tied to "0V". This puts a positive 15V on the chip, as required. The 15V and 30V zeners limit the -15V and +30V supplies as necessary. I suggest using 5W devices.
You must use an INVERTING driver so that when the input signal is low, the bottom fet gets 15V on it, turning it on, and pulling the IGBT gate to -15V when the coil is off between bangs.
The p-channel fet is driven through the .1uF capacitor, i got this clever trick from Jimmy. This inherently limits the maximum ON time of the IGBT due to the 1k resistor eventually draining the charge from the .1uF cap (and the Pch gate capacitance). But typically this time is considerably longer than you would ever operate at normally.
Re: what do you think of these bricks?
Reaching, Mon Aug 07 2006, 09:00PM
ok, you helped me a lot, i ordered some parts and when they arrive i can start building the drivers. i nearly finished the low inductance wiring, heres a pic , i have some more double sided pcbs which i want to use for the middle contacts, then a layer of 2mm copper sheet or so, will see.
for the mosfets i ordered some irf530n channel and some irf 9530 p channel mosfets, both are 100volt 14,12A have a gate cap of around 800pf, the n channel is faster than the p channel, but that was what i found, hope that works
Reaching, Mon Aug 07 2006, 09:00PM
ok, you helped me a lot, i ordered some parts and when they arrive i can start building the drivers. i nearly finished the low inductance wiring, heres a pic
, i have some more double sided pcbs which i want to use for the middle contacts, then a layer of 2mm copper sheet or so, will see. for the mosfets i ordered some irf530n channel and some irf 9530 p channel mosfets, both are 100volt 14,12A have a gate cap of around 800pf, the n channel is faster than the p channel, but that was what i found, hope that works
Re: what do you think of these bricks?
Marko, Mon Aug 07 2006, 11:21PM
This driver is what I talked about 'with P-channel mosfet's.
It's input is designed ot be easy-to-drive, but if you want, you can simply use some standard UCC-driver board and drive MOSFET's trough a GDT with it. GDT would only need to be small driving a simplepush-pull of mosfet's recommended by steve.
Tons of conbinations..
Marko, Mon Aug 07 2006, 11:21PM
This driver is what I talked about 'with P-channel mosfet's.
It's input is designed ot be easy-to-drive, but if you want, you can simply use some standard UCC-driver board and drive MOSFET's trough a GDT with it. GDT would only need to be small driving a simplepush-pull of mosfet's recommended by steve.
Tons of conbinations..
Re: what do you think of these bricks?
Steve Ward, Mon Aug 07 2006, 11:23PM
The 530 and 9530 should work just great.
Looks like you are off to a good start. You plan to eventually add copper to the PC board for the bridge, right? I dont know that the clad alone will handle the currents. Im using some pretty thin copper for mine, i think its about .5mm thick. To solder it down, i first put little "beads" of solder down on the clad board. A blob of solder every 1-2 square inches is enough. Then get out a propane torch, set it for low flame, and slowly heat up an area of the copper until the beads of solder melt underneath. When you get it to the right temp, you can see solder flow along the outer edge of the copper (similar to if you have done plumbing work, when you "sweat" the pipes). I got my copper sheeting sold as roof flashing at the hardware store.
Oh, and i wanted to mention, there is probably no need to go with more than the 2 x 12000uF caps. 4 of the caps is more energy storage than you really need and may actually lower your power factor somewhat.
Steve Ward, Mon Aug 07 2006, 11:23PM
The 530 and 9530 should work just great.
Looks like you are off to a good start. You plan to eventually add copper to the PC board for the bridge, right? I dont know that the clad alone will handle the currents. Im using some pretty thin copper for mine, i think its about .5mm thick. To solder it down, i first put little "beads" of solder down on the clad board. A blob of solder every 1-2 square inches is enough. Then get out a propane torch, set it for low flame, and slowly heat up an area of the copper until the beads of solder melt underneath. When you get it to the right temp, you can see solder flow along the outer edge of the copper (similar to if you have done plumbing work, when you "sweat" the pipes). I got my copper sheeting sold as roof flashing at the hardware store.
Oh, and i wanted to mention, there is probably no need to go with more than the 2 x 12000uF caps. 4 of the caps is more energy storage than you really need and may actually lower your power factor somewhat.
Re: what do you think of these bricks?
Reaching, Tue Aug 08 2006, 04:35AM
okay, thanks steve, ,
the upper and lower power rails are finished as you see in the pic, there is a double layer pcb first and then a 2mm thick brass sheet screwed on it, that is what i want to do for the wiring in the middle, with some thick screws to connect the primary on it. in a few days i can test the bridge to see how far it goes i hope for a fres of around 50khz, . i calculated some secondaries and came upt with a 8"x40" Former wound with around 0,3 or 0,4mm wire, with a big enough toroid this will give me a resonant frequency of around 48khz, low enough i think, and big enough to create some really long bright power discharges
Reaching, Tue Aug 08 2006, 04:35AM
okay, thanks steve, ,
the upper and lower power rails are finished as you see in the pic, there is a double layer pcb first and then a 2mm thick brass sheet screwed on it, that is what i want to do for the wiring in the middle, with some thick screws to connect the primary on it. in a few days i can test the bridge to see how far it goes i hope for a fres of around 50khz, . i calculated some secondaries and came upt with a 8"x40" Former wound with around 0,3 or 0,4mm wire, with a big enough toroid this will give me a resonant frequency of around 48khz, low enough i think, and big enough to create some really long bright power discharges
Re: what do you think of these bricks?
