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

[Pars]

If these are the lt1021 refs, you can buy these direct from linear for quite a bit less than Digikey...


Sent from my iPhone using Tapatalk

[sorenb]

1 hour ago, Tinkerer said:

... As far as I could tell, the things that blew were consistently.

...

Might save someone a bit of time if something goes wrong.

is it a description of what you needed to repair after the amp blew up the last time? I guess before you replaced the ½W with something more bold?

 

[Tinkerer]

Yeah, didn't finish pulling, testing and cleaning everything until now. New 12V diodes are 5W and along with the new current resistors should beef that output board stuff up considerably. Plus being able to turn that on second after giving the driver boards a bit to power up should help too.

 

56 minutes ago, Pars said:

If these are the lt1021 refs, you can buy these direct from linear for quite a bit less than Digikey...

I got them from Arrow. With the shipping difference it's about the same for small quantities. But thanks for the tip on the direct line. Might need that for a 7V ref for a different project. Those things can be a pain to find in stock at places for cheaper prices.

[kevin gilmore]

soren came up with this to limit the output current of the hv900 supplies

http://www.mouser.com/ds/2/205/DS100183B(IXTA-TP-TH6N100D2)-368848.pdf

put one on a heatsink and adjust to limit output current to say 50ma

[Tinkerer]

Just a couple small questions. What type of case is best for the device for how much heat it's going to be shedding? TO-220 or TO-247? What value for the trimmer pot? Can we get a quick diagram of the circuit modifications with all alterations on the HV900? I just want to make sure I'm doing this right the first time. 50mA should be .6V across the 12ohm resistor. And about what's the minimum AC voltage on a variac you can set the current for this and the output boards before powering them up all the way?

[kevin gilmore]

no, not a modification of the power supply, that current limit is before the regulated cap.

talking about adding a to-247 on a heatsink with a single resistor. since these are not sharp cutoff devices, the value of the resistor is unknown, would have to test it with lower voltages say 30v. but about 3 volts across the resistor so about 60 ohms

The adapter is added on the + side of the power supply line. If it works which I suspect it will, can add it to a new version of the board, but the board might have to get bigger.

[sorenb]

3 hours ago, Tinkerer said:

Just a couple small questions. What type of case is best for the device for how much heat it's going to be shedding? TO-220 or TO-247? What value for the trimmer pot? Can we get a quick diagram of the circuit modifications with all alterations on the HV900? I just want to make sure I'm doing this right the first time. 50mA should be .6V across the 12ohm resistor. And about what's the minimum AC voltage on a variac you can set the current for this and the output boards before powering them up all the way?

[Tinkerer]

Won't that need to be a 1kv'ish rated resistor though? Or were you guys thinking more of putting a couple smaller ones in series?

Edited April 27, 2017 by Tinkerer

[Tinkerer]

Bumping this because I got my HV900 supplies all back up and running and need to make my last parts order to do some testing soon. I need to know the general voltage and current rating for the 60ish ohm resistor on the new current limiting adapter.

Now that I looked at it closer, I think I get it. I didn't see the normally open part of the mosfet before.

Also, if this works, I would assume the big resistors are not needed on the output boards. Though since I already have them, I'll probably install them anyway as long as I can find the sink space.

[sorenb]

5 hours ago, Tinkerer said:

Bumping this because I got my HV900 supplies all back up and running and need to make my last parts order to do some testing soon. I need to know the general voltage and current rating for the 60ish ohm resistor on the new current limiting adapter.

Now that I looked at it closer, I think I get it. I didn't see the normally open part of the mosfet before.

Also, if this works, I would assume the big resistors are not needed on the output boards. Though since I already have them, I'll probably install them anyway as long as I can find the sink space.

60ohm -> RN60D or similar
If you bought the mosfet Kevin mentioned you need to flip source/drain, as it is an n-channel
Also it has exposed drain on the backside - ceramic isolation and washers that goes all the way through is needed

[Tinkerer]

Gonna test the current limiter on the HV 900's today. Just double checking that this is the right setup. If it works, I should have at least one channel up and running today.

 

EDIT: I'm just trying to puzzle it out. Diode direction for an N channel mosfet should be Source -> Drain, right. Then the resistor is 60 ohms to get a voltage drop of -20VDC to activate the depletion mode gate at around forty something miliAmps to give some buffer room for the device to fully close before the 50mA limit on the output boards.

 

I know this is simple stuff, but when it's a hundred bucks in exploded parts and worse, the time to pull everything apart, test it and fix it up again, I just want to be 100%.

I had one other question though. The output stage bias trimmer on the driver board. We keep that shorted since the two pots on the Output V2 control the bias current directly now? And is the extra DC offset resistor still required if feedback is run from the output board like the silk says instead of from the driver board?

 

EDIT 2: Nevermind, I get it now. I need to swap them around. The body diode is supposed to point back the opposite way to act as a suppressor.

