[jensting]

All,
 
I just couldn't resist the idea of having a go at refurbishing (if not modifying) my Stax amp. I wanted to do "just" the electroytics and the valves, those being the only components which I could reasonably assume had aged. (The rest of the components look fine, and I struggle to believe that Stax would have picked parts which are somehow inferior in such a nicely made piece of kit.)
 
I got a pair of NOS RTT/Matsu****a 6CG7 tubes on e-bay, so that's probably just fine.
 
I wanted to replace the power supply caps as well, and here availability of replacements look, ehrm, interesting. For a supplier which looks like it will be able to ship and which sells in small numbers, I found a French sub-site of Mouser Electronics which at least do a mean website.
 
Capcitor specs: I just took the lid off the amp and read the numbers (didn't look at the solder side of the PCB). The main high voltage supply caps are 100 uF / 500V polarised (I assume electrolytics) with some sort of radial mount (can't see if it's leads or snap-in), these are marked C11 - C14. There's also a pair of low voltage 10 uF 50V marked C9 - C10 which I might as well replace when I'm in there.
 
(Finally there are a few  electrolytics in the delay circuit, but I'm leaving those alone - it would be a strange delay circuit indeed which made a difference to anything other than the switch-on delay, and I can't imagine that improving from where it is

).
 
The high voltage caps are 30 mm D / 60 mm L dimensions, marked 85 deg C.
 
Now, my main doubt: capacitor terminals. The ones I can find on mouser are all snap-in not leads. Is that what the PCB takes? If the PCB doesn't take those, can I simply and cruelly drill a hole or two holes in the correct diameter, and go ahead and snap in the new caps?
 
The rest of the specs I'm merely slightly in doubt: higher temp rating versus higher ripple current. Temps available are 85 and 105 deg C. For the caps that are in stock, the data sheet all say 3000 hours life time / stress test time.
 
-  85 deg C, 3000 hours : Cornell Dubilier 380 LX, ESR : 2 Ohm, max ripple: 1.1 A
- 105 deg C, 3000 hours: United Chemi-con KMS, max ripple: 0.82 A
- 105 deg C, 3000 hours: Nichicon GN, rated ripple: 0.65 A
 
No ESR spec on the two latter. (For comparison, Kendeil K06 are 85 deg C, 1 Ohm, 0.7 A but I can't find a place to buy those.)
 
Physical dimensions are all a small variety, 22x45 up through 30x30 (mm D x mm L). They're going to lok a little short, but at least if I get 30 mm D they're going to look like they belong...
 
So, any comments? I honestly don't know how hot it gets in there - the ventilation looks carefully thought out - so the extra max temp night not improve matters for me. The increased max current of the Cornell caps ought to go with lower ESR which ought to be A Good Thing [tm], right? Oh, and they all cost about the same - that's the downside to not shopping in Audiophile stores where there is a siginificant charge for flowery verbage about the sound qualities

 
Best Regards
 
Jens
 
PS: background : electronics engineer but working on integrated analogue circuits, not discrete or audio. Last DIY audio was back at Uni in the late 1980s but I was doing ambitious stuff back then. Funnily enough, one of my first DIY projects (back in evening classes while at secondary school) was the Wireless World electrostatic headphones for which I did put together the power supply but no more - I was first told to keep one hand in my lap while poking a live circuit and then found out why that was a good idea...

 

[KimLaroux]

If looking for a supplier it would help to tell in which country you are.
 
Also, I would look bellow the board before ordering new caps. I'd make sure they are 2 poles and not 4 poles, and measure the lead spacing. At the price those caps sell for, I would not buy on a guess. And since I have a feeling capacitor sizes shrunk since the SRM-T1S was designed, all you'll be worried about is lead spacing.
 
Ripple current ratings are related to ESR: A higher ripple current rating will mean a lower ESR value.
 
This would do nicely, if the lead spacing is correct:
http://www.digikey.com/product-detail/en/LGX2H101MELZ40/493-6750-ND/3664019

 

[jensting]

Hi Kim,
 
I'm in France, so in practice any supplier that ships from Europe would be fine.
 
Digikey had a series which Mouser didn't - good hint, thanks. So did Farnell.
 
I take the point about making sure that it's not more than two pins. I'm still optimistic that some of the Stax experts who have been there before can clue me in.
 
Best Regards
 
Jens

 

[jensting]

.. make that 100 uF 400V . Ooops. This makes a significant difference to the connectors - I can get radial leads for this voltage. That's nice

.
 
Off to gather courage to take the bottom off to look at the PCB (easier said than done, the lower cover and the feet bolt through to nuts sitting on top of the PCB => everything becomes a little loose when undoing the bolts). Also, must remember to disconnect power and leave for a while.