Reaching, Thu Aug 10 2006, 06:28PM
my parts arrived and i build up the driver and tested it, but there are a few problems.
here my test condition:
i use the brick igbt driver from steve ward, the circuit is correct and with the same parts steve used in his shematic, i tested with a signal generator (squarewave 50khz 5volt out) and with a voltage between 15 and 30volt. if i set the voltage to around 15 volts i get no output (measured between the emitter and base connections without a brick installed) i only get a positive voltage of around 3 volts. if i set the voltage to about 20volts i get a weak squarewave falling to 0 in about 1 or 2 seconds, in place with a heavy current draw of over 1 A and the 10R resistor on the p and n channel mosfets is smoking. if i vary the voltage between 0 and 30 volts very fast i get a nice squarewave , ..what do i wrong??????
Reaching, Thu Aug 10 2006, 06:28PM
my parts arrived and i build up the driver and tested it, but there are a few problems.
here my test condition:
i use the brick igbt driver from steve ward, the circuit is correct and with the same parts steve used in his shematic, i tested with a signal generator (squarewave 50khz 5volt out) and with a voltage between 15 and 30volt. if i set the voltage to around 15 volts i get no output (measured between the emitter and base connections without a brick installed) i only get a positive voltage of around 3 volts. if i set the voltage to about 20volts i get a weak squarewave falling to 0 in about 1 or 2 seconds, in place with a heavy current draw of over 1 A and the 10R resistor on the p and n channel mosfets is smoking. if i vary the voltage between 0 and 30 volts very fast i get a nice squarewave , ..what do i wrong??????
Re: what do you think of these bricks?
Steve Ward, Thu Aug 10 2006, 07:49PM
Your gate driver chip probably has some undervoltage lock out, but i dont know if that would explain it. You have the polarity of the voltages correct? My gate drivers never seemed to care how slowly i brought up the voltage, or even if there was not enough voltage on the gate driver chip, it would just shut down.
Are you sure you wire the P-channel fet in right? The source connects to the +30V, the drain connects to the IGBT gate (through the resistor). Its easy to flip the drain and source if you arent paying attention, or used to working with P-channel stuff.
Steve Ward, Thu Aug 10 2006, 07:49PM
Your gate driver chip probably has some undervoltage lock out, but i dont know if that would explain it. You have the polarity of the voltages correct? My gate drivers never seemed to care how slowly i brought up the voltage, or even if there was not enough voltage on the gate driver chip, it would just shut down.
Are you sure you wire the P-channel fet in right? The source connects to the +30V, the drain connects to the IGBT gate (through the resistor). Its easy to flip the drain and source if you arent paying attention, or used to working with P-channel stuff.
Re: what do you think of these bricks?
Steve Conner, Fri Aug 11 2006, 11:03AM
It's probably a bad idea to try testing the circuit with a continuous drive signal from a signal generator. If you try to drive a brick continuously, the power dissipation will be huge and things will smoke. These drivers are really only designed to work at DRSSTC duties (ie, a 300us burst followed by a 10ms rest)
Steve Conner, Fri Aug 11 2006, 11:03AM
It's probably a bad idea to try testing the circuit with a continuous drive signal from a signal generator. If you try to drive a brick continuously, the power dissipation will be huge and things will smoke. These drivers are really only designed to work at DRSSTC duties (ie, a 300us burst followed by a 10ms rest)
Re: what do you think of these bricks?
Reaching, Fri Aug 11 2006, 12:00PM
but steve, its a very important circuit and i dont like the thought that it only works for short burst lenghts or so. 1 failure in one of 4 brick drivers and the whole bridge can get damaged.
btw, i got the driver working, i wired the p channel wrong, okay never made anything with p channel mosfets so steve ward was right, . but is it normal that the squarewave output goes back to zero voltage after about 10 seconds of runtime in continious mode?. maybe i should test it with an interrupter in place to simulate the drsstc mode
whats about this driver?
this is what ive build and tested and it works just fine, i can push the big bricks up to 170khz with a relatively good squarewave, on 50khz the squarewave is perfect, and it runs continious without problems. now the difficulty is to get a single supply voltage for this driver, can i use a capacitive divider like steve did or wil it result in a output going zero after a few seconds? i want the driver to be bombproof so if its difficult to divide 40volt properly to drive the ucc driver ic than i´ll go with 2 supply voltages instead
Reaching, Fri Aug 11 2006, 12:00PM
but steve, its a very important circuit and i dont like the thought that it only works for short burst lenghts or so. 1 failure in one of 4 brick drivers and the whole bridge can get damaged.
