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

[Tinkerer]

Still fiddling with this. Here's my issue.

 

The driver board lights up fine, current is set at 10mA (.5V across the 50 ohm both sides). Balance is within .01 volts. But the offset is still giving me fits.

 

EDIT:Got it working but the drift is like 150V every minute up and down as it warms up.  That normal without feedback or servo?

Edited September 30, 2017 by Tinkerer

[kevin gilmore]

I would expect that to be normal. no way it can work without the servo

[Tinkerer]

I really need to know how to hook the servo up to the V2 Circlotron stage. There's no opto terminal on that.

[Tinkerer]

So, I got a little impatient and went ahead and hooked up everything for one channel but didn't power up the circlotron stages to see if the feedback would help stabilize the driver boards.

 

It does, best I've seen. After about half an hour, balance had only drifted .5V and offset only 1.0V. Absolute night and day. Even better than the feedback mods to run the driver as a carbon. Might actually be able to power this thing up without a servo.

 

Since this thread hasn't had any good pictures in awhile. I can show how mine stands at least. None of the wiring has been prettied up. All the previous looming minus a couple small runs had to be removed for the new mods, especially with the current limiters. Same with the paste. Will clean it up better once I'm sure everything is good. I've pulled those brackets enough to stop fooling with it for now.

 

You can't see it very well but the transistors bolted to the sink have their legs bent about 45 degrees away from the sink and are mounted on thick ceramic insulators. The 62ohm resistor pairs that replace the 120 on the circlotron board are insulated and ceramic backed so are bolted directly to the heatsink. Their leads are soldered through the test point holes since the 120 resistor holes are too small for the HV wire.

 

Tail resistor for the offset on the driver board was replaced with a 300 ohm and the pot with a 200 ohm for more range. 10mA is also the lowest the current pots would go on the driver board, so anybody that wants lower will probably have to change some bits.

 

 

 

 

Anything that looks completely stupid and any good reasons I shouldn't be able to power up the output stages with a variac next?

[Tinkerer]

Went ahead and gave the circlotron stage a little juice with a variac with the drivers running so I could set the current. With 123 ohm current resistors, it should be 3.075V across for 25mA I think. With the pots maxed out, it was still running up into the low 40's mA with like 20VAC line from the variac. I didn't run it up any higher than that for fear of blowing something.

 

This another case where the resistors are too low? Should I try swapping the 100 ohm cmf60 tail with a 180 I have laying around or something?

[sorenb]

from your posts I am not sure exactly what you are doing, or how you have connected the various parts.
do you have ~zero out of the driver? and -15V present at the output board?

[Tinkerer]

Yes. You can see the -15 for the output board is that naked green wire that goes to the GRLV in the bottom of the case.

 

The way it's set up is

1.Servo disconnected on driver board (10k resistor and little parallel cap removed from driver)

2.All wires to correct terminals (Output FB to Driver FB)

3.Current limiting transistor setup on the +900 lines of the HV900. (NO and working, haven't pushed them into shutoff yet)

4.120 ohm current resistors on output board replaced with paired noninductive 62ohm resistors on heatsink (unnecessary if number 3 works but had already spent the money on them so figured why not)

5. Driver set to 10mA each side (as low as it will go), offset controls changed to 200 ohm trimmer and 300 ohm tail to make it zero out

Interesting thing. The balance and offset measure from the driver board or the jack on the other side of the output stage is exactly the same. I checked it twice to make sure, but measurements are a lot easier that way.

 

What I'm doing is running driver at full mains (GRHV450 and GRLV15 full up, Driver boards full up, -15 to Output boards) and then using a variac to slowly power up the HV900's for the output stage.

 

With the variac at app. 20VAC into the HV900 transformer, which is the HV900 spitting out about 400VDC IIRC (I really need more meters, three is not cutting it), the read voltage from the output stage test points is over 5.3VDC with the V2 current control trimmer pots at maximum resistance. Which is just about 43mA through the Crees on the Output board. And it just keeps scaling straight up.

 

So my guess, based on the driver trimmers that were a little off what they needed to be from what the silk said (no way I could get 8mA or such without swapping some values around), that the Output V2 boards have a similar issue. I just need to know which resistor and what value. Or maybe it's something to do with the old output stage bias trimmer on the driver board that I still don't know what I need to do with if anything. Short it to ground, leave it be, whatever.

[sorenb]

I believe you need to reconsider your approach a bit.

• consider having the FB local on the driver rather - makes things less complicated
• make the servo working at the driver (Kevin mentioned this a few posts back) - having the driver in balance with the servo working is key
   
• check the opto's on the output boards whether the emitters are pulled towards ground when the current in the bridge is rising (from your description it sounds like one or both isn't working).

 

[Tinkerer]

The V2 Output boards don't have servo output. And the mod you made to feed the servo locally with the diodes and 2M resistors way back on page 3 doesn't work. I've tried it before. That's why I've asked three times how to get the servo to work with the V2 boards, and I still haven't gotten an answer.

