Kevin Gilmore Circlotron Output, Large Enclosure [high-slew] Electrostatic Amplifier

[JoaMat]

Thanks for the information.

I have made some attempts with the Circlotron. Blown  both PSUs and outputs. Small black fragments all over the place. Seems that things easily might get wrong and then both PSU and output are effected.

 

[fat joey]

On 2015/10/4 at 11:35 PM, kevin gilmore said:

so the default would be 25ma on the output stage and 15ma on the driver stage

so 90 watts for the output stage and 54 watts for the driver stage (both at 900/450v)

plus what ever differential on the power supply so close to 200W is probably right

 

Hi there, speaking of CREE driver, here's a pricy edition of c2m1000170d
still 1700V, but rates at 72A

http://www.mouser.com/ProductDetail/Wolfspeed-Cree/C2M0045170D

Will this high current version be stable at 100ma or even 200ma? (Maybe at that level, have to mate with a $$$$$ source to hear the small difference compared to 25ma)

Looks like there will be higher voltage CREE driver coming out in the future, let's say, 2.5kV 5A (HV version of c2m1000170d), is the 2.5kV Cree driver still have benefits in the design, ie. of the electrostatics amps?

[Tinkerer]

So, I kept working on a few things on the right channel while I'm waiting on parts. Got all the balance and offset adjusted before hooking up the output board. Double checked all connections by hand and with multimeter. Checked PSU's, the whole 9 yards. Powered up the output board to check it and adjust current. I started it down low compared to last time. It lasted all of about three seconds before blowing (blew right as it hit 25mA). Same deal, left 120ohm resistor and both mosfets exploded. This time however, it didn't take any of the HV900 PSU's with it.

This didn't happen when the offset was really high, and like I said the left channel output board exploded when I was adjusting the offset down to zero as well. After what JoaMa said, this is starting to look like a board problem. Something to do with moving the feedback to the driver board maybe? Is 3W not a high enough rating for the 120ohm resistors? I could really use some help.

[spritzer]

6 hours ago, fat joey said:

 

Hi there, speaking of CREE driver, here's a pricy edition of c2m1000170d
still 1700V, but rates at 72A

http://www.mouser.com/ProductDetail/Wolfspeed-Cree/C2M0045170D

Will this high current version be stable at 100ma or even 200ma? (Maybe at that level, have to mate with a $$$$$ source to hear the small difference compared to 25ma)

Looks like there will be higher voltage CREE driver coming out in the future, let's say, 2.5kV 5A (HV version of c2m1000170d), is the 2.5kV Cree driver still have benefits in the design, ie. of the electrostatics amps?

That one is wholly incompatible with electrostatic use. 

[JoaMat]

48 minutes ago, Tinkerer said:

So, I kept working on a few things on the right channel while I'm waiting on parts. Got all the balance and offset adjusted before hooking up the output board. Double checked all connections by hand and with multimeter. Checked PSU's, the whole 9 yards. Powered up the output board to check it and adjust current. I started it down low compared to last time. It lasted all of about three seconds before blowing (blew right as it hit 25mA). Same deal, left 120ohm resistor and both mosfets exploded. This time however, it didn't take any of the HV900 PSU's with it.

This didn't happen when the offset was really high, and like I said the left channel output board exploded when I was adjusting the offset down to zero as well. After what JoaMa said, this is starting to look like a board problem. Something to do with moving the feedback to the driver board maybe? Is 3W not a high enough rating for the 120ohm resistors? I could really use some help.

Sorry for your second failure. After my own failures I've been like gun-shy dog – haven’t touched the Circlotron project for a year.

But there must be a way to get the Circlotron safe even for the common DIYers, mustn’t it?

[spritzer]

There could maybe be some small improvements here and there but the over all nature of this amp would be classified as "tricky".  Super high voltage plus the PSU being an integral part of the output stage will always make it so.  I did study ways of making the amp safe to use for anybody but in the end it just wasn't worth it. 

[Tinkerer]

But could you explain why the output boards ran fine for almost half an hour at +450VDC offset but the minute offset drops below +15VDC they explode.

With feedback moved to the driver board, the feedback terminals on the outboard board are unused. Does the feedback setup have something to do with it?

[spritzer]

Could very well be.  Hard to know exactly what is going on though...

[JoaMat]

I’m actually a bit surprised that it didn’t explode at +450V. From memory - according simulation I’ve done in LTspice  I think it will all explode at +450V and probably even at +15V. The unsuccessful work I’ve done on the  Circlotron is with the first generation servo.