Edited May 15, 2017 by Tinkerer

[sorenb]

On 5/14/2017 at 3:19 PM, Tinkerer said:

Gonna test the current limiter on the HV 900's today. Just double checking that this is the right setup. If it works, I should have at least one channel up and running today.

 

EDIT: I'm just trying to puzzle it out. Diode direction for an N channel mosfet should be Source -> Drain, right. Then the resistor is 60 ohms to get a voltage drop of -20VDC to activate the depletion mode gate at around forty something miliAmps to give some buffer room for the device to fully close before the 50mA limit on the output boards.

 

I know this is simple stuff, but when it's a hundred bucks in exploded parts and worse, the time to pull everything apart, test it and fix it up again, I just want to be 100%.

I had one other question though. The output stage bias trimmer on the driver board. We keep that shorted since the two pots on the Output V2 control the bias current directly now? And is the extra DC offset resistor still required if feedback is run from the output board like the silk says instead of from the driver board?

 

EDIT 2: Nevermind, I get it now. I need to swap them around. The body diode is supposed to point back the opposite way to act as a suppressor.

nice drawing. 

assuming the fet Kevin pointed to: the RN60 goes between gate-source, and switch the text of input/output(like you mention in your edit) and you should be good to go.

[Tinkerer]

Bumping this as I got some more multi KV rated silver plated wire in, enough to give all this a good test in a day or two. Had one of my reconstructed HV900 supplies blow under stress testing, bad resistor I missed I think, so will only be testing one channel. I'll let you guys know how it goes. It wouldn't be Fourth of July weekend without some fireworks after all.

 

If anybody else has been working on this, now would be the time to pipe up if the latest current limiting mosfet additions worked for you or didn't.

[sorenb]

2 hours ago, Tinkerer said:

Had one of my reconstructed HV900 supplies blow under stress testing

How do you "stress testing"?

2 hours ago, Tinkerer said:

bad resistor I missed I think

Which resistor do you suspect being the culprit?

[Tinkerer]

Just loaded it with proper resistance the same way you did way back at the beginning of the thread. Looked like one of the CMF60's (the 100R next to the 1N4007 diode) had lost a chunk of insulation on the underside only and I didn't see it. Either that or the new SMD STN9360's were garbage. It's the only PSU that needed new ones since I had enough salvaged of the others. Also possibly the socket for the voltage reference as that seemed to have melted and soldered itself to the chip. Hard to tell what was what gave out and what got damaged afterward, especially when one of the resistors burned and covered a good chunk of things with soot.

 

All the other three ran fine. This one got up to regulated voltage, lasted about ten seconds, then popped. All the solder joints were good and the board itself was squeaky clean before it got powered up.

Edited July 2, 2017 by Tinkerer

[sorenb]

9 hours ago, Tinkerer said:

Just loaded it with proper resistance the same way you did way back at the beginning of the thread. Looked like one of the CMF60's (the 100R next to the 1N4007 diode) had lost a chunk of insulation on the underside only and I didn't see it. Either that or the new SMD STN9360's were garbage. It's the only PSU that needed new ones since I had enough salvaged of the others. Also possibly the socket for the voltage reference as that seemed to have melted and soldered itself to the chip. Hard to tell what was what gave out and what got damaged afterward, especially when one of the resistors burned and covered a good chunk of things with soot.

 

All the other three ran fine. This one got up to regulated voltage, lasted about ten seconds, then popped. All the solder joints were good and the board itself was squeaky clean before it got powered up.

I never had troubles with the HV900's, and haven't had any that blew, so the following is based on what I've found useful troubleshooting/checking the regular Golden Reference HV.
Often it can be helpful to work at lower voltages when trouble shooting. Paralleling the high resistor (is it 2x 450kohm in the HV900?) in the probe string to get the circuit in action at a lower output DC. If regulation doesn't work for some reason and things go bad it usually isn't really bad.
Parallel the R8(2x 450k) with one ~100k (one across both) will provide some ~100VDC.

Initial check:

1. The CCS's
   Use a lab supply at some ~15-20V, through a DMM at mA: put the (+) at the drain (middle leg) of the 10m90, and the (-) at the bottom of the CSS where R2(100) and R1(4k5) meets ...you should get ~2mA ....if you see much more, try replace the DN2540.
   Do the same at the other CSS: (+) to drain (middle leg) at the 10m90, and (-) at the bottom (pin 6 at the voltage reference) ...you should get ~1.3mA ...if not the 10m90 is probably damaged. 
   Check the current limiting CSS you added (your drawing) the same way (Kevin suggested 50mA for this)

When powering up the supply, hold the variac when you reach some ~40-50VDC output, and check: 

1. the pin6 of lt1021 against "ground" (pin4) ....you should see 10V ...if not, the lt1021 has probably gone south
   Before raising the VARIAC further:
2. Attach a DMM across the 2k at the bottom of the cascoded CSS ... you should see ~0V (when at ~50VDC output) and only before regulation kicks in, the voltage jumps to some ~4V; it happens just before you reach the desired output voltage
3. Attach a DMM across the 3k at the source leg of the output CSS (right most leg) ...should be some steady ~4V

 

[s_r]

On 7/1/2017 at 6:29 AM, Tinkerer said:

Bumping this as I got some more multi KV rated silver plated wire in, enough to give all this a good test in a day or two. Had one of my reconstructed HV900 supplies blow under stress testing, bad resistor I missed I think, so will only be testing one channel. I'll let you guys know how it goes. It wouldn't be Fourth of July weekend without some fireworks after all.

