Do not underestimate the power of the brick. :)
brtaman, Sun Mar 07 2010, 08:32PMHello,
Just wanted to post some findings I have come across today, mainly in the hopes that it will be of use to anyone building a DRSSTC. Well my main finding today is do not underestimate a brick IGBT no matter how "bad" its ratings my seem.
The victim I chose for today is the SEMIKRON SKM50GB100D
Short summary:
1000V
50A nominal
100A pulsed
Gen1=SLOW
The reason being I have quite a few on hand and I thought them to be useless. Since I am waiting for wire for a nice big secondary and had a 200kHz(with topload) laying around, I decided to see what these babies are made off. To be honest was expecting a rather loud bang when i started working on the setup.
To my surprise during the voltage test, they seem to have absolutely no problem coping with this high frequency (extremely high for GEN1). It basically proves what the experts have been saying all along, the resonant soft switching the IGBTs see in DRSSTC duty can/is much faster than datasheets suggest.
Second surprise...I am pumping 700amps through the primary (7x pulsed rating), with no discernable heating (ok I may be cheating on that part with my 30x20cm 5kg heatsink, but still
). The sparks at the moment max out 50cm 1:1 with the secondary, but there is still much tuning to be done.To sum things up, IGBT bricks regardless of generation are heavy duty devices and seem to be rather underrated compared other less industrial packages. If you have one on hand, or can get one cheap even the tiny 50amp bricks pack quite a surprise so don't be afraid to use them. Just make sure to do a low-voltage test or if not, at least get some lexan for protection.
Best Regards,
brtaman
Re: Do not underestimate the power of the brick. :)
Steve Conner, Sun Mar 07 2010, 08:49PM
Thanks brtaman!
Theory predicts that slow IGBTs will switch faster in resonant mode, but the thing is, nobody seems to know exactly how much faster! So any experimental data that anyone can provide is very useful.
Steve Conner, Sun Mar 07 2010, 08:49PM
Thanks brtaman!
Theory predicts that slow IGBTs will switch faster in resonant mode, but the thing is, nobody seems to know exactly how much faster! So any experimental data that anyone can provide is very useful.
Re: Do not underestimate the power of the brick. :)
vasil, Sun Mar 07 2010, 08:53PM
God idea to try this. Nice work.
vasil, Sun Mar 07 2010, 08:53PM
God idea to try this. Nice work.
Re: Do not underestimate the power of the brick. :)
Herr Zapp, Sun Mar 07 2010, 09:58PM
brtaman -
Can you describe the measurement set-up you are using to monitor peak current in your tank circuit?
Regards,
Herr Zapp
Herr Zapp, Sun Mar 07 2010, 09:58PM
brtaman -
Can you describe the measurement set-up you are using to monitor peak current in your tank circuit?
Regards,
Herr Zapp
Re: Do not underestimate the power of the brick. :)
brtaman, Sun Mar 07 2010, 10:48PM
Thought I would throw in a picture of the "set-up"...jerry rigged to the limit, but I honestly thought there is no way the IGBT would function. Notice its still just a 4046 and a 555 controlling everything. :D (drsstc board is in the works, just waiting for the damn terminal blocks...-.-)
Mr. Conner: Thank you! :) The only reason I even tried these bricks out, is that I remembered reading one of your posts a while back stating the theory behind faster resonant switching.
Herr Zapp: The pretty standard CT config was used. 1:33:33 with a 10.6ohm resistor (calibrated DMM), waveform was monitored on the uber-elite old HUNG CHANG "dso" calibrated at least 5 years ago if not more. So take the current rating with a grain of salt. I am honestly more surprised with the fact that they are functioning at this high frequency than with the current (which is indeed a bit fishy considering the sparks are ~50cm), then again IIRC Jimmy Hynes had CM150 bricks running at 1500amps+.
brtaman, Sun Mar 07 2010, 10:48PM
Thought I would throw in a picture of the "set-up"...jerry rigged to the limit, but I honestly thought there is no way the IGBT would function. Notice its still just a 4046 and a 555 controlling everything. :D (drsstc board is in the works, just waiting for the damn terminal blocks...-.-)
Mr. Conner: Thank you! :) The only reason I even tried these bricks out, is that I remembered reading one of your posts a while back stating the theory behind faster resonant switching.
