The DIY Electrostatic amp thread

[chinsettawong]

I like your idea "Staxes for the masses". Looking forward to seeing them.

Wachara C.

[spritzer]

I certainly have a lot of fun building the "lesser" amps just because they cost next to nothing. It's also a lot of fun to put discarded parts back to use.

[spritzer]

Since I just did one of these today I figured I'd post a "how to convert a SRM-T1 to use ECC99's". Not much to this really since the ECC99 is very similar to the 6CG7, the heaters draw a bit more current but the transformer seems to have no trouble with this. The only difference is the setup of the heaters since the 6CG7 is just a 6.3V unit but the ECC99 is a 6.3/12.6V. Easy enough to accomplish, cut two traces, add two wires and bridge two pins on the sockets.

On the top side all the anode resistors have to be changed for 30K/2W units. Rebias and you are done.

[Currawong]

I've finally gotten around to pulling out the KGSS boards and am plugging the values into a Mouser project, following the guide in the first post, starting with the resistors. However, while I was tweeting away about it, I was offered a Blue Hawaii built by pabbi1, I think which, according to a friend of the owner, just needs the LEDs changed and a case to get it working again. I imagine the LEDs are similar in purpose in the BH as in the KGSS?

[nattonrice]

If you mean red leds then yes they are just 1.7v 5mm jobbies.

Yep same idea in usage.

[luvdunhill]

If you mean red leds then yes they are just 1.7v 5mm jobbies.

Yep same idea in usage.

Agreed. Look for 1.7v though, the 1.8v will result in a different operating point unless you adjust a few resistors. You'll probably have to buy a few different kinds and measure them. You may even have better luck with 3mm devices.

[Kerry]

I use these LTL-307ELC in my Blue Hawaii. They are about 1.7V. They have them at Mouser.

[nattonrice]

In a push I've used 3mm ones before to some effect and got away with it (it sorta worked out in the wash tolerances and all).

The ones kerry list work great.

I bought about 100 of them and regret not getting double.

Grab a small 12v supply (I use a tiny cheap meanwell smps) a 1k resistor and a little pair of sip sockets.

Together with a dmm it is the fastest check and match job you'll ever do.

[luvdunhill]

Together with a dmm it is the fastest check and match job you'll ever do.

Second only to CRDs, but maybe i am the only one that matches those.. I recently matched a ton of 2sc3381 and those aren't really that hard. You need two meters, one that can do .1uA accuracy and a resistor and a pot... The key is making a jig that just requires you to pull the device out and rotate it so you don't have to duplicate the jig for the second side...

[nattonrice]

I'd considered how to match those dual package devices and gave up due to the sips... well not sip'ing nicely when there's 7 of them.

Obviously using one set and rotating didn't occur to me doh!

[luvdunhill]

I'd considered how to match those dual package devices and gave up due to the sips... well not sip'ing nicely when there's 7 of them.

Obviously using one set and rotating didn't occur to me doh!

You can even get zag sockets, but if you use standard .1" veroboard you'll have to do some tricks with the pins. But yeah, rotating is the trick. I'll take a pic for you tonite, and if anyone is interested I could colate the matching data and post that.

[nattonrice]

Yes please I'd love a pic.

The less amount of fiddling that needs to be done the better I say.

[luvdunhill]

double

Edited June 12, 2011 by luvdunhill

[luvdunhill]

I matched them at 6V and Ib = 3.5uA and used a 1M pot in series with a 1M resistor to set the current. Once the base current was set, I just read off the collector current and matched these values. Typically what I do is set a target collector current using the pot, then read off the base current and use this to calculate hfe, but this approach doesn't work well with high hfe devices. In this case, it's better to just fix the base current...

[Currawong]

Ok, with Nattonrice's help, I've made some progress, if only a year late .

With Justin's KGSS boards from Headamp.com, I'm stuck on the transistors. I was lucky to pick up a BH in pieces with a heap of spares which has helped, but there are still gaps, and do they all need to be matched?

I'm posting where I'm at here as I haven't investigated outside of Mouser yet and I thought it might end up helping someone else in the future if I post stuff here.

Small transistors:

2SK389 x4: As per the first post, I'll get 2SK170s matched from AMB.

2SA1970 x4, 2SC1815 x2, 2SC2240 x4: Enough came with the BH, but it'd be handy to know where to get them or alternatives in the future in case other people ask.

2SC2705 x4: EOL but in stock in Mouser.

Big transistors:

2SC3675 x12: Subtituted with 2SC4686A in the KGSSHV (thanks Yun).

2SA1156 x4: Can use the higher-voltage 2SA1486, but have to get them from bdent. Thanks Stax Mafia.

