The DIY Electrostatic amp thread

[JimL]

On April 20, 2016 at 6:33 AM, Craig Sawyers said:

The 1967 Quad 33 used the same circuit, but moved across to silicon BC109 - which were equally expensive back then.  The 12 transistors in the 33 would have been the determining factor in the unit's cost.

You can make that 1965/7 performance much better by adding a transistor (to up the open loop gain, and give enough drive current for the RIAA), something they probably knew - but that would have added a lot to the semiconductor cost.

Which is all horribly irrelevant, since it is not going to be used!

On this side of the Atlantic the Dynaco PAT-4 used a similar topology  IIRC.

[oogabooga]

I want to replace the internal hookup wire in the SRA-3S - some of the wiring insulation is looking black/heated. How much do I have to worry about the voltage rating of the wire insulation? Is the 300V stuff I have on hand good enough or does it need to be 600V-rated?

FWIW I'm going to make one of the jacks pro bias @ 560V.

 

[spritzer]

600V is a must for the HV stuff, bias and output wiring.  I would also add 5K1 resistors to the output per the modern way to do things

Bias is easy, you could just take a 4M7 resistor off the B+ but I'd put something in there that limits the possible current.  A high impedance voltage divider that shaves off just a few volts in normal operation but comes in handy if there is a short. 

In other news, I just bought a mini SRA-3S so there will be a reforestation project soon.  I might just go crazy with this one though... 

[oogabooga]

I was afraid you'd say that. A 100 ft spool of 22 AWG teflon-insulated wire costs ~$75 at Mouser. So much for doing this repair on the cheap!

For the 5k1 - is a 2W or 3W resistor about right? Buy 25 and match 2 pairs by hand?

What's a mini SRA-3S?? How the heck can they make that thing any smaller!!

Edited April 25, 2016 by oogabooga

[GeorgeP]

Sameer, how much do you need? I can send you some in a few different colors.

[Kerry]

I use bulkwire.com or McMaster.com

I remember Craig had some sources in the UK. I think they were in the T2 thread.

[spritzer]

1 hour ago, oogabooga said:

I was afraid you'd say that. A 100 ft spool of 22 AWG teflon-insulated wire costs ~$75 at Mouser. So much for doing this repair on the cheap!

For the 5k1 - is a 2W or 3W resistor about right? Buy 25 and match 2 pairs by hand?

What's a mini SRA-3S?? How the heck can they make that thing any smaller!!

Just ebay 600V teflon wire and you should find plenty. 

1/2W 500V 5K1 resistors will work perfectly. 

ooops that was supposed to say "mint"...   

[DefQon]

Yep plenty of mill-spec 24awg 600v teflon insulated spools on the US ebay domain. I bought my 2000ft spool off the US ebay was only like $70AUD after shipping.

[oogabooga]

Thanks all for the suggestions, I've got a few choices now. Been avoiding eBay mostly because of delays with customs in Canada, but it looks like I can either have it cheap or fast...

George - the offer is much appreciated but I'm gonna just buy a spool and get it over with

[oogabooga]

Removed all the PSU caps, to find two resistors in the PSU (circled in red) looking terrible. One has it's casing blown clear off! The Vishay BC PR01 metal film resistors (1W) I have are much smaller than what Stax used. Is 3 W enough/safe here? I figure the voltage is 280V here, so that 47k resistor should top out at 1.7 W if it's other end was ground. 

Bias resistors are also cracking, going to replace these with 1/2W Vishay BC high voltage, ditto for the pro bias. 

I hope that's it for parts - one more order from Digikey and I can start putting it all back together! Thanks everyone for the help

 

(Does anyone know that the two 'double arrows' just above and to the right of that red circle are? Never seen that symbol before and I can't find it on the board, or on Google.) 

[spritzer]

Not sure what those two arrows mean but I'd guess the text in Japanese has something to do with it. 

Those resistors will see some load especially if their values have drifted over the years.  3W units will be overkill but why not. 

[Pars]

Usually arrows like that would mean an offboard connector, or maybe jumper stake pins?

[spritzer]

I don't remember anything like that but it's been some years since I last had a SRA-3S here... 

Been working on the SRA-7S complete rebuild.  It was in horrible shape when it came here so it was full restoration time.  No original components left... 

[oogabooga]

On April 30, 2016 at 2:20 PM, Pars said:

Usually arrows like that would mean an offboard connector, or maybe jumper stake pins?