Herr Zapp: The pretty standard CT config was used. 1:33:33 with a 10.6ohm resistor (calibrated DMM), waveform was monitored on the uber-elite old HUNG CHANG "dso" calibrated at least 5 years ago if not more. So take the current rating with a grain of salt. I am honestly more surprised with the fact that they are functioning at this high frequency than with the current (which is indeed a bit fishy considering the sparks are ~50cm), then again IIRC Jimmy Hynes had CM150 bricks running at 1500amps+.
Re: Do not underestimate the power of the brick. :)
brtaman, Mon Mar 15 2010, 06:48PM
Well I seem to have run into an interesting problem in my continued testing of this tiny brick. After finally completing an actual drsstc driver (ward style) and boxing it up, I set up my coil. Everything is just one massive jerry-rig as I wait for parts for my actual planned drsstc.
It seems to be running at ludacris (again 700amp) primary current at a very small ~33us pulse width and produces next to nothing (5-10cm max) in sparks...giving it more pulse width results in stronger arcs however the current shoots well over 1kA...very strange indeed.
Here is the waveform, measurements taken with 1:33:33 with 10.6ohm burden ~1v=100A (+- 15%?), scope set to 2v divide:
At first I thought it was a matter of primary tunning, however I have tried many different configurations ranging from high to low coupling combined with different primary frequencies, from 1:1 to 5%, 10%, 15%, 20% under secondaries Fo. Nothing seemed have any measurable effect on the function of the coil. I should add that I managed to get 70cm sparks with the PLL driver running in VCO mode
. Though it seems the tuning in that case was quite different, that direct primary feedback.
As mentioned this is with PW at almost the minimum of the interrupter. OCD and all other components of the system are functioning as they should, though OCD is set to max due to the inherent triggering it was facing.
My thoughts:
-Obviously for some reason I am not transferring the energy into the secondary as I should. Phasing and all other factors have been checked and rechecked. Primary heating is a major problem, so in that sense do not doubt the scope-shot however would personally derail the presumed current by 15%.
Anyone have a same problem in setting up their coil? Any tips?
-There seems to be some lagging in the current and switching times. This is why I replaced the 1:33:33 with a 1:200 for feedback, thinking the cascaded CTs were the source of the phase shift on feedback, didn't help at all...
Obviously being the genius that I am I forgot to take scope shots of Vce in relation to primary current (don't like to have my camera in the same room, thank god for DSO :)). All seems "well" in the switching department, there seems to be a bit of lag in the primary current, result is that I am not getting nice ZCS, however ringing seems to be quite minimal based on my 100VDC input shots.
Any advice on getting closer to perfect ZCS?
Best Regards,
brtaman
brtaman, Mon Mar 15 2010, 06:48PM
Well I seem to have run into an interesting problem in my continued testing of this tiny brick. After finally completing an actual drsstc driver (ward style) and boxing it up, I set up my coil. Everything is just one massive jerry-rig as I wait for parts for my actual planned drsstc.
It seems to be running at ludacris (again 700amp) primary current at a very small ~33us pulse width and produces next to nothing (5-10cm max) in sparks...giving it more pulse width results in stronger arcs however the current shoots well over 1kA...very strange indeed.
Here is the waveform, measurements taken with 1:33:33 with 10.6ohm burden ~1v=100A (+- 15%?), scope set to 2v divide:
At first I thought it was a matter of primary tunning, however I have tried many different configurations ranging from high to low coupling combined with different primary frequencies, from 1:1 to 5%, 10%, 15%, 20% under secondaries Fo. Nothing seemed have any measurable effect on the function of the coil. I should add that I managed to get 70cm sparks with the PLL driver running in VCO mode
. Though it seems the tuning in that case was quite different, that direct primary feedback.As mentioned this is with PW at almost the minimum of the interrupter. OCD and all other components of the system are functioning as they should, though OCD is set to max due to the inherent triggering it was facing.
