Is this a full or half bridge inverter?
nzoomed, Wed Sept 09 2015, 02:30AMOK, im pretty new here, but i dont fully understand the difference between a full bridge and a half bridge inverter, other than people saying you get double the voltage output but you need more components.
What confuses me more is that i see people referring to their inverter as a Full H-Bridge, whats that supposed to mean?
Thats like saying you have a full half bridge inverter, how can it be both?
Im going to use CM300DY-24H series IGBT modules in a very similar fashion to Loneoceans DRSSTC3 but i cant find the exact schematic, but it looks rather simple in the way they are wired up basically the GDT outputs connect straight to the gates of each IGBT.
Id imagine the schematic below is a full bridge, but its unclear to me on steve's page if its a full bridge or half bridge configuration, if it is referring to the power supply using a voltage doubler or not then i will feel like a complete idiot! lol
Anyway besides that, i am unsure what D14-D18 or D6-D13 are for as i dont see anyone using them on their units.
I also dont know what side of the primary coil the current transformers are supposed to connect to and im interested to know if it matters what way around the primary is connected (i.e the internal winding and outside tapped cable) I would imagine it would change the phasing of the secondary, not that i can see any issue with this.
TIA
Re: Is this a full or half bridge inverter?
Dr. Slack, Wed Sept 09 2015, 08:30AM
That's an 'H-bridge' inverter, aka 'full H-bridge', aka 'full bridge', aka 'bridge' inverter. The four switching devices and the load make up the 5 short lines of the letter 'H'.
A 'half bridge' inverter, with no aliases that I know of, replaces the switches in one of the uprights of the H with capacitors, so contains only 2 switches overall, hence 'half'. It only delivers half the voltage swing at the load as the full bridge inverter. Sometimes the two caps are small and are sized to replace the function of resonant cap C1. Sometimes the two caps are big and are used also also as the DC bus. Sometimes intermediate values are used, in addition to MMC and PSU bus caps.
If D14 to D18 are specified as fast diodes, they are there to back up the operation of the body diodes. It can result in a lower forward drop, it can result in shorter reverse recovery, it does result in higher device capacitance. There's no clear advantage or disadvantage in the general case, but you may find an advantage with specific devices at specific powers and switching frequencies. Some people insist on using them, some abhor them.
Here D14 to D18 are specified as TVS, which are used to clamp inadvertent spikes from destroying the switches. They will add extra capacitance. Again there's no consensus on whether they are always a good idea. To use them takes up extra space, and they won't protect well unless they are very close to the device. That really depends on your mindset, do you want a stripped-down H bridge that's cheaper and slightly faster but more fragile, or the opposite of all those things.
D6 to D13 are to protect the gates, most are specified only to +/- 20v. In theory, if you have limited the drive to the GDT, you should not have excess gate volts. In practice, whoopsies do happen. If you make the GDT's removable, then putting D6 to D13 directly on the devices protects the gates from static charge accumulation when they are not being shorted by the GDTs. They are cheap, and a good insurance policy, most people use them.
In a full bridge, both ends of the load are equally 'hot', but the end of the primary connected to the MMC is much hotter. Tap the CT into either end of the load, but not the MMC-primary connection. In a half bridge, if your caps are the PSU caps, that point will be cold and the best one to use. If your caps are the MMC, then that point will be very hot and will be the worst one to use.
Dr. Slack, Wed Sept 09 2015, 08:30AM
That's an 'H-bridge' inverter, aka 'full H-bridge', aka 'full bridge', aka 'bridge' inverter. The four switching devices and the load make up the 5 short lines of the letter 'H'.
A 'half bridge' inverter, with no aliases that I know of, replaces the switches in one of the uprights of the H with capacitors, so contains only 2 switches overall, hence 'half'. It only delivers half the voltage swing at the load as the full bridge inverter. Sometimes the two caps are small and are sized to replace the function of resonant cap C1. Sometimes the two caps are big and are used also also as the DC bus. Sometimes intermediate values are used, in addition to MMC and PSU bus caps.