btw, i got the driver working, i wired the p channel wrong, okay never made anything with p channel mosfets so steve ward was right, . but is it normal that the squarewave output goes back to zero voltage after about 10 seconds of runtime in continious mode?. maybe i should test it with an interrupter in place to simulate the drsstc mode
whats about this driver?
this is what ive build and tested and it works just fine, i can push the big bricks up to 170khz with a relatively good squarewave, on 50khz the squarewave is perfect, and it runs continious without problems. now the difficulty is to get a single supply voltage for this driver, can i use a capacitive divider like steve did or wil it result in a output going zero after a few seconds? i want the driver to be bombproof so if its difficult to divide 40volt properly to drive the ucc driver ic than i´ll go with 2 supply voltages instead
Re: what do you think of these bricks?
Marko, Fri Aug 11 2006, 03:00PM
It's not a problem to divide voltages at all, and there's actually no sense to use a bunch of separate supply voltages (wich could trouble you because of higher supply inductance, you would again need decoupling and etc..)
It will be completely OK with etc. 1000uF and 2200uF caps, if you are really willing to be 1V precise then use 3 1000uF caps or similar.
You can use much smaller caps if you are using SMPS supply or have good initial filtering.
Capacitive divider does not cause 'draining after 10 secs' problem you are talking about, unless you wired something wrong.
Caps recharge with each cycle and it must work forever in CW if it is done right.
Running CW your drive mosfet's may dissipate a bit more heat but they are pretty beefy and hard to damage.
You must just care not to terribly overheat them if you aren't using heatsink...
Marko, Fri Aug 11 2006, 03:00PM
It's not a problem to divide voltages at all, and there's actually no sense to use a bunch of separate supply voltages (wich could trouble you because of higher supply inductance, you would again need decoupling and etc..)
It will be completely OK with etc. 1000uF and 2200uF caps, if you are really willing to be 1V precise then use 3 1000uF caps or similar.
You can use much smaller caps if you are using SMPS supply or have good initial filtering.
Capacitive divider does not cause 'draining after 10 secs' problem you are talking about, unless you wired something wrong.
Caps recharge with each cycle and it must work forever in CW if it is done right.
Running CW your drive mosfet's may dissipate a bit more heat but they are pretty beefy and hard to damage.
You must just care not to terribly overheat them if you aren't using heatsink...
Re: what do you think of these bricks?
Reaching, Fri Aug 11 2006, 04:32PM
okay, but i´ll stay away from a capacitive divider or so, i experimented a bit with it and it seems that they cannot supply enough current to be on the safe side, the voltage drop is to high and the caps recharge too slow,no sorry no way i´ll use 2 supply voltages instead,.
now to another problem, everyone uses a seperate powersupply for one high side driver, is this important or can i use the same powersupply for all four driver pcbs?
Reaching, Fri Aug 11 2006, 04:32PM
okay, but i´ll stay away from a capacitive divider or so, i experimented a bit with it and it seems that they cannot supply enough current to be on the safe side, the voltage drop is to high and the caps recharge too slow,no sorry no way i´ll use 2 supply voltages instead,.
now to another problem, everyone uses a seperate powersupply for one high side driver, is this important or can i use the same powersupply for all four driver pcbs?
Re: what do you think of these bricks?
Marko, Fri Aug 11 2006, 05:01PM
Using separate supplies for all 4 drivers is the point.
Bridges cannot work without some kind of isolated gatedrive.
You just doubled the trouble by tapping the power supply again for each board, that's 8 separate windings you need to wind on some toroidal transformer or etc.
That's why most people use a SMPS there, simply because small ferrite transformers are easier to wind.
Electrolytic caps themselves probably won't produce enough peak current anyway, and you are going to need somedecoupling caps (tantalum or ceramic) as close as posible to mosfets.
Using a tapped supply won't help there.
I also didn't understand the thing about the voltage drop you are talking about, but surely I'l let you be with your own design.
Marko, Fri Aug 11 2006, 05:01PM
now to another problem, everyone uses a seperate powersupply for one high side driver, is this important or can i use the same powersupply for all four driver pcbs?
Using separate supplies for all 4 drivers is the point.
Bridges cannot work without some kind of isolated gatedrive.
You just doubled the trouble by tapping the power supply again for each board, that's 8 separate windings you need to wind on some toroidal transformer or etc.
That's why most people use a SMPS there, simply because small ferrite transformers are easier to wind.
Electrolytic caps themselves probably won't produce enough peak current anyway, and you are going to need somedecoupling caps (tantalum or ceramic) as close as posible to mosfets.
Using a tapped supply won't help there.
I also didn't understand the thing about the voltage drop you are talking about, but surely I'l let you be with your own design.
Re: what do you think of these bricks?