My thoughts:
-Obviously for some reason I am not transferring the energy into the secondary as I should. Phasing and all other factors have been checked and rechecked. Primary heating is a major problem, so in that sense do not doubt the scope-shot however would personally derail the presumed current by 15%.
Anyone have a same problem in setting up their coil? Any tips?
-There seems to be some lagging in the current and switching times. This is why I replaced the 1:33:33 with a 1:200 for feedback, thinking the cascaded CTs were the source of the phase shift on feedback, didn't help at all...
Obviously being the genius that I am I forgot to take scope shots of Vce in relation to primary current (don't like to have my camera in the same room, thank god for DSO :)). All seems "well" in the switching department, there seems to be a bit of lag in the primary current, result is that I am not getting nice ZCS, however ringing seems to be quite minimal based on my 100VDC input shots. Any advice on getting closer to perfect ZCS?
Best Regards,
brtaman
Re: Do not underestimate the power of the brick. :)
Arcstarter, Wed Mar 17 2010, 08:00PM
Yo dude, you rock, lemme send you all my igbts for testing and some slim jims...thanks!
But seriously, nice work. I am glad you did this, maybe i could get these IGBTs to do 500-700 amps at 60KHz. Kinda wussy 75 amp bricks, but semikrons are faster than mine.
Arcstarter, Wed Mar 17 2010, 08:00PM
Yo dude, you rock, lemme send you all my igbts for testing and some slim jims...thanks!
But seriously, nice work. I am glad you did this, maybe i could get these IGBTs to do 500-700 amps at 60KHz. Kinda wussy 75 amp bricks, but semikrons are faster than mine.
Re: Do not underestimate the power of the brick. :)
brtaman, Wed Mar 17 2010, 08:05PM
Thanks arcs, will be waiting for the Slim Jims to arrive. ;)
Well...I finally got working...kind of...I had been having a really tough time tuning the primary on this coil. I was estimating Fo through JavaTC, but threw that out the window as it was getting me nowhere... so current situation...Sec Fo ~220 Pri Fo ~290 (JAVATC....!?, forgot to scope waveform with new tuning, but should be less than 200, based on what I was measuring before).
Currently, 3 turns with a 166nF primary.
Really need to rethink my MMC...moved down to single string of two 4kv 0.33 FKP1 snubbers (8kVDC 0.166uF), anyone have a good deal on CDE 942s?
Video:
CT last I checked still read 700amp with OCD set a hair above that.
pretty mediocre performance for the current in the primary, but I thought I would post since video is worth 10^99 words. Really pushing this GEN1 above and beyond, have gone over 200us with no difference in output only made sparks brighter, now it seems it is all down to tuning and a new primary, the run in the video made 6AWG cable extremely hot...
Strike target is ~1m away so only 2x Sec length on the sparks.
EDIT: Took some photos, improving ever so slightly:
EDIT2: Added new video:
brtaman, Wed Mar 17 2010, 08:05PM
Thanks arcs, will be waiting for the Slim Jims to arrive. ;)
Well...I finally got working...kind of...I had been having a really tough time tuning the primary on this coil. I was estimating Fo through JavaTC, but threw that out the window as it was getting me nowhere... so current situation...Sec Fo ~220 Pri Fo ~290 (JAVATC....!?, forgot to scope waveform with new tuning, but should be less than 200, based on what I was measuring before).
Currently, 3 turns with a 166nF primary.
Really need to rethink my MMC...moved down to single string of two 4kv 0.33 FKP1 snubbers (8kVDC 0.166uF), anyone have a good deal on CDE 942s?
Video:
CT last I checked still read 700amp with OCD set a hair above that.
pretty mediocre performance for the current in the primary, but I thought I would post since video is worth 10^99 words. Really pushing this GEN1 above and beyond, have gone over 200us with no difference in output only made sparks brighter, now it seems it is all down to tuning and a new primary, the run in the video made 6AWG cable extremely hot...
Strike target is ~1m away so only 2x Sec length on the sparks.
EDIT: Took some photos, improving ever so slightly:
EDIT2: Added new video:
Print this page