If D14 to D18 are specified as fast diodes, they are there to back up the operation of the body diodes. It can result in a lower forward drop, it can result in shorter reverse recovery, it does result in higher device capacitance. There's no clear advantage or disadvantage in the general case, but you may find an advantage with specific devices at specific powers and switching frequencies. Some people insist on using them, some abhor them.
Here D14 to D18 are specified as TVS, which are used to clamp inadvertent spikes from destroying the switches. They will add extra capacitance. Again there's no consensus on whether they are always a good idea. To use them takes up extra space, and they won't protect well unless they are very close to the device. That really depends on your mindset, do you want a stripped-down H bridge that's cheaper and slightly faster but more fragile, or the opposite of all those things.
D6 to D13 are to protect the gates, most are specified only to +/- 20v. In theory, if you have limited the drive to the GDT, you should not have excess gate volts. In practice, whoopsies do happen. If you make the GDT's removable, then putting D6 to D13 directly on the devices protects the gates from static charge accumulation when they are not being shorted by the GDTs. They are cheap, and a good insurance policy, most people use them.
In a full bridge, both ends of the load are equally 'hot', but the end of the primary connected to the MMC is much hotter. Tap the CT into either end of the load, but not the MMC-primary connection. In a half bridge, if your caps are the PSU caps, that point will be cold and the best one to use. If your caps are the MMC, then that point will be very hot and will be the worst one to use.
Re: Is this a full or half bridge inverter?
nzoomed, Thu Sept 10 2015, 12:15AM
Thanks, that answers alot!
Here i was thinking people were referring to a half bridge when they mention H-Bridge, so its good to have that clarified! :)
I assume C2 and C3 are for protection from primary strikes?
I believe that most people nowdays connect a 68nf cap from the negative rail to ground for this, but there appears to be many different views on this.
Thats good to know what those diodes are for and will help me with my design considerations.
nzoomed, Thu Sept 10 2015, 12:15AM
Dr. Slack wrote ...
That's an 'H-bridge' inverter, aka 'full H-bridge', aka 'full bridge', aka 'bridge' inverter. The four switching devices and the load make up the 5 short lines of the letter 'H'.
A 'half bridge' inverter, with no aliases that I know of, replaces the switches in one of the uprights of the H with capacitors, so contains only 2 switches overall, hence 'half'. It only delivers half the voltage swing at the load as the full bridge inverter. Sometimes the two caps are small and are sized to replace the function of resonant cap C1. Sometimes the two caps are big and are used also also as the DC bus. Sometimes intermediate values are used, in addition to MMC and PSU bus caps.
If D14 to D18 are specified as fast diodes, they are there to back up the operation of the body diodes. It can result in a lower forward drop, it can result in shorter reverse recovery, it does result in higher device capacitance. There's no clear advantage or disadvantage in the general case, but you may find an advantage with specific devices at specific powers and switching frequencies. Some people insist on using them, some abhor them.
Here D14 to D18 are specified as TVS, which are used to clamp inadvertent spikes from destroying the switches. They will add extra capacitance. Again there's no consensus on whether they are always a good idea. To use them takes up extra space, and they won't protect well unless they are very close to the device. That really depends on your mindset, do you want a stripped-down H bridge that's cheaper and slightly faster but more fragile, or the opposite of all those things.
D6 to D13 are to protect the gates, most are specified only to +/- 20v. In theory, if you have limited the drive to the GDT, you should not have excess gate volts. In practice, whoopsies do happen. If you make the GDT's removable, then putting D6 to D13 directly on the devices protects the gates from static charge accumulation when they are not being shorted by the GDTs. They are cheap, and a good insurance policy, most people use them.
In a full bridge, both ends of the load are equally 'hot', but the end of the primary connected to the MMC is much hotter. Tap the CT into either end of the load, but not the MMC-primary connection. In a half bridge, if your caps are the PSU caps, that point will be cold and the best one to use. If your caps are the MMC, then that point will be very hot and will be the worst one to use.