Reaching, Fri Aug 11 2006, 05:53PM
i only build the driver from steve ward and had trouble with it , cause i had the p channel mosfet wired wrong. after correcting the failure the circuit worked but after some seconds the output dropped to 0 volt ,dont know why, i checked everything but the circuit is exactly what steve build. i experimented with the capacitive voltage divider and it seemed to me that the divider caps cannot recharge fast enough to supply the driver ic though the voltage dropped and output goes zero,
i supplied the driver with a normal 0-30volt powersupply, no such weird smps
Reaching, Fri Aug 11 2006, 05:53PM
i only build the driver from steve ward and had trouble with it , cause i had the p channel mosfet wired wrong. after correcting the failure the circuit worked but after some seconds the output dropped to 0 volt ,dont know why, i checked everything but the circuit is exactly what steve build. i experimented with the capacitive voltage divider and it seemed to me that the divider caps cannot recharge fast enough to supply the driver ic though the voltage dropped and output goes zero,
i supplied the driver with a normal 0-30volt powersupply, no such weird smps
Re: what do you think of these bricks?
Steve Ward, Fri Aug 11 2006, 06:50PM
Its NOT a CAPACITIVE DIVIDER as this is a DC input voltage! Its a resistive divider with filtering. If you left out the 1k resistors across the caps, it will cause the voltage to not distribute evenly, and eventually one of the voltage sections will just drop to 0V.. I left the 1k off the bottom cap for the -15V supply because the gate driver will draw about that much current as a 1k resistor anyway.
Anyway, ive tested this driver in CW mode driving CM600s. The gate resistor did eventually burn up (only 1/2W) but without the gate resistor, it ran CW just fine. You have to have proper mosfet heatsinking if you want to run CW, as the fets will have to dissipate a few W probably depending on the load. Be sure your gate resistor is very beefy!
Steve Ward, Fri Aug 11 2006, 06:50PM
can i use a capacitive divider like steve did or wil it result in a output going zero after a few seconds?
Its NOT a CAPACITIVE DIVIDER as this is a DC input voltage! Its a resistive divider with filtering. If you left out the 1k resistors across the caps, it will cause the voltage to not distribute evenly, and eventually one of the voltage sections will just drop to 0V.. I left the 1k off the bottom cap for the -15V supply because the gate driver will draw about that much current as a 1k resistor anyway.
Anyway, ive tested this driver in CW mode driving CM600s. The gate resistor did eventually burn up (only 1/2W) but without the gate resistor, it ran CW just fine. You have to have proper mosfet heatsinking if you want to run CW, as the fets will have to dissipate a few W probably depending on the load. Be sure your gate resistor is very beefy!
Re: what do you think of these bricks?
Marko, Fri Aug 11 2006, 08:07PM
There seems to be misunderstanding around here.
'Capacitive divider' (or whatever you call it) gets AC component from the halfbridge since both caps are discharged with each cycle (but it is supplied with DC)
Since drive isn't symetric your resistive divider (equalising reisstors?) are used to 'drain' bad oltage distribution on the caps.
As you already guessed reaching's problem is probably because of leaving out the resistors (?).
(or maybe I should just shut up
)
Marko, Fri Aug 11 2006, 08:07PM
There seems to be misunderstanding around here.
'Capacitive divider' (or whatever you call it) gets AC component from the halfbridge since both caps are discharged with each cycle (but it is supplied with DC)
Since drive isn't symetric your resistive divider (equalising reisstors?) are used to 'drain' bad oltage distribution on the caps.
As you already guessed reaching's problem is probably because of leaving out the resistors (?).
(or maybe I should just shut up
)Re: what do you think of these bricks?
Reaching, Fri Aug 11 2006, 08:31PM
i build everything exactly ,with resistors etc, but it doesnt work properly and i dont want to toast my bricks so i made my own driver, and it works and i can slap myself around if something bad happens. okay, the disadvantage is that i must use 2 different supply voltages, but who cares.
next problem is that im a total noob in smps so i want to use a normal transformer, selfwound of course for 4x12 and 4x40volts, a lot easier for me than a smps, thats nothing for me.i experimented a bit but i only get crap and i wanted it to be regulated but it have to be isolated and i have no parts here to build it like i want and im out of money, so >normal transformer
Reaching, Fri Aug 11 2006, 08:31PM
i build everything exactly ,with resistors etc, but it doesnt work properly and i dont want to toast my bricks so i made my own driver, and it works and i can slap myself around if something bad happens. okay, the disadvantage is that i must use 2 different supply voltages, but who cares.
next problem is that im a total noob in smps so i want to use a normal transformer, selfwound of course for 4x12 and 4x40volts, a lot easier for me than a smps, thats nothing for me.i experimented a bit but i only get crap and i wanted it to be regulated but it have to be isolated and i have no parts here to build it like i want and im out of money, so >normal transformer
Re: what do you think of these bricks?
Steve Ward, Fri Aug 11 2006, 09:12PM
Oh, i just noticed something with my schematic. While it technically is the right schematic for what i drive my CM300 IGBTs with, there are some things that might not allow it to work CW. Resistors R2 and R3 might be too big for CW operation as they would limit the current provided to the gate driver circuit. Also, i wanted to mention i used a 9.4V zener to clamp the negative voltage, so the real drive on my bricks is -9V to +30V right now, not -15V as i might have stated before. So the DC input would be lower accordingly.