 

From what you're saying though, it seems like the cleanest solution would probably be to just take some actual Carbon Amp boards and jumper the headphone ballast resistor spots instead of even having specialized driver boards in the first place.

 

The 4N25's in the V2 Output are brand new as well. But I'll pull them and check both them and the sockets.

[sorenb]

Sorry, I haven't noticed that you have mentioned that the "recipe" didn't work ...
Anyways, I had a look at my notes from back then, and it seems I forgot to add a few things I did in addition to get it working; I've edit the post accordingly.

[Tinkerer]

Much appreciated. If I'm lucky, parts will get in Saturday and I can give an update.

[Tinkerer]

Still working on things but had an HV900 question, which soren asked a long time ago, but I didn't see answered. How close do the the paired HV900's have to be to each other and how close do the HV900's have to be to the HV450's +/- together.

 

I only ask because one of the HV900's I built with available Dale resistors (used a CMF604K3200FKR664 at a slightly lower value instead of the Xicon) regulates at +905VDC while the others are at +910VDC. GRHV's are +452VDC and -450VDC respectively.

[Tinkerer]

Got the servo working with Soren's new notes. Tried to run global feedback first and servo is sitting right at -4VDC to -5VDC. It gets exactly to -4 after everything warms up for a few minutes. Can somebody explain why it's at that instead of zero or what I should adjust to fix it? I made sure before I engaged the servo that offset was about +14VDC warmed up with global feedback since you're supposed to have about +10 or a little more to kick the OP27G on right?

 

Edited October 19, 2017 by Tinkerer

[sorenb]

is the OpAmp postive input grounded? - as I recall I shorted the trimmer

[kevin gilmore]

offset currents in the opamp, no opamp is perfect. If it likes to be at -4 leave it alone.

[Tinkerer]

Yeah soren, dialed to ground. And thanks for the advice on the opamp Kevin, makes me feel a lot better. I remember my old KGSSHV servo was kind of similar but with this amp I never knew if a few volts offset away from zero might make it explode.

 

Okay, got some good news. Got some music! Some problems with the sound but will get to that in a minute. Brought the Output V2 stage up slow on the variac to be able to set the current. It started working low level as soon as the variac hit about 10AC and made it to line voltage without anything blowing up or any big changes in behavior. I was kind of worried about how when you power off the driver stage you get +350ish offset until the PSU bleeds down, so I powered off the Output stage first and let the HV900's bleed down before powering off the rest of the amp. But I think that the added diodes in the Output V2 are specifically to counter this, right? So no alterations needed?

 

Here's the thing though, the current keeps skipping around a bit. 30-45mA, continuously, doesn't steady out. You'll also hear some static when it moves and the offset will jump a few volts. Kinda sounds like when I had a leaky resistor in my old SRA-12S through the headphones. Similar symptoms too. Static and a change in offset at the same time. That means the driver local feedback is necessary? And could there be a resistor problem? Could the variation be the resistors bolted to the heat sink not maintaining a steady temperature or not being isolated enough despite being an isolated package

 

Also, more good news. The current limiters work. Any time the board was about to bust 47mA from a current jump, the transistors would close and it would take a second or so for them to open back up all the way.

[Tinkerer]

Changed to local feedback on the driver board and that leveled the current out on the output board. Now sits at 4.7V ( app. 38mA) consistently at lower level on the variac and can be adjusted but that's as low as the pots go. Current still starts jumping around and offset as well, once you bring up the AC into the HV900 transformer up to about 50VAC on the 115 primaries. Current swings are worse at those higher levels with the local feedback. I've seen 12V across the 123ohm resistor for a split second before the current limiters kick in.

 

I was getting serious 60hz hum I think from the variac. I could actually feel the coil vibrating through the case. GRHV/GRLV transformer is dead quiet, so I assume the HV900 toroids should be good too. Couldn't feel or hear any vibration in the amp case.

Troubleshooting tips at this point? Output boards have brand new 4N25's from my spares. The Cree's on the output are salvaged from old HV900's that didn't blow them when they died. They showed up fine on my tester but could they be damaged and that's why the current is jumpy at higher voltages?

 

Otherwise, same problem as before. Big static pulses that coincides with a jump in current and offset after output boards are about halfway powered up.

Edited October 21, 2017 by Tinkerer

[Tinkerer]

Holy crap, finally got this thing running right after like two years. My last screwup was actually having the multimeters reading voltage at the test points for the current on the output stage as I turned the supply voltage up. The leads were leaking or something even though they were supposed to be rated for that voltage. So when they were getting close to breakdown is when the noise and meters would start acting up. No multimeters attached except for balance and offset at the jack and it powered up perfectly. 60hz hum was entirely from the variac too.

 

Listened to it for about fifteen minutes, left channel as regular carbon (driver + pair of 5k ballast resistors) and right as circlotron, and was impressed I could actually hear the difference. I was just hoping no noise problems at this point honestly.

 

A couple minor notes on stuff that might be helpful for anyone else.