 

If anybody else has been working on this, now would be the time to pipe up if the latest current limiting mosfet additions worked for you or didn't.

Where did you source the high voltage wire from? Probably won't be starting a circlotron build myself all that soon but it'd be nice to have beforehand.

[Tinkerer]

I got the 20awg PTFE stuff from naavships on ebay. It was the only multiKV silver plated stuff I could readily find. Unfortunately, only comes in one color so it's helpful to add some other markings to easily differentiate what goes where. Other stuff I just used the more standard silver plated teflon coated that's good to 600V or so. And I used two conductor silver plated shielded for the unamplified audio with the shield tied to ground.

 

Been bringing an old IBM mainframe terminal and it's controllers back online lately so haven't got around to testing the new stuff yet. I did put in an order for replacement parts for my HV900 though. Hopefully, I'll get around to more testing later this month.

 

Thanks for the troubleshooting stuff Soren, but there's so much energy in this thing, the 10M90S always gets toasted along with the 10V reference if anything else blows. The ones in the power supply that died were all brand new. I appreciate the slow power up info though. When I picked up my old Conrad variac last year, I wouldn't imagine I'd get so much use out of it on all sorts of projects, but it's become almost as indispensable as my multimeters.

[Tinkerer]

Okay, finally got a bit of time to work on this again. All the HV900 supplies are running in good order. The current limiter mods for them seem to work fine too. I have them bolted with thick ceramic insulators directly to the heatsink and they haven't had any problems with the voltage unloaded. Can't actually know for sure until the circlotron output stage boards are hooked up, but that's what I need the next few questions answered for.

 

So here's what I need to know. I removed all the mods from the driver boards except the couple removed components that disconnect the servo. Brought it back to basically stock. What is the value of the offset supposed to be on O+/O- coming directly out of the driver board before you get to the circlotron stage? Because without the mods to turn it into a carbon amp board, it's sitting at around +450 VDC offset. This is how it was way back when I first assembled it too.

 

What do you do with the output stage bias trimmer on the current source board? I assume the -15VDC from the Low Voltage supply to the V2 Circlotron Output replaces the original purpose but what do you do with the trimmer on the driver board? And how do you measure whatever you're supposed to do with it?

 

Is there any way to get the servo to run on the current driver board with some kind of simple mod to the Circlotron Output V2 board?

 

Also, a little side note. A Dale CMF60 4.32k resistor works as a drop in replacement for the Xicon 4.42k in the HV900 boards. Closer to +900 to +905VDC output too instead of everything sitting at about 910 to 915.

 

EDIT: Just realized that the "output bias" pot on the driver board is probably the current control for the output board but the V2 lets you do that on the output board itself now. Still need the answer to the question what to do with that trimmer.

Edited September 22, 2017 by Tinkerer

[sorenb]

you probably need to lower the tail resistor for the driver output ...or dial down the current ....anyways, you need 0 offset at output (Kevin recommends  ~8mA)

[Tinkerer]

Current is only set at 10mA on mine. Kevin said that as the upper value somewhere I think in here. But I can drop it down if that's the recommended value. I paralleled a resistor to the tail resistor before to fix this offset problem when I ran the driver board as a carbon amp. I can find new value that works based on that. But just to double-check, it is supposed to be 0 offset from the driver board to the output board? There's no other stuff there to fool with? Does the feedback from the output board affect the offset at all? I decided to run it that way instead of locally.

 

What about the output bias pot on the driver board and rigging input to the driver board opto servo from the V2 output board?

[sorenb]

not sure what you are up to, as far as I remember the silk value for the tail resistor wasn't far off. 10mA or 8mA isn't day or night  anyways.

[Tinkerer]

Kevin said he set it high intentionally in the original stuff so it would kick over the servo, but I agree that I don't think it was intended to be that high. All I know is a vanilla driver board with no connection to an output board has +455VDC offset with positive probe on O+ and negative probe on ground with a 500ohm (silk value) tail resistor. You can get the offset to 0 with a 120-150ish ohm tail resistor.

It was just so far off I had to ask if it was intentional for something in the output stage.

[sorenb]

lower the current or the tail resistor ...

[Tinkerer]

I am. Just wanted to be sure. With all the problems this amp has been, I'm beyond assuming what should be correct until I see it power up completely.

 

Edit:looks like the resistor value should be closer to 300ohm. 115 was with the increased current modification.

Edited September 23, 2017 by Tinkerer