Reaching, im not quite sure how your circuit:
could work with 40VDC applied across the half-bridge while only using a 15V drive from the gate driver chip (from what i can figure, it shouldnt be able to operate both fets, one or the other or both would be ON the whole time). And umm, your 30V TVS is shorting out any negative gate voltage, unless you used a bi-directional TVS, but you didnt indicate that in your schematic.
Steve Ward, Fri Aug 11 2006, 09:12PM
Oh, i just noticed something with my schematic. While it technically is the right schematic for what i drive my CM300 IGBTs with, there are some things that might not allow it to work CW. Resistors R2 and R3 might be too big for CW operation as they would limit the current provided to the gate driver circuit. Also, i wanted to mention i used a 9.4V zener to clamp the negative voltage, so the real drive on my bricks is -9V to +30V right now, not -15V as i might have stated before. So the DC input would be lower accordingly.
Reaching, im not quite sure how your circuit:
could work with 40VDC applied across the half-bridge while only using a 15V drive from the gate driver chip (from what i can figure, it shouldnt be able to operate both fets, one or the other or both would be ON the whole time). And umm, your 30V TVS is shorting out any negative gate voltage, unless you used a bi-directional TVS, but you didnt indicate that in your schematic.
Re: what do you think of these bricks?
Plasmaniac, Sat Aug 12 2006, 09:40AM
Eh, you dont need to make 4 separate supplies. 2x2 IGBTs are connected anyway, so you need just 2 separate PSUs for your brick drivers. I'm a noob when it comes to build a DRSSTC but I always try to save as many parts as I can, so I'd use just 2 separated PSUs, one for the electronics and one brick driver pair and another one for the other pair of brick drivers. Also, I'd make just 2 ~25V windings in total and a voltage doubler like this:
You can use a 7815 and 7915 to get the Voltages you need for your drivers and electronics (+-15 and +30?).
Plasmaniac, Sat Aug 12 2006, 09:40AM
Firkragg wrote ...
Using separate supplies for all 4 drivers is the point.
Bridges cannot work without some kind of isolated gatedrive.
You just doubled the trouble by tapping the power supply again for each board, that's 8 separate windings you need to wind on some toroidal transformer or etc.
Using separate supplies for all 4 drivers is the point.
Bridges cannot work without some kind of isolated gatedrive.
You just doubled the trouble by tapping the power supply again for each board, that's 8 separate windings you need to wind on some toroidal transformer or etc.
Eh, you dont need to make 4 separate supplies. 2x2 IGBTs are connected anyway, so you need just 2 separate PSUs for your brick drivers. I'm a noob when it comes to build a DRSSTC but I always try to save as many parts as I can, so I'd use just 2 separated PSUs, one for the electronics and one brick driver pair and another one for the other pair of brick drivers. Also, I'd make just 2 ~25V windings in total and a voltage doubler like this:
You can use a 7815 and 7915 to get the Voltages you need for your drivers and electronics (+-15 and +30?).
Re: what do you think of these bricks?
Reaching, Sat Aug 12 2006, 12:25PM
meh, i can definately say thats way too high for me, negative voltage, positive voltage,umg, smps and everything like this. mhh, omg, have to try some normal gdt drivers or so,
okay, update, gdt tests ended in crap, now i got my high side driver version working, but why should it not work steve, i think its fine, here are a few pics
(the brick driver fits perfectly on my bricks)
(on testing)
(gatewaveform on 50khz measured directly on the brick, v/div 5 volt 20ys/div input voltage 15 volt driver ic and 15 volt bridge)
(same but with 100khz input signal)
from the pics and the longtime test for 15 minutes im sure that it works, drawing around 0,2A 15 volts (okay have to make some tests with 30volts or so on the bridge but why should that cause trouble?
Reaching, Sat Aug 12 2006, 12:25PM
meh, i can definately say thats way too high for me, negative voltage, positive voltage,umg, smps and everything like this. mhh, omg, have to try some normal gdt drivers or so,
okay, update, gdt tests ended in crap, now i got my high side driver version working, but why should it not work steve, i think its fine, here are a few pics
(the brick driver fits perfectly on my bricks)
(on testing)
(gatewaveform on 50khz measured directly on the brick, v/div 5 volt 20ys/div input voltage 15 volt driver ic and 15 volt bridge)
(same but with 100khz input signal)from the pics and the longtime test for 15 minutes im sure that it works, drawing around 0,2A 15 volts (okay have to make some tests with 30volts or so on the bridge but why should that cause trouble?
Re: what do you think of these bricks?
Steve Conner, Sat Aug 12 2006, 05:25PM
Plasmaniac, you need at least 3 separate power supplies. Sure, the collectors of the top two IGBTs are connected together, but the gate drive has to be referenced to the emitters, which are connected to the RF outputs.
You can use a common power supply for the two bottom gate drivers, but like most people, I used 4 completely isolated supplies and drive signals. It's not really that much more hassle and you don't have to worry about the L*di/dt voltages that tend to appear between things that are supposed to be "connected together" in power electronics.
