JoaMat

Lowered the resistor value at OP27 output. Now both offset and balance are in the mV area. Very happy with the outcome.

Can you put that in a high altitude extended holding for a while? Need to save fuel. Almost finished the layout for the other channel. This is the first board layout I’ve done from schematic in the software. By doing it this way I hopefully can avoid mistakes in the process of copy, mirror from the right channel board to the left … Thanks Kevin, Kerry and others for new ideas that I can exploit.

Your wife is most certainly right, but to be five years you are really good.

Happy birthday!

Welcome MattN, great work. The offset servo is a new version and I'm not familiar with the board you are using. I guess the offset servo is working and reason the trimmer seems not affecting the offset might be because the resistor in series with the trimmer has a bit high value? One way to disable the servo might be to remove the ptza06 next to the 12V Zener diode.

Some progress. The terrible unbalance was due to an unconnected small tube pin. During the milling process the CNC machine drilled a couple of unauthorized holes which I completely forgot when soldering the board. When corrected the amplifier works. The offset is done by opto servo and it works as supposed. Balance control is also done by servo and I've no trimmer. I've same balance servo here as in my amplifiers with LSK389 input and there it works without any problem. With the tube input you can probably expect more differences between the sides. First tube I tried gave 260 V unbalance (offset worked, so +130V and -130V) and the second tube gave 30V unbalance. The balance servo (in current version) is not capable to make it balanced if the tube isn't enough matched. The input is supplied by +100V/-15 and all components there are surface mounted. Current source is 01N100D/LT1021. GG to the left and tube input GG to the right.

No blowing up - no fun. One HV up, other HV down - so no fun there either.

Did this today. "Twin holes" along the right side of board - forgot to mirror before drilling the board, but this is a prototype and it will probably blow up anyhow…

It was actually very easy to rebuild the large T2 with the DC/DC converters. Used same daughter board layout as for as for my modified T2, only had to move one hole 50th. I also have plenty of space since I only use one and not three output transistors and I found some nice spots for the daughter boards. One channel has the DC/DC converters and the other the newly made 01N100D/LT1021 boards. Both channels seem to work OK. Below: Daughter boards with LT1021 and 01N100D on heat sink. Boards with DC/DC converters and LT3045. Now it's late, done for today.

That’s very tempting. I’ve to figure out how that is best done on the original board.

Today I’ve equipped my modified T2 with the 01N100D/LT1021 current source Kerry has simulated, see previous posts. From beginning it had a straight 10M90S set to 10 mA. When JimL came with the cascaded version I removed the two resistors at 10M90S and added a small board with the DN2540 and corresponding resistors(red circle below). Today I made new daughter boards containing LT1021, 1k resistor and 47pF cheremic capacitor (green circle). Have listen to this for an hour. It works and I’m satisfied. Thanks Kerry for coming up with this. Below solar cell. Only for experimental purpose.

Megatron is probably 120 mA, that's plus 50 % compared to your Carbon.

01N100D with LT1021 to the left. Got the idea from Kerry. LT3042 is a nice piece, but 1021 is a bit bigger and then easier to deal with.

emitter only needed during night and cloudy conditions?

Ops! Kevin pointed out the resistors in this post. Sorry.

Nice picture. If a1486 had been a kV device the T2 current source would be a better one? Most of the variation of the green curve must be due to the 150k resistors.

I do believe that the T2 current source is the one to aim for. But that’s only for the squirrels and the boys with big pockets filled with c4686. If you want a CCS based on components from Mouser or equivalent then I think 10m90s is good enough for most people, including myself. But we should look for alternatives for 10m90s so in that respect I’m grateful to Kerry’s work. I’ve had a T2 with c4686 and isolated switching power supplies for a while and I really like that amplifier. Probably the cascaded current source for driving the LED is preferred but I like the isolated things. Maybe I should try current source for driving LED, anyone have a schematic?

Two current sources. One “normal CCS” and one “Kerry’s in progress regulator CCS”.

Thank you, good to know.

Interesting current source. Would 01N100D/DN2540 cascoded work (in LTSpice)?

It seems that G3 is tied to the plate on kgsshvtubesandwich while G3 is tied to the cathode on BH and GG. Any particular reason?

Agree, very impressive. Can't be done much smaller.

It has been a rainy day so I modified the batteries on one channel. Replaced the four 2sa1486 with two 2sa1968. Battery voltage, offset and balance just fine but sonically I got unpleasant hum. Now everything is back with a1486. Why change something that works well?

Thanks a lot. Good to know I can save my small stock of 2sa1968 for more important missions.

I've rebuilt a couple of batteries on original DIY T2 boards with LT1021 - works. Have looked for schematic without success but it is somewhere on my computer… Anyhow, how do you like this one: LT1021 current set to 1.2 mA. Hard to find 2SA1968 and easy to find smd HN4C51. Fewer components then the original battery. Better? probably not.