In Soren's servo notes, the 5K resistor is the 500 ohm. Probably a 5000 code RN60 or something was the confusion I guess?

I just bent the #2 opto leg up and away from the socket and soldered the 200ohm resistor to it directly so I wouldn't have to cut any traces.

Everything else went together easy just as written. Almost everything has an open through hole slot from pulled parts or an easy place to hook up under a terminal.

 

While the servo keeps the driver at about dead 0 balance and offset alone, through the output board it sits at about -5VDC offset and .2VDC imbalance. But it keeps it right there so I haven't fooled with it like Kevin recommended.

 

Other than that, since last time I put new DIP sockets in the Output board, moved the sink mounted current resistors to a bracket on the Output board to shorten the wires and reduce the parasitic capacitance and make the output board easier to pull and service. Heat dissipation isn't as good as being directly on the sink, but still plenty I would think.

 

 

Should hopefully finish all the stuff for the left channel tomorrow and see how smoothly things go when the output stages are turned on at line voltage from the internal relay instead of slow variac power up.

[Tinkerer]

Well, some good news some bad news. Did the left channel and slow powered it as well and it worked fine. Then I got everything all hooked up to mains through the relay.

 

Fired it up two stage like you're supposed to. Main switch in the back powers on the GRHV/GRLV and PSU for the LED's and relay and lets everything warm up for a minute. Front panel engages the HV900's for the Output and lights the LED's up.

 

The left channel struggled a little bit but came around. Right sounded like it was struggling, then went severely out of balance and blew. Looks like the right side opto couldn't take the fast power up. Literally blew its top off.

 

Good news is this is the first time I've ever had an Output board blow where both the HV900's are still perfectly intact and functional. Due to the current limiters I guess? That makes them worth the cash right there, a lot better when you only have to rebuild one board instead of three. Going to try buying higher rated DIP sockets and look if there's more robust 4n25's to pick from.

 

Any other solutions? Should I try a themistor or something to force the Outputs to power up more slowly? Like I said, they did work when it was slow and they had time to warm up and let the servo keep up. I just need something that works for a daily amp. It would be silly to have to power up half the thing with a variac every single time I turn it on, though I guess there's room in the case to panel mount one if it really is the only way.

 

Also, I hadn't realized just how much power this thing eats. I was originally going to run it through my 350watt UPC and it threw that into overload without anything else hooked up. Had to actually run it from the wall.

[sorenb]

@kevin gilmore would applying  -15V gradually from 0V, make it easier for the opto´s to balance the bridge?

[kevin gilmore]

I always bring up the low voltage power supplies first. and turn them off last

[sorenb]

57 minutes ago, kevin gilmore said:

I always bring up the low voltage power supplies first. and turn them off last

...and does that also apply to second edition of the output board?  ...wouldn't the sources be at -12V until the HV900 is applied?

[Tinkerer]

Yeah. The Output stage stays -12 and gates closed until you start to apply the HV900's. Then moves to about -10 then creeps up steadily with the applied voltage until about -5 where the servo keeps it.

 

Well, I tried a small workaround. Got a couple of old unmounted Superior Electric 120VAC 270VA Variacs for cheap and hooked one up for each channel with a temporary cross bar mount. They're small and good quality. Also swapped completely to nylon screws for the Crees on the Output board and added additional isolators to the current resistors as well as swapped the 4N25's for the newer fairchild 4n25-X001 that handles double the current.

 

Seemed to be working fine, but had one of the HV900's start to smoke after a minute. Nothing actually failed Output board wise. I guess that HV900 was damaged after all from the last time that Output blew. Just couldn't tell until it was under load. Gonna have to rebuild that before I try again.

 

Had another important question that might relate to problems. Are the little Stax jack boards rated for this? Because even when I'm powering up a single channel, the noise from the output board bleeds into both channels, and I don't get if it's the jack or the ground or what. Also, what is the sound of the Output supposed to be like as it's powering up? This is how it seems to me, and was this way when powering from the external big variac.

 

5VAC, quiet but can hear distortion compared to plain carbon

10VAC, regular 60hz hum, can hear repeating pulses in it

10-40VAC, hum gets louder, pulses closer together until they become a continuous single noise, noise equal in both channels despite power only coming from one

40-50VAC, hum goes mostly away, but still there (this would be when the HV900's are hitting about 450VDC)

50-80VAC, hum returns but not as loud as lower voltages

90VAC+, clear but slight underlying hum (HV900's about at full voltage)

 

Does that line up with you guys at all?

[kevin gilmore]

not sure which stax jack board, but if the traces are more than .1 inch away they are good for 1kv

now if there is too much flux, then who knows

 

Edited November 7, 2017 by kevin gilmore

[Tinkerer]

Works fine at carbon voltages. Barely any flux but it's also hard to clean off as it had to be soldered with the sockets already mounted inside.

 

But then do the output to socket wires need to be 1kv rated? Because they're just 600VDC teflon insulated.

Edited November 8, 2017 by Tinkerer