I also used a string of zeners instead of resistors, to regulate the supply to 36V and split it into +24 and -12. I see other people use higher voltages, but I think +24 is enough: I ran my OLTC gate driver at that voltage and ended up putting about 3000-4000A peak through each 600A brick with no problems.
Steve Conner, Sat Aug 12 2006, 05:25PM
Plasmaniac, you need at least 3 separate power supplies. Sure, the collectors of the top two IGBTs are connected together, but the gate drive has to be referenced to the emitters, which are connected to the RF outputs.
You can use a common power supply for the two bottom gate drivers, but like most people, I used 4 completely isolated supplies and drive signals. It's not really that much more hassle and you don't have to worry about the L*di/dt voltages that tend to appear between things that are supposed to be "connected together" in power electronics.
I also used a string of zeners instead of resistors, to regulate the supply to 36V and split it into +24 and -12. I see other people use higher voltages, but I think +24 is enough: I ran my OLTC gate driver at that voltage and ended up putting about 3000-4000A peak through each 600A brick with no problems.
Re: what do you think of these bricks?
Steve Ward, Sat Aug 12 2006, 05:26PM
OK, yes, your driver will work fine at 15V, but at about 20V, i predict the "sh*t will hit the fan". The reason you dont see a problem is probably because you arent too familiar with the workings of a P-ch mosfet. For a Pch to be off, you must have 0V or a positive G-S voltage. To turn them ON, the gate has to be lower than 0V relative to the source (typically -5V to turn it on). Now, your gate driver can apply 15V of total "swing" to both of the gates. Tell me, with 40V across the 2 sources of the bridge, how do you plan to turn both fets off at any point in time with only a 15V swing?
My driver avoids this problem in a sneaky way, by capacitively coupling the drive to the Pch fet, but not the Nch. Due to the self-biasing (provided by the resistor and diode across the G-S), the capacitive coupling only pulls the mosfets gate negative, and then pushes it back up to 0V.
Unless the IGBTs share a common emitter, they MUST have isolated power for the gate drivers. So, i suppose if you wanted, you could use one supply for the "bottom" 2 IGBTs. Id worry about ground loop issues (which could be a real problem). The top IGBTs have a common collector, but their emitter voltage is going up and down between 0V and Vin, so they each need their own isolation.
Steve Ward, Sat Aug 12 2006, 05:26PM
OK, yes, your driver will work fine at 15V, but at about 20V, i predict the "sh*t will hit the fan". The reason you dont see a problem is probably because you arent too familiar with the workings of a P-ch mosfet. For a Pch to be off, you must have 0V or a positive G-S voltage. To turn them ON, the gate has to be lower than 0V relative to the source (typically -5V to turn it on). Now, your gate driver can apply 15V of total "swing" to both of the gates. Tell me, with 40V across the 2 sources of the bridge, how do you plan to turn both fets off at any point in time with only a 15V swing?
My driver avoids this problem in a sneaky way, by capacitively coupling the drive to the Pch fet, but not the Nch. Due to the self-biasing (provided by the resistor and diode across the G-S), the capacitive coupling only pulls the mosfets gate negative, and then pushes it back up to 0V.
Eh, you dont need to make 4 separate supplies. 2x2 IGBTs are connected anyway, so you need just 2 separate PSUs for your brick drivers.
Unless the IGBTs share a common emitter, they MUST have isolated power for the gate drivers. So, i suppose if you wanted, you could use one supply for the "bottom" 2 IGBTs. Id worry about ground loop issues (which could be a real problem). The top IGBTs have a common collector, but their emitter voltage is going up and down between 0V and Vin, so they each need their own isolation.
Re: what do you think of these bricks?
Reaching, Sat Aug 12 2006, 06:56PM
yeah steve youre right, im not familiar in p channel, its the first time ever i used one. but its no problem to add this cap,resistor,diode to my current circuit, than it should work or not?
okay, made a few changes in my shematic and now everything works fine even with more than 15 volt applied to the bridge,
heres a pic of the gate waveform with 22volt applied to the bridge and 12 volt applied to the driver ic, same as above, 50khz 20ys/div 5volt/div
hehe, now everything is right,hope so
Reaching, Sat Aug 12 2006, 06:56PM
yeah steve youre right, im not familiar in p channel, its the first time ever i used one. but its no problem to add this cap,resistor,diode to my current circuit, than it should work or not?
okay, made a few changes in my shematic and now everything works fine even with more than 15 volt applied to the bridge,
heres a pic of the gate waveform with 22volt applied to the bridge and 12 volt applied to the driver ic, same as above, 50khz 20ys/div 5volt/div
hehe, now everything is right,hope so
Re: what do you think of these bricks?
Steve Conner, Sat Aug 12 2006, 10:26PM
That waveform looks great. You still ought to have some sort of resistor or zener string across your two electrolytic caps, though.