That's an 'H-bridge' inverter, aka 'full H-bridge', aka 'full bridge', aka 'bridge' inverter. The four switching devices and the load make up the 5 short lines of the letter 'H'.
A 'half bridge' inverter, with no aliases that I know of, replaces the switches in one of the uprights of the H with capacitors, so contains only 2 switches overall, hence 'half'. It only delivers half the voltage swing at the load as the full bridge inverter. Sometimes the two caps are small and are sized to replace the function of resonant cap C1. Sometimes the two caps are big and are used also also as the DC bus. Sometimes intermediate values are used, in addition to MMC and PSU bus caps.
If D14 to D18 are specified as fast diodes, they are there to back up the operation of the body diodes. It can result in a lower forward drop, it can result in shorter reverse recovery, it does result in higher device capacitance. There's no clear advantage or disadvantage in the general case, but you may find an advantage with specific devices at specific powers and switching frequencies. Some people insist on using them, some abhor them.
Here D14 to D18 are specified as TVS, which are used to clamp inadvertent spikes from destroying the switches. They will add extra capacitance. Again there's no consensus on whether they are always a good idea. To use them takes up extra space, and they won't protect well unless they are very close to the device. That really depends on your mindset, do you want a stripped-down H bridge that's cheaper and slightly faster but more fragile, or the opposite of all those things.
D6 to D13 are to protect the gates, most are specified only to +/- 20v. In theory, if you have limited the drive to the GDT, you should not have excess gate volts. In practice, whoopsies do happen. If you make the GDT's removable, then putting D6 to D13 directly on the devices protects the gates from static charge accumulation when they are not being shorted by the GDTs. They are cheap, and a good insurance policy, most people use them.
In a full bridge, both ends of the load are equally 'hot', but the end of the primary connected to the MMC is much hotter. Tap the CT into either end of the load, but not the MMC-primary connection. In a half bridge, if your caps are the PSU caps, that point will be cold and the best one to use. If your caps are the MMC, then that point will be very hot and will be the worst one to use.
Thanks, that answers alot!
Here i was thinking people were referring to a half bridge when they mention H-Bridge, so its good to have that clarified! :)
I assume C2 and C3 are for protection from primary strikes?
I believe that most people nowdays connect a 68nf cap from the negative rail to ground for this, but there appears to be many different views on this.
Thats good to know what those diodes are for and will help me with my design considerations.
Re: Is this a full or half bridge inverter?
loneoceans, Thu Sept 10 2015, 12:45AM
C2 and C3 are snubber capacitors which are typically polypropylene film caps and placed as closed as possible across the + and - bus terminals of the IGBTs. Many large IGBT modules come in half bridge configuration each, where one terminal is typically connected to V+, one to V- and one as the middle junction. If you look at my DRSSTC 3
, these 'C2 and C3' equivalent caps are the black boxy caps I place directly over the terminals of the IGBTs. More information here:
Primary strike protection is a different issue and connects to RF ground which is different from V negative. In that case, a bypass capacitor is typically used to connect one of the V+ or V- or both rails to ground via a good quality capacitor. This provides an 'easy' path for the streamer energy to travel to ground if it hits the primary circuit. I use 68nF or 100nF 2kV caps just because I have a bunch of them lying around.
loneoceans, Thu Sept 10 2015, 12:45AM
C2 and C3 are snubber capacitors which are typically polypropylene film caps and placed as closed as possible across the + and - bus terminals of the IGBTs. Many large IGBT modules come in half bridge configuration each, where one terminal is typically connected to V+, one to V- and one as the middle junction. If you look at my DRSSTC 3
, these 'C2 and C3' equivalent caps are the black boxy caps I place directly over the terminals of the IGBTs. More information here: Primary strike protection is a different issue and connects to RF ground which is different from V negative. In that case, a bypass capacitor is typically used to connect one of the V+ or V- or both rails to ground via a good quality capacitor. This provides an 'easy' path for the streamer energy to travel to ground if it hits the primary circuit. I use 68nF or 100nF 2kV caps just because I have a bunch of them lying around.