[G600]

My god.

This DIY is becoming dangerous !

Tinkerer, be careful, I wish you fast success.

[Tinkerer]

3 hours ago, spritzer said:

Could very well be.  Hard to know exactly what is going on though...

Maybe I'm phrasing my question wrong. What modifications need to be done to the Driver board in order to work with the V2 Output board?

Mine is currently set up like what soren posted on page 3 minus the parallel 332ohm resistor to increase current and the ballast resistors on the driver output since it's going to the output board and not headphones. No feedback connection between driver and output board.

[fat joey]

On 2015/7/3 at 7:48 PM, kevin gilmore said:

you need

4 x 900v floating power supplies (could split into singles if it makes more sense)

2 x 450v ground referenced power supplies (likely the fat power supplies)

2 x 15v ground referenced power supplies (on one of the fat power supplies)

1 x bias  (on one of the fat power supplies)

 

Regarding those big caps on the fat HV carbon board and the circlotron 900V board

Here's the list per ear

fat hv
550v 680uf  *2  in parallel = one 550v 1360uf

circlotron
470uf 550v  *2 in series = one 117.5uf 1100v
470uf 500v *2  in series = one 117.5uf 1000v
here's one set, per channel needs two sets
---> and then = two sets = two 117.5uf 1100v + two 117.5uf 1000v (these four caps are separated)

I met a resident in my building, according to him, he owns a small house to produce caps, consumed caps up to 2.8kv whatever normal value, is it a good idea that I order
two 550v 1360uf
eight 117.5uf 1100v
for the circlotron? Now it's 10 caps instead of 20

 

On 2015/8/27 at 0:45 AM, kevin gilmore said:

10 ma for the driver is more than enough.

needs to be 3 transformers, 2 of which have 4 windings of 380vac each

the other is the kgsshv-carbon transformer

 

 

Shouldn't be 360vac?

[Tinkerer]

Pretty sure the cap values were selected for what was available and affordable. That way you could get all the caps for this for just a few hundred bucks.

 

In other news, was testing some things and realized all the HV900 PSU's are blown. The 3 other don't look it, but they're only spitting out about +60-200VDC. Also my solid state relay locked closed even though it's rated for 25amps and bolted to a heatsink.  I'll be replacing stuff for awhile before I can do any tests again. Will have more questions soon.

[JoaMat]

Just curious. Did you use back to back zeners on the output mosfets suggested by Soren and Kevin? Also what value of the current sense resistors - 5.1 ohms or 12 ohms for "further protection”?

[Tinkerer]

On 2/6/2017 at 8:24 AM, JoaMat said:

zeners on the output mosfets

That's incorporated into the V2 Output boards which I used

 

On 2/6/2017 at 8:24 AM, JoaMat said:

5.1 ohms or 12 ohms for "further protection”?

3W 12 ohms on the HV900 supplies. Still exploded anyway.

 

There's a few things I'm probably going to fool with once things get repaired. One is that the output stage bias is required to be shorted to ground in the modded driver board, so that may affect something. I still haven't gotten a clear answer to if the feedback connections to the V2 Output board are necessary and what alterations to the V1 Driver board are necessary to work with the V2 Output. Another was I didn't use a variac to power the output board up separately from the rest of it so in further testing, hopefully I can find out if something is wrong or unbalanced before anything gives out.

 

EDIT:Since this is going to take awhile, I went ahead and hooked it up like a regular Carbon with a set of ballast resistors so I'd have something to listen to while working on everything else. Sounds really good, better than any other amp I've owned. The precision pot works good too, and the bypass switch works fine. So at the least, I got a big fat Carbon that takes way too long to warm up.

Edited February 8, 2017 by Tinkerer

[Tinkerer]

Had a couple issues I needed to ask about. Even with 100ppm/C resistors, drift from cold start to fully warm is about 5 volts in balance and 20 volts in offset. Same for either channel. Considering how sensitive the output stage is, doesn't that mean the servo is necessary and now there's no way to hook it up with the V2 output board? Or is the behavior because my driver boards are in some way damaged from the output boards blowing? They still seem to work and sound fine.

 

EDIT: Or is much tighter resistor matching needed

Edited February 8, 2017 by Tinkerer

[JoaMat]

You can’t get the servo to work? The opto servo on the normal Carbon controls the offset perfect – no drift.

[spritzer]

20V of drift is nothing to worry about. 

[Tinkerer]

55 minutes ago, JoaMat said:

You can’t get the servo to work? The opto servo on the normal Carbon controls the offset perfect – no drift.