You're right! Traced the paths - the two arrows connect to pins on the Mag PU card slot. There's a jumper bridging the two pins on the card, so when it's plugged in, the 63k resistor is parallel with the 47k. 

 

[oogabooga]

Refurbished my SRA-3S PSU board today (except the HV caps that I'll place tomorrow after putting it back together). Three copper traces were already in a bad way when I started, and I broke a few more. Ran the blue 22AWG teflon-coated wire to fix. 

 

I've attached a picture so you can make fun of my soldering skills. I left a few mm on the component leads in case I need to reseat them... there'll be enough room in the case anyways thanks to those Rikens. 

 

For reference, I ended up replacing: 

diodes with 1N4007s,

the five PSU resistors with Vishay BC PR02/PR03 2W/3W

the three bias resistors with Vishay BC VR37 1/2W

three LV PSU electrolytics with Nichicon VX 22-47 uF (the HV caps will be 180 uF 400V)

four output film caps with Panasonic ECW-F(L) 0.22 uF 630V

 

[nopants]

clean that flux

[spritzer]

Stax never bothered with that...

[oogabooga]

PSU done! With no source connected, the "A" line (aka B+) is running a bit high at 603V, the "B" line measured ~56V, and normal bias is ~260V, measured before the 5M1 resistor. 

In the original design, the 10K input resistors are soldered across the contacts of the input selector switch. I'm not using the selector, so I put new resistors on perfboard and stuck that into the plugin slot for the Mag PU card. Connected a source and played a 1 kHz sine wave: maxed out at ~380Vrms output to the left and right channels. Played some music and the distortion I had before is gone! Pot is no longer scratchy, either (contact cleaner). 

Next step is to redo the input/pot audio wiring. Stax used shielded wire but the ends look like crap and the wires are too short to trim, so it's all got to go. Will single conductor pick up anything from inside the amp? 

Also, why the heck did Stax design this amp to lay on its 'side'? It just seems like sub-optimal cooling for the tubes...

 

 

[spritzer]

You are probably running higher mains than the original design called for which explains the higher voltages.  This happens all the time here as the gear was designed for 220V but we are edging closer to 240V. 

When using contact cleaner on the pot make sure it has some lubrication.  Some are just cleaners and they will kill the pot quickly by washing away all of the lubrication. 

I would use shielded wire due to the small nature of the box.  You could just use a super thin microphone cable with two contacts and a shield. 

Edit:  Side mounding the tubes is actually better in many ways.  The chassis acts like a cooling tunnel for the tubes. 

Edited May 6, 2016 by spritzer

[oogabooga]

 

Created HV bias from 603V B+ into a 23K-390K-390K voltage divider to get 585V measured just before the 5M1 ballast resistor. The plug-in card refurb will wait until I've enjoyed the amp for a bit :-)

Thanks again for your suggestions Spritzer. I used lubricating contact cleaner on the pot, and ended up finding enough shielded wire from the previous stuff to run the audio. Oh - as to that 'side' mounting, after poking around a bit, I'm convinced that the giant 50K resistors are a major reason Stax designed the amp on it's 'side'. They get DAMN hot and would probably melt if they were under the PCB.  

I measured the frequency response using a sine wave generator on my Mac and the AC Vrms reading of my Fluke via the headphone sockets. I set the volume pot so that a 1 kHz wave was 250 Vrms and measured at 19 frequencies on both channels. Attached log plots shows was normalized to 0 dBV @ 1 kHz. The bass rolloff is slightly more than I calculated for 0.22 uF caps into a 145KOhm impedance load, but -2 dB @ 20 Hz is more than great for me. The high frequency rolloff is about 4x stronger than on my SRM-323 and is probably why this amp sounds less fatiguing than the SRM-323 with my Lambda Pros :-)  Almost makes one think Stax designed the Lambda Pro to go with a tube amp... 

PS - I recreated the circuit in falstad.com: http://tinyurl.com/zlxjj9d  The simulator seems to work (better) if I replace the the triodes with N-mosfets, but the output signal still looks screwy. That being said it's still a great website for smaller circuits. 

 

 

 

 

[DefQon]

I changed the 300k resistor to 390k which gave me around 230ish v for the normal bias before the 4.7m resistor to output and I got a tapped 583v just after the diode.