Re: Is this a full or half bridge inverter?
nzoomed, Thu Sept 10 2015, 01:14AM
Thanks for your reply Loneoceans, yes im totally forgetting about the snubber caps! That reminds me i need to find some aerovox RBPS caps soon.
I was going to email you about this earlier, I have been in contact with you a bit over the last 6 months :)
I see your coil doesnt have the diodes discussed on the inverter, i suspect that the only crucial ones are the ones that go across each 5.1K resistor to each gate.
With proper care taken, i dont expect my Universal driver to put over voltage on the gates.
nzoomed, Thu Sept 10 2015, 01:14AM
loneoceans wrote ...
C2 and C3 are snubber capacitors which are typically polypropylene film caps and placed as closed as possible across the + and - bus terminals of the IGBTs. Many large IGBT modules come in half bridge configuration each, where one terminal is typically connected to V+, one to V- and one as the middle junction. If you look at my DRSSTC 3 , these 'C2 and C3' equivalent caps are the black boxy caps I place directly over the terminals of the IGBTs. More information here:
Primary strike protection is a different issue and connects to RF ground which is different from V negative. In that case, a bypass capacitor is typically used to connect one of the V+ or V- or both rails to ground via a good quality capacitor. This provides an 'easy' path for the streamer energy to travel to ground if it hits the primary circuit. I use 68nF or 100nF 2kV caps just because I have a bunch of them lying around.
C2 and C3 are snubber capacitors which are typically polypropylene film caps and placed as closed as possible across the + and - bus terminals of the IGBTs. Many large IGBT modules come in half bridge configuration each, where one terminal is typically connected to V+, one to V- and one as the middle junction. If you look at my DRSSTC 3
, these 'C2 and C3' equivalent caps are the black boxy caps I place directly over the terminals of the IGBTs. More information here: Primary strike protection is a different issue and connects to RF ground which is different from V negative. In that case, a bypass capacitor is typically used to connect one of the V+ or V- or both rails to ground via a good quality capacitor. This provides an 'easy' path for the streamer energy to travel to ground if it hits the primary circuit. I use 68nF or 100nF 2kV caps just because I have a bunch of them lying around.
Thanks for your reply Loneoceans, yes im totally forgetting about the snubber caps! That reminds me i need to find some aerovox RBPS caps soon.
I was going to email you about this earlier, I have been in contact with you a bit over the last 6 months :)
I see your coil doesnt have the diodes discussed on the inverter, i suspect that the only crucial ones are the ones that go across each 5.1K resistor to each gate.
With proper care taken, i dont expect my Universal driver to put over voltage on the gates.
Re: Is this a full or half bridge inverter?
Dr. Slack, Thu Sept 10 2015, 01:07PM
Yes, 'H' is not an abbreviation for 'half'.
One further possible source of confusion. Loneoceans referred to pair of series connected devices in one package. I have seen these series-connected things referred to by the manufacturer as a 'full half-bridge', aaaaaagh!
... 5.1 ohm resistor, you'll have a lot of trouble with 5.1k
Dr. Slack, Thu Sept 10 2015, 01:07PM
Yes, 'H' is not an abbreviation for 'half'.
One further possible source of confusion. Loneoceans referred to pair of series connected devices in one package. I have seen these series-connected things referred to by the manufacturer as a 'full half-bridge', aaaaaagh!
... 5.1 ohm resistor, you'll have a lot of trouble with 5.1k
Re: Is this a full or half bridge inverter?
nzoomed, Thu Sept 10 2015, 09:20PM
OK, thanks for that!
Ive been learning so much its not funny.
Loneoceans has also been a great help, i cant wait to upload some photos when the coil is up and running!
nzoomed, Thu Sept 10 2015, 09:20PM
Dr. Slack wrote ...
Yes, 'H' is not an abbreviation for 'half'.
One further possible source of confusion. Loneoceans referred to pair of series connected devices in one package. I have seen these series-connected things referred to by the manufacturer as a 'full half-bridge', aaaaaagh!
... 5.1 ohm resistor, you'll have a lot of trouble with 5.1k
Yes, 'H' is not an abbreviation for 'half'.