Steve Conner, Sat Aug 12 2006, 10:26PM
That waveform looks great. You still ought to have some sort of resistor or zener string across your two electrolytic caps, though.
Re: what do you think of these bricks?
williamn, Sun Aug 13 2006, 02:18AM
If you know how to build a simple SSTC you can build a SMPS. There is no scary Voodoo Here is a pic of one I built about six months ago, nothing fancy, no PWM or anything just free running at 30khz. One of the output windings can charge a 6800uf capacitor to 45VDC in about 2.5 seconds I have beat on this thing hard and it works fine in my medium DRSSTC. I have since moved on to seperate onboard PWM SMPS for each gatedrive card, but a simple one as above will do.
williamn, Sun Aug 13 2006, 02:18AM
If you know how to build a simple SSTC you can build a SMPS. There is no scary Voodoo
Here is a pic of one I built about six months ago, nothing fancy, no PWM or anything just free running at 30khz. One of the output windings can charge a 6800uf capacitor to 45VDC in about 2.5 seconds I have beat on this thing hard and it works fine in my medium DRSSTC. I have since moved on to seperate onboard PWM SMPS for each gatedrive card, but a simple one as above will do.
Re: what do you think of these bricks?
Reaching, Sun Aug 13 2006, 08:02AM
yeah, i build several small smps but the results were always dissapointing and i dont know what i do wrong or if something is wrong. i found a program to calculate the turn ratio for smps transformers and build a transformer for 4x15volt out, . i used the halfbridge converter with a simple 555 driving 2 mosfets at 100khz (to prevent inducing voltage spikes etc in my 50khz drive signal). i use a double E core with about 5x5cm and 1 cm thickness , that should be enough for 100watts
the only thing i dont understand, the powersupply charges the cap without load to about 20volts, with a 1k! load the voltage drops to 12 volt, and with a small motor connected the voltage drops to 11 volt drawing 200mA thats only about 2 watts! this is really dissapointing
Reaching, Sun Aug 13 2006, 08:02AM
yeah, i build several small smps but the results were always dissapointing and i dont know what i do wrong or if something is wrong. i found a program to calculate the turn ratio for smps transformers and build a transformer for 4x15volt out, . i used the halfbridge converter with a simple 555 driving 2 mosfets at 100khz (to prevent inducing voltage spikes etc in my 50khz drive signal). i use a double E core with about 5x5cm and 1 cm thickness , that should be enough for 100watts
the only thing i dont understand, the powersupply charges the cap without load to about 20volts, with a 1k! load the voltage drops to 12 volt, and with a small motor connected the voltage drops to 11 volt drawing 200mA thats only about 2 watts! this is really dissapointing
Re: what do you think of these bricks?
Marko, Sun Aug 13 2006, 11:40AM
Most-loved 'unregulated' SMPS is mazzili driver, just used with low-voltage secondary winding.
It offers a lot of power and voltage stability is pretty good with big loads (pulsed brick drivers aren't even considered a heavy load).
Marko, Sun Aug 13 2006, 11:40AM
Most-loved 'unregulated' SMPS is mazzili driver, just used with low-voltage secondary winding.
It offers a lot of power and voltage stability is pretty good with big loads (pulsed brick drivers aren't even considered a heavy load).
Re: what do you think of these bricks?
Reaching, Sun Aug 13 2006, 02:52PM
yeah, got it, bzw the small smps is ready to use and works fine. its just a simple flyback converter using a single irf630. outputs are 4x 15 volts 0,3A max 0,6A
heres a pic of the finished psu
i spend the whole day on it to make it work and i think i made my homework good the psu already survived the longtime test with 4x 120mm fans as a load on all 4 outputs drawing 15volt 0,6A for 15 minutes, the mosfet only got warm ,not hot or so.
Reaching, Sun Aug 13 2006, 02:52PM
yeah, got it, bzw the small smps is ready to use and works fine. its just a simple flyback converter using a single irf630. outputs are 4x 15 volts 0,3A max 0,6A
heres a pic of the finished psu
i spend the whole day on it to make it work and i think i made my homework good
the psu already survived the longtime test with 4x 120mm fans as a load on all 4 outputs drawing 15volt 0,6A for 15 minutes, the mosfet only got warm ,not hot or so. Re: what do you think of these bricks?
Marko, Sun Aug 13 2006, 03:28PM
Flyback converters without regulation sometimes do make trouble on high duty cycles.
The voltage spike that occurs will avalanche the mosfet unless it is dained by the load or some kind of clamping network.
Mosfet shouldn't get much more than room temperature with just few watts of load, and something is telling me that you are avalanching it.
Using resistor-diode clamping network and adequate snubber capacitor/network across mosfet could help him a lot staying cool.
You can also use a higher voltage mosfet sacrafycing a bit of ON resistance (altough I don't think it's really needed) to give more time for spike to drop.
Marko, Sun Aug 13 2006, 03:28PM
Flyback converters without regulation sometimes do make trouble on high duty cycles.
The voltage spike that occurs will avalanche the mosfet unless it is dained by the load or some kind of clamping network.
