Body Diode isolation
Tom540, Tue Jul 11 2006, 02:17AMI noticed the other day that most of the SSTC circuits I find posted don't include any shottky's for body diode isolation. How important are they?
I built my coil after finding Richie's site and he stated that they were important and that the body diodes shouldnt be alowed to conduct. So I added them and but then realized at some point all 4 shottky's got toasted and were just a short. So for all I know they could have been that way since I first fired it up. I took 2 dead shottky's out the ones on the +v rail and managed to kill my mosfets later that day when I decided to increase on time. Of course the on time proabably killed them but either way they still died and the shottky's weren't doing much. I'm wondering if I should even bother buying some big beefy brick package type shotkky's to replace the old ones.
Re: Body Diode isolation
Steve Ward, Tue Jul 11 2006, 03:58AM
The problem with body diodes is that they have a slow recovery. But for the record, none of my old SSTCs used a schottky and ultrafast diode to disable the internal diode. If you use feedback from the secondary, you can get pretty close soft switching (where you switch at the primary current zero crossing), so the diode recovery is better (it gets worse with higher currents).
What FETs are you using, and what sort of spark length are you going for? Its possible that you are simply just exceeding the ability of the FET.
Steve Ward, Tue Jul 11 2006, 03:58AM
The problem with body diodes is that they have a slow recovery. But for the record, none of my old SSTCs used a schottky and ultrafast diode to disable the internal diode. If you use feedback from the secondary, you can get pretty close soft switching (where you switch at the primary current zero crossing), so the diode recovery is better (it gets worse with higher currents).
What FETs are you using, and what sort of spark length are you going for? Its possible that you are simply just exceeding the ability of the FET.
Re: Body Diode isolation
Tom540, Tue Jul 11 2006, 04:19AM
Well I am using ultrafast diodes and they were fine. They actually had smaller voltage and current rating the the shottky's I used on the drains of the FET's. I'm using IXFN48n50's and they seemed pretty beefy. I managed to not blow any for awhile and I was running the coil at ~3 bps. I know that's really slow The heatsinks stayed cool as well. I got a little over 3 ft arcs. I'm actually using one of your circuits from your site. It's the mini sstc but I added two more gate drivers and made it a full bridge. I added a voltage doubler as well.
As far as the soft switching, If I use secondary feedback for the resonant frequency it should be easier on the fet's? I wonder if its even necessary at such a slow rate. That would be using a CT right? That might help with the weird spurts I get when im slowly cranking the variac up. It seems to change burst rate, but that's another issue.I wonder if its even necessary at such a slow rate? I am a total noob at this high power stuff.
Tom540, Tue Jul 11 2006, 04:19AM
Well I am using ultrafast diodes and they were fine. They actually had smaller voltage and current rating the the shottky's I used on the drains of the FET's. I'm using IXFN48n50's and they seemed pretty beefy. I managed to not blow any for awhile and I was running the coil at ~3 bps. I know that's really slow The heatsinks stayed cool as well. I got a little over 3 ft arcs. I'm actually using one of your circuits from your site. It's the mini sstc but I added two more gate drivers and made it a full bridge. I added a voltage doubler as well.
As far as the soft switching, If I use secondary feedback for the resonant frequency it should be easier on the fet's? I wonder if its even necessary at such a slow rate. That would be using a CT right? That might help with the weird spurts I get when im slowly cranking the variac up. It seems to change burst rate, but that's another issue.I wonder if its even necessary at such a slow rate? I am a total noob at this high power stuff.
Re: Body Diode isolation
teravolt, Tue Jul 11 2006, 04:46AM
Tom I am also trying to do this and I am using FRED's specifically rurg80100 and they are 80A 1Kv 55ns recovery time by Fairchid. I got 4 of them on ebay for about 20. I prefer FRED's because they have a higher voltage rating than shottky
teravolt, Tue Jul 11 2006, 04:46AM
Tom I am also trying to do this and I am using FRED's specifically rurg80100 and they are 80A 1Kv 55ns recovery time by Fairchid. I got 4 of them on ebay for about 20. I prefer FRED's because they have a higher voltage rating than shottky
Re: Body Diode isolation
Tom540, Tue Jul 11 2006, 05:45AM
Yeah I'll have to check Ebay again for those. This means another board layout. :|
Tom540, Tue Jul 11 2006, 05:45AM
Yeah I'll have to check Ebay again for those. This means another board layout. :|
Re: Body Diode isolation
Steve Ward, Tue Jul 11 2006, 06:47AM
My latest SSTC uses IXFN80N50 mosfets. I use no extra diodes. I could only get about 3 foot sparks from my coil, so the fact that you achieved a little more is really great! I run my coil at 30-60bps though, not 3
. So far ive only lost 2 fets, and i think that was due to a large DC offset current on the primary coil due to non-50/50 drive signal from my PLL. Adding some DC blocking caps has allowed for my longest sparks yet (36") and no failures.