One further possible source of confusion. Loneoceans referred to pair of series connected devices in one package. I have seen these series-connected things referred to by the manufacturer as a 'full half-bridge', aaaaaagh!
... 5.1 ohm resistor, you'll have a lot of trouble with 5.1k
OK, thanks for that!
Ive been learning so much its not funny.
Loneoceans has also been a great help, i cant wait to upload some photos when the coil is up and running!
Re: Is this a full or half bridge inverter?
Hydron, Thu Sept 10 2015, 10:23PM
I believe the current consensus is that the TVS diodes across the bridge don't do a lot (though no harm either) - it's much better to put the effort into bus layout etc. The diodes on the gates however are a very easy and cheap insurance policy.
Getting those Aerovox RBPS snubbers (or equivalent) in NZ is a bit of a pain unfortunately. I ended up using them for my MMC too, so I got a big (>50) lot of them from CTR surplus for ~$100 USD, but shipping via NZ post's freight forwarding service doubled the cost (CTR doesn't ship internationally).
Edit:
I'll second the comment about loneoceans - he's made some great info/writeups/photos available to the TC community!
Hydron, Thu Sept 10 2015, 10:23PM
I believe the current consensus is that the TVS diodes across the bridge don't do a lot (though no harm either) - it's much better to put the effort into bus layout etc. The diodes on the gates however are a very easy and cheap insurance policy.
Getting those Aerovox RBPS snubbers (or equivalent) in NZ is a bit of a pain unfortunately. I ended up using them for my MMC too, so I got a big (>50) lot of them from CTR surplus for ~$100 USD, but shipping via NZ post's freight forwarding service doubled the cost (CTR doesn't ship internationally).
Edit:
I'll second the comment about loneoceans - he's made some great info/writeups/photos available to the TC community!
Re: Is this a full or half bridge inverter?
klugesmith, Fri Sept 11 2015, 02:30AM
Here is a crude illustration of what Dr Slack said, about why it's called an H-bridge topology.
The same schematic also has a FULL wave BRIDGE rectifier.
(the large type is 72 point Garamond Bold, enlarged to 270% in MS Paint)
klugesmith, Fri Sept 11 2015, 02:30AM
Here is a crude illustration of what Dr Slack said, about why it's called an H-bridge topology.
The same schematic also has a FULL wave BRIDGE rectifier.
(the large type is 72 point Garamond Bold, enlarged to 270% in MS Paint)
Re: Is this a full or half bridge inverter?
nzoomed, Fri Sept 11 2015, 06:27AM
They are cheap on ebay, but cost 10X what they are worth to send them which is stupid ($40 post.)
nzoomed, Fri Sept 11 2015, 06:27AM
Hydron wrote ...
I believe the current consensus is that the TVS diodes across the bridge don't do a lot (though no harm either) - it's much better to put the effort into bus layout etc. The diodes on the gates however are a very easy and cheap insurance policy.
Getting those Aerovox RBPS snubbers (or equivalent) in NZ is a bit of a pain unfortunately. I ended up using them for my MMC too, so I got a big (>50) lot of them from CTR surplus for ~$100 USD, but shipping via NZ post's freight forwarding service doubled the cost (CTR doesn't ship internationally).
Edit:
I'll second the comment about loneoceans - he's made some great info/writeups/photos available to the TC community!
I believe the current consensus is that the TVS diodes across the bridge don't do a lot (though no harm either) - it's much better to put the effort into bus layout etc. The diodes on the gates however are a very easy and cheap insurance policy.
Getting those Aerovox RBPS snubbers (or equivalent) in NZ is a bit of a pain unfortunately. I ended up using them for my MMC too, so I got a big (>50) lot of them from CTR surplus for ~$100 USD, but shipping via NZ post's freight forwarding service doubled the cost (CTR doesn't ship internationally).
Edit:
I'll second the comment about loneoceans - he's made some great info/writeups/photos available to the TC community!
They are cheap on ebay, but cost 10X what they are worth to send them which is stupid ($40 post.)
Print this page