Mosfet shouldn't get much more than room temperature with just few watts of load, and something is telling me that you are avalanching it.
Using resistor-diode clamping network and adequate snubber capacitor/network across mosfet could help him a lot staying cool.
You can also use a higher voltage mosfet sacrafycing a bit of ON resistance (altough I don't think it's really needed) to give more time for spike to drop.
Re: what do you think of these bricks?
Steve Ward, Sun Aug 13 2006, 06:17PM
Eeep! i dont see any isolation between those windings on that flyback converter! I wouldnt trust that magnet wire to 900V or whatever.
Steve Ward, Sun Aug 13 2006, 06:17PM
Eeep! i dont see any isolation between those windings on that flyback converter! I wouldnt trust that magnet wire to 900V or whatever.
Re: what do you think of these bricks?
Reaching, Sun Aug 13 2006, 06:27PM
ehm, 900volt?? those are 600volt igbts so i can only get up to rectified mains 309volts exactly and its double isolated magnet wire rated at 600volts, . or is there more voltage on the gate emitter than i expect, the potenzial should be no more than 30volts or so
on the other side, this powersupply only powers the gate driver ics on the driver boards, not the halfbridge, so there should be no voltage or isolation problem between those magnet wires.
for powering the halfbridges i dont think i´ll use a smps, i´ll use a normal transformer with 4 isolated windings. i dont know how much power the halfbridge draws on pulsed mode but i think its more than 1A or so, a smps with 200watt is difficult to build so i´ll go with a normal transformer, better for a good dc voltage without any induced hf though ..
Reaching, Sun Aug 13 2006, 06:27PM
ehm, 900volt?? those are 600volt igbts so i can only get up to rectified mains 309volts exactly and its double isolated magnet wire rated at 600volts, . or is there more voltage on the gate emitter than i expect, the potenzial should be no more than 30volts or so
on the other side, this powersupply only powers the gate driver ics on the driver boards, not the halfbridge, so there should be no voltage or isolation problem between those magnet wires.
for powering the halfbridges i dont think i´ll use a smps, i´ll use a normal transformer with 4 isolated windings. i dont know how much power the halfbridge draws on pulsed mode but i think its more than 1A or so, a smps with 200watt is difficult to build so i´ll go with a normal transformer, better for a good dc voltage without any induced hf though ..
Re: what do you think of these bricks?
Hazmatt_(The Underdog), Sun Aug 13 2006, 07:50PM
What a pretty job you did Reaching. Nice PCB!
Hazmatt_(The Underdog), Sun Aug 13 2006, 07:50PM
What a pretty job you did Reaching. Nice PCB!
Re: what do you think of these bricks?
Steve Ward, Sun Aug 13 2006, 09:16PM
Oops, i forgot this is going to be a 350VDC input system. Ok, well my point still stands. With voltage spikes, you will probably be up around 550V. The power supplies are isolated for a reason, they need to stand off this 550V. I guess if the magnet wire has not been damaged anywhere, it *might* be ok for this voltage... i still dont like it though. Many of the GDTs i made with magnet wire shorted out.
Each gate driver IC has its "ground" connection to the Source of the bottom fet in the half-bridge driver, which is referenced to the emitter of the main IGBT, whose voltage is swinging up and down 300++ volts. So even this power supply needs 4 isolated windings for each chip, and the isolation between the windings should be good for 600V id say.
My SMPS driving 4 drivers for a CM600 bridge is powered from a 12W transformer. It probably wouldnt handle CW use of course. But its fine at up to 10% duty cycle.
Steve Ward, Sun Aug 13 2006, 09:16PM
Oops, i forgot this is going to be a 350VDC input system. Ok, well my point still stands. With voltage spikes, you will probably be up around 550V. The power supplies are isolated for a reason, they need to stand off this 550V. I guess if the magnet wire has not been damaged anywhere, it *might* be ok for this voltage... i still dont like it though. Many of the GDTs i made with magnet wire shorted out.
on the other side, this powersupply only powers the gate driver ics on the driver boards, not the halfbridge, so there should be no voltage or isolation problem between those magnet wires.
Each gate driver IC has its "ground" connection to the Source of the bottom fet in the half-bridge driver, which is referenced to the emitter of the main IGBT, whose voltage is swinging up and down 300++ volts. So even this power supply needs 4 isolated windings for each chip, and the isolation between the windings should be good for 600V id say.
i dont know how much power the halfbridge draws on pulsed mode but i think its more than 1A or so, a smps with 200watt is difficult to build so i´ll go with a normal transformer, better for a good dc voltage without any induced hf though ..
My SMPS driving 4 drivers for a CM600 bridge is powered from a 12W transformer. It probably wouldnt handle CW use of course. But its fine at up to 10% duty cycle.
Re: what do you think of these bricks?
Reaching, Mon Aug 14 2006, 04:26AM
mhh, okay steve you made me careful, no problem i´ll solve this "problem"
Reaching, Mon Aug 14 2006, 04:26AM
mhh, okay steve you made me careful, no problem i´ll solve this "problem"
Print this page