Somebody correct me if I'm wrong but I think you need the opto input from the Output V1. V2 doesn't have that.

 

48 minutes ago, spritzer said:

20V of drift is nothing to worry about. 

Thanks.

[sorenb]

43 minutes ago, Tinkerer said:

Somebody correct me if I'm wrong but I think you need the opto input from the Output V1.

you can tap it off the drains of the SiC's at the driver board using some resistors.

20V offset will whack the bridge out of balance and raise the current flowing.
The zeners at the gates provides some protection in that regard, but I suspect that those are slower in comparison to the SiC's and that is the culprit.
I suspect the HV900 current limiter isn't robust enough for being literally 'shorted' for a very brief moment when the driver at power up/down swing to B+

[JoaMat]

On ‎2017‎-‎02‎-‎04 at 6:15 PM, Tinkerer said:

Maybe I'm phrasing my question wrong. What modifications need to be done to the Driver board in order to work with the V2 Output board?

Mine is currently set up like what soren posted on page 3 minus the parallel 332ohm resistor to increase current and the ballast resistors on the driver output since it's going to the output board and not headphones. No feedback connection between driver and output board.

Maybe I got it all wrong but I think that the modification suggested by Soren, beside onboard feedbacks, is an onboard offset servo. My idea is to have a fixed offset at zero. Then you can concentrate on the outboard (version 2) and its PSUs, preferable with help of a variac. From my experience (very limited though) is if something goes wrong neither zeners nor current limiter will save you from flying black fragments.

If you just are looking for some excitement you can try joggling three bottles of nitroglycerine.

Edited February 8, 2017 by JoaMat

[Tinkerer]

On 2/8/2017 at 1:50 PM, sorenb said:

I suspect the HV900 current limiter isn't robust enough for being literally 'shorted' for a very brief moment when the driver at power up/down swing to B+

How would you fix this? Bigger resistor? Two stage startup? But for me the outputs blew after around three to five seconds. It wasn't immediately at startup.

 

On 2/8/2017 at 4:17 PM, JoaMat said:

Maybe I got it all wrong but I think that the modification suggested by Soren, beside onboard feedbacks, is an onboard offset servo.

I would say you're probably right, but it doesn't work on mine. Offset drifts all over the place. Like he said, maybe the diodes. Will try removing all that stuff and paring it down to just the couple feedback resistors.

 

Also, I've been puling busted parts and on the Output V2 boards it's the exact same failure. Both mosfets, left 12V diode and current resistor are wrecked. Maybe that tells something to somebody who understands this better than I do. Still sorting out the HV900 supplies, but on the one that actually exploded, at minimum the current resistor, the 1000V diode by the voltage reference, the 10M90S on the same trace as the mosfet, the mosfet, and the 100 ohm resistor.

 

EDIT: Took out all the servo mod stuff and just left in the feedback resistors. Offset drift isn't all over the place anymore and balance is better.

Edited February 10, 2017 by Tinkerer

[JoaMat]

Great that you got a more stable driver by removing the servo.

After a closer look at what you have done I think the servo needs to be further modified to work. My experience is that the opto servo on the KGSShv Carbon works perfect. I use the same opto servo on my SMD Grounded Grid except I changed pzta06 to a to-126 device and removed the trimmer and its serial resistor and that servo works without any drift – high voltage 400, 450 or 500 volts, output current 10 – 20 mA still no drift and offset fixed at zero volts.

The driver will probably work as you have it now. Let it get warm and stabilized before turning on the output for the first time for adjusting.

One might consider to have the driver on and working while turning on/off the output…

[sorenb]

On 2/9/2017 at 6:24 PM, Tinkerer said:

How would you fix this? Bigger resistor? Two stage startup? But for me the outputs blew after around three to five seconds. It wasn't immediately at startup.

A Resistor capable of withstanding more current briefly. A more bold and faster transistor in comparison to the tiny 1-watt BC-thing

As I have stated on page 3, powering up the driver and let it settles before dialing in the output bridge will bring the Albatross up in the Air; landing is likewise a bit delicate, and turning off the HV900's before the drivers works.
I suspect that the 120ohm's also needs to be of a more bold character to prevent this thing for going boom - maybe 100W bolded to the heatsink. 

[Tinkerer]

I had been looking at 1N5349BG for the new 12V diodes. What do you look at for speed?

 

EDIT:Also, something like this for the resistor?

http://www.mouser.com/ProductDetail/ARCOL-Ohmite/HS100-120R-J

Edited February 12, 2017 by Tinkerer