I havent added this coil to my website formally, but there is information about it here:
The PLL driver takes some understanding of everything going on, but once you get it finely tuned, it works great.
Also, just paralleling an UF diode with the FET isnt likely to fix anything. Diodes share current pretty well (even with slightly different forward voltages) so its pretty likely that the body diode is also conducting. So this is why you must have the schottky in series with the drain of the fet.
Steve Ward, Tue Jul 11 2006, 06:47AM
My latest SSTC uses IXFN80N50 mosfets. I use no extra diodes. I could only get about 3 foot sparks from my coil, so the fact that you achieved a little more is really great! I run my coil at 30-60bps though, not 3
. So far ive only lost 2 fets, and i think that was due to a large DC offset current on the primary coil due to non-50/50 drive signal from my PLL. Adding some DC blocking caps has allowed for my longest sparks yet (36") and no failures. I havent added this coil to my website formally, but there is information about it here:
The PLL driver takes some understanding of everything going on, but once you get it finely tuned, it works great.
Also, just paralleling an UF diode with the FET isnt likely to fix anything. Diodes share current pretty well (even with slightly different forward voltages) so its pretty likely that the body diode is also conducting. So this is why you must have the schottky in series with the drain of the fet.
Re: Body Diode isolation
Steve Conner, Tue Jul 11 2006, 09:37AM
A while back I did some experiments looking at whether body diode isolation is worth the hassle.
First I tested to see if it's worth just putting a fast diode across the FET with no schottky. Like Steve Ward said, this turned out to be totally pointless: the body diode takes practically all the current. (it's kind of a murphy's law of diodes that the faster they are, the higher a forward voltage they have for a given current.)
Next I investigated whether the body diodes actually need isolated. It turns out that if a SSTC is properly tuned and making sparks, the diodes hardly suffer from forced recovery at all. Even in Richie's nightmare scenario where you operate slightly below the resonant frequency, the streamer loading limits the capacitive current that causes forced recovery, enough that the magnetizing current of the primary can cancel it out, and make the overall current pretty much inductive at any frequency. So I built my DWSSTC without any isolation diodes and started trying to persuade everyone else to dump them too
I also promoted the PLL driver as being so accurate that it did away with the need for the isolation diodes. But empirically, people seemed to find that it worked fine with plain feedback drivers too. I guess the magnetizing current thing gives you a little more leeway for late switching, so even a plain feedback driver is plenty good enough.
Steve Conner, Tue Jul 11 2006, 09:37AM
A while back I did some experiments looking at whether body diode isolation is worth the hassle.
First I tested to see if it's worth just putting a fast diode across the FET with no schottky. Like Steve Ward said, this turned out to be totally pointless: the body diode takes practically all the current. (it's kind of a murphy's law of diodes that the faster they are, the higher a forward voltage they have for a given current.)
Next I investigated whether the body diodes actually need isolated. It turns out that if a SSTC is properly tuned and making sparks, the diodes hardly suffer from forced recovery at all. Even in Richie's nightmare scenario where you operate slightly below the resonant frequency, the streamer loading limits the capacitive current that causes forced recovery, enough that the magnetizing current of the primary can cancel it out, and make the overall current pretty much inductive at any frequency. So I built my DWSSTC without any isolation diodes and started trying to persuade everyone else to dump them too
I also promoted the PLL driver as being so accurate that it did away with the need for the isolation diodes. But empirically, people seemed to find that it worked fine with plain feedback drivers too. I guess the magnetizing current thing gives you a little more leeway for late switching, so even a plain feedback driver is plenty good enough.
Re: Body Diode isolation
Tom540, Tue Jul 11 2006, 05:57PM
Wow fast responses. Those diodes were kind of expensive anyway. It was the shottky's in series with the drains I had problems with. The ultrafast diodes were fine but if they arent doing anything then theyre comming out or at least im not putting them in again.
Steve what are the dimensions on that secondary? It looks HUGE! haha
Tom540, Tue Jul 11 2006, 05:57PM
Wow fast responses. Those diodes were kind of expensive anyway. It was the shottky's in series with the drains I had problems with. The ultrafast diodes were fine but if they arent doing anything then theyre comming out or at least im not putting them in again.
Steve what are the dimensions on that secondary? It looks HUGE! haha
Print this page