stax t8000 clone (well sorta)

[kevin gilmore]

Well its not like the stax mafia aren't busy

and we certainly are not going to do an output stage with 8ma of current, and therefore no reason for a 4 stage amp, and since tube inputs seem to be the big thing lately, well might as well take the kgsshv-carbon design and replace the input fet with an input triode. bonus, no low voltage supplies.  not a bonus, ksa1156 and ksc5026 run at close to 1 watt each.

I should have a real t8000 end of june, but this should keep some of you busy. I expect joamat to have one built in less than 24 hours.

no, no circuit board yet

about 70 volts on the plate of the tube, same as t2.

optical servo from kgsshv-carbon will work fine (but then you need the low voltage supplies)

 

kgsshvcarbontubeinput.PDF

Edited May 12, 2017 by kevin gilmore

[JimL]

Gee, and that took less than the 3 years of development time that Stax says they used for the T8000.  Now the next step is to change to GG EL-34 outputs, and you'll have a modernized, simplified T2. 

Edited May 12, 2017 by JimL

[kevin gilmore]

I could publish what I think the t8000 is, but take my design, and replace the output stage with stages 3 and 4 of the srm727 and it's going to be real close. Probably a stacked current source in stage 3 with current production transistors.

and yes, you can replace the output fet with a tube in addition

more fun ahead

[joehpj]

Ahh... light speed. Hat off to Kevin. 

[mwl168]

I need a bigger house for all the KG ES amps!

[kevin gilmore]

You need a bigger house, hell I need a much bigger house

[chinsettawong]

Keep them coming.  I'll buy a bigger house.  

[joehpj]

I wonder if it's possible to add a voltage dropper like megatron did to lower 1156 and 5026 Pc and push the output voltage a little bit further? Say 475V~? ( I assume 10M90S+DN2540 can take 950V in total to be safe.) Or even -500V since EL34 output will need approx 35V across the tube.

Edited May 12, 2017 by joehpj

[JimL]

1 hour ago, kevin gilmore said:

I could publish what I think the t8000 is, but take my design, and replace the output stage with stages 3 and 4 of the srm727 and it's going to be real close. Probably a stacked current source in stage 3 with current production transistors.

and yes, you can replace the output fet with a tube in addition

Yeah, my guess was a 6DJ8 diff amp into the rest of the SRM727.

1 minute ago, joehpj said:

I wonder if it's possible to add a voltage dropper like megatron did to lower 1156 and 5026 voltage and push the output voltage a little bit further? Say 475V~? ( I assume 10M90S+DN2540 can take 950V in total to be safe.)

There is only a few volts across the DN2540 (the 10M90S/DN2540 cascode basically runs the DN2540 at constant voltage), so I wouldn't push the combination past +/-450, and would want to have it somewhat under to allow for over voltage on the AC line.  Anyway, the difference between, say +/-425 volts and +/-500 volts is less than 1.5 dB in ultimate output.

[Whitigir]

So are we going down the road of an improved And simplified T2 soon ?

[kevin gilmore]

A simply regulated +200v rail will reduce the power, same with the input current source. Then up to 450v rails

what I published I know works, now time to perfect it.

[spritzer]

It's been a long while since we did the simplified and improved T2 but it won't sound like the real deal.  Better yes, but not the same. 

[jose]

Yes, I need to buy a bigger house... and a nuclear power plant too.

[Ali-Pacha]

Something to follow, for sure

Ali

[JoaMat]

I want the output current source to be c4686 with DC/DC converter.

[nopants]

did you figure out a way around the inherent noise?

[JoaMat]

No, I haven’t. I guess it’s some sort of EMI. One question is how to stop it and another is how it effects the quality of the amplifier. Now I’m most interested in if the DC/DC converter can be used at all. One T2 is modified with DC/DC converters and it seems to work alright. The idea with DC/DC converter is IMO absolutely worth being tested.

[joehpj]

I googled for some substitute for c4686 which has been obsolete. And found MJW18020 seems to be a nice current production one. 1kv Vceo with much higher Pd makes it easier to do 500V 25mA. Any thoughts? 

Edit:No. output capacitance way too high.

Edited May 13, 2017 by joehpj

[kevin gilmore]

i'm not sure what the mjw18020 is designed for, but for dc use, its going to blow up over 450v

 

[JimL]

20 hours ago, spritzer said:

It's been a long while since we did the simplified and improved T2 but it won't sound like the real deal.  Better yes, but not the same. 

Can you be more specific?  Better in what way?  Thanks in advance.

[spritzer]

Better CCS, batteries work now and it is a shit ton less parts.  Also far less noisy which is a large problem with the T2. 

[mypasswordis]

Need to get around to swapping out some 10m90s for breadboarded cascode one of these days. Bought a hand tap that actually works, which makes things easier

Gotta add the t8000 clone to the loooong queue too....  

[Kerry]

I've been playing a bit with the design.  Trying to keep it to only +/-450V rails + filaments.  I also wanted to see if I could keep the servo.

Here's what I came up with.

 

The 6DJ8's anodes are sitting at about 67V.  I think this works well, but could easily shift the operating point for these.

Q9 & 10 are dissipating about .46W, but we could do a thermal pad on the top of the board and should be good 

There's still some tweaking on some of the parts.  Also, thought to do the 47 ohm hum bucker resistors to ground on the filaments. 

I've gotten the 200V supply down to about 5mA so not to bad on the pass fet (1.25W).  I would embed this with each channel and mount to the heat sink.  Could probably use FQPF8N80C here instead.

The servo uses the HV opamp we use in the PS, etc.  I'm running about 1mA though it.  Models well, but need to test this. I've also brought the bottom up with the 400K resistor so we could use lower voltage transistors in it.  Right now is sits at about 40V between R34 and R9.  The output is sitting around -4V.  I'll need to play with this a bit more, though this isn't a real issue.

Definitely worth a build.  Thanks Kevin

EDIT:  @JoaMat  Have you finished building this yet

EDIT 2:  I'd love to experiment with the CCS with batteries as well.

[joehpj]

I always wanted something that can beat T2 output power. 2SC4686 has been obsoleted and need 3/ch to do 500V/20ma. I check the datasheet, c4686 has 2.2pf output capacitance so it will be 6.6pf in total. Fairchild announced SiC BJT but was never available on the market. GeneSiC has some available on the market with some silly price. Among those, this has the lowest Coss. http://www.genesicsemi.com/images/products_sic/sjt/GA04JT17-247.pdf 16pf Coss about 2.5 times than c4686. 

But I don't know why this thing should be BJT(alike?) but those pin out are labeled as DGS. 

BTW. series caps are need for 600V and means we'll need 4x caps than 450V to get the same capacitance. Much more expensive to build. I don't know if this has that much difference on the sound. Normally no one will and should drive it to clipping.

Edited May 14, 2017 by joehpj

[JoaMat]

Nice Kerry and Kevin!

How about Grounded Grid with tube input, c4686 CCS....?? Lots of opportunities – hoping for a rainy summer.