2SA1968 x2: I have plenty spare from the BH, but I gather the alternative is 2SA1486.

800V Mosfet: The BH PSU has something with "4N60" on it. Alternative is FQP8N80C (thanks Tom).

Other stuff:

RLED: LED lamp. Was a weird shape so thought I should check.

1N914: I picked the first on the list.

Big PSU caps: I bought some expensive Nichicons because I figured, why not?

Edited July 12, 2011 by Currawong

[livewire]

The RLED is a red light emitting diode.

In KG's designs it is used as a voltage reference as well as an operational indicator lamp.

Be sure to buy the correct operating voltage spec for this part.

I dont know what Justin used in his KGSS design. KG's original KGSS used 1.7 volt LED's.

Lately he is specifying 2.1 volt LED's in his KGSSHV and T2 designs.

The 1N914A diodes arent nothing special,

in the KGSSHV, KG currently specs the Mouser p/n: 512-1N914A part.

In case you havent seen it yet, basic KGSS info is available here:

The LED's function in this circuit is explained in detail.

http://gilmore2.chem.northwestern.edu/projects/showfile.php?file=gilmore2_prj.htm

[spritzer]

You can get most of the transistors from bdent.com. The 3675 is end of life but you should have no issues tracking some down. Plenty of alternatives too and if you are in a bind, PM me as I have a few hundred. Ditto on the 1486 and 1968.

The 1156 should be replaced with the 1486 if you are using that one in the PSU.

The RLED is actually the chassis led indicator and has nothing to do with the circuit. Just pick a LED and feed the forward voltage and desired current through an online calculator. It's fed off the +15V supply.

[Nevod]

I've read that SRX design with EL34 as output (same as Singlepower ES-1) is good.

But how about some more power? Would KT88s be adequate, or it would be better to use 2 EL34s in parallel? Or anyway that would be an overkill and one EL34 is enough, considering load capacitance should be comparable with SR-Omega?

Also, what if the anode resistors would be replaced by some availible kit CCSes?

Primary reason I'm considering instead of KGSSHV all that is because I don't want to mess with finding rare parts, hard to source stuff here, while tubes are reaadily available. Plus the price reason, SRX seems to be gradually upgradeable.

Also, I don't really well understand the first stage of SRX. By looking at it, I'd call it a cross-coupled cascode. What is it's role, to compensate out tube nonlinearities?

[kevin gilmore]

The purpose of the cross coupled thing is to increase gain. This is not an easy thing to do.

You need at least 70 db of open loop gain, and that has to be at 20khz. Mikhail's versions

without modifications suffered greatly from this which is why there were severe channel

to channel imbalances.

[Nevod]

So then anode current sources seem to be even more fit? Some I've seen are rated for up to 1kV, enough for even somewhat boosted SRX..

[kevin gilmore]

10m90s on a heatsink is WAY better than any resistor. Good for 900 volts.

Other versions good for more than 1kv. If a tube ever shorts, nothing bad happens.

If you use the resistor plate load and a tube shorts, fireworks.

Anything more than about 20 ma will push the output tubes too hard.

blubliss has a gilmore schematic fully modified singlepower with the ixys parts

you can ask him how much better it sounds than the original.

[Nevod]

And one more somewhat silly question.

What about plate current source wired as mu-follower/SRPP? (they are pretty much the same, aren't they?). Of course, biased so that other side doesn't get close to zero current through CCS.

Or better leave tube current variable?

[kevin gilmore]

If you do SRPP you end up in the same troubles as the ray samuels A10.

Same sort of thing as mikhail's original disaster.

Voltage swing on the top tube is rail to rail. Which means that unless you

want 4 seperate filament supplies for each of the top tubes you quickly

exceed the cathode to filament voltage specifications.

Which in additon means seperate tubes for the outputs. 8 in all.

Gets big and complicated in a hurry.

[luvdunhill]

If I understand what Nevod is asking correctly, I think the answer is yes, it will work and likely sound better than how the BHSE for example is wired using a single depletion mode MOSFET. I'm not sure what adjustments will need to be made though, other than jus setting the dc offset properly, assuming there is sufficient range in the circuit already.

Edited September 20, 2011 by luvdunhill

[spritzer]

Just checked the first page and I really suck at posting stuff I meant to post... Anyway, I'm always fixing up some Stax amps so here is a SRA-3S I just gave some much needed TLC. It was just the usual stuff, replace all diodes, film and electrolytic caps with equal or uprated parts.

Nice curve on the PCB and previous service attempts complicate matters but it all went well in the end. It's a bit startling just how much stuff was crammed into this tiny box as it it only a bit bigger then a SRD-7. I also added Pro bias to one of the sockets and it sounds pretty nice driving a SR-007Mk2....