 

[elmoe]

Hi,
 
You might find what you want here: http://www.audiophonics.fr/capacitors-electolytic-capacitors-c-51_176.html
 
A few "audio grade" caps, with prices to match.

 

[jensting]

Well, I had the bottom cover off. Less intrusive than I feared, since the screws through the bottom cover only attach to spacers (they do not go through to the PCB).
 
It looks like the power supply caps are snap-in with 12.5mm spacing between the centres of the leads on the PCB side (i.e. probably hole spacing), the leads are flat, 1.8mm width, not exactly centered (huh?) and there appears to be components mounted on top of the PCB underneath the caps.
 
Given that all the snap-ins I've found have hole spacing of 10mm, I think the smartest move is to find some caps with plain radial leads, if possible with 12.5mm spacing, and make as neat a job of it as possible.
 
I'm keeping photos of everything, but I don't appear to be able to post any just yet. Ah well, it'll come.
 
Best Regards
 
    Jens

 

[jensting]

Looking like UPW2W101MRD from Nichcon, low impedance, 100 uF, 450V, radial leads @ 12.5mm spacing, 8000h @ 105 deg C reliability and 350 mA max ripple current. Available from Mouser, now I just have to see if I need to buy any more components to cut down the burden of shipping (or if I run out of patience...).

 

[jensting]

Well, finally got it done.
 
Spritzer kindly shared some of his experience and on his advice I went with bigger caps.
 
Again, it was Nichicon but in 270 uF 420 V spec, UPTW6271MRD. The rated ripple current is 1.2A which is, ahemm, surely "sufficient" - in the same way that horsepower in Aston Martin engines is "sufficient."
 
I took the lid and bottom sheet off and measured the DC (and AC) voltages on the existing caps. The four caps showed:
 
+313 (0.09), +318 (0.82), -329 (0.11), -335 (0.97)
 
with the first closest to the front marked C12. I guess one pair of supplies is more lightly loaded than the other.
 
The unsoldering was a little scary. I used a desolder pump to get the old solder off, but not all of it came off (of course). It turned out that I could straighten out the legs on the old caps anyway, tearing them off the solder joint (if you see what I mean). Happily, there was good PCB track underneath and the solder pads re-tinned nicely. The only annoying thing about the swap is that the pins on the new caps (radial leads) are placed differently relative to the circular outline compared to the old caps (snap-in), so I end up with two caps rubbing against each other. I didn't try to force the issue, and concentrated on getting the caps ligned up more or less neatly. Soldering the new caps in went like a breeze.
 
Measuring the voltages on the new caps, I got
 
+315 (0.03), +321 (0.3), -332 (0.03), -338 (0.37)
 
So I guess the bigger size (and maybe the lower ESR) of the new caps paid off - 10 dB less noise

 
With a nice little component tester I measured the capacitance and ESR : old caps 91 uF, ESR=0.8 Ohm, new caps 270 uF, ESR=0.1 Ohm. All the old caps had the same measured values, so I guess they weren't dead - at the same time, they defintely did not have as good specs.
 
There are also two smaller electrolytics. I observed that the solder side of the PCB was very busy around them (leads run off right next to them) so I decided to leave well alone and not change them.
 
Replacing the valves went easily enough, but adjusting the bias was a nightmare. I guess one has to get used to the whole idea. I used the handy guide in http://www.head-fi.org/a/adjusting-bias-on-stax-tube-amplifiers and all I can say is that not only do the results drift but the trim pots sure are sensitive!
 
(I see that my charisma rating appears to be so sky-high that I can post photos. Stand by for slide show 

)
 
Oh, and the headphones still sound amazing. Not only is A/B testing right out on this manipulation, but I suspect that the differences here are a little too subtle for me. I don't mind - I'm having a good time

.
 
Best Regards
 
    Jens

 

[jensting]

A few photos:


The Before picture

. At the back another ebay bargin (the pair of Martin Logans cheaper than the headphones + amp...).
 


The caps we're looking to replace.


Oh, and they're not kidding.
 
You really, REALLY, want to be very disciplined around the insides of this. Apart from trimming the DC there's no reason to have power anywhere near the box while the lid (or the bottom) is off. Always keep one hand away from the bench / box while there is power applied. As below:
 

Using a clip-on probe so as to leave one hand free to do absolutely nothing (OK, in this case I was taking a photo, but you see what I mean). And the probe is a so-called safety probe, i.e. together with a safety banana plug (oh yes) the voltage at the measuring point is not exposed to the person using the equipment[1].
 


After a bit of not very creative swearing (oh, and perhaps some de-soldering, I only remember the important details) we have all the caps off. Check out the tear marks in the tin on the pads on the left (the other three pairs have been re-tinned and are pretty). Some scuff marks on the PCB surface coating - not as neat as I had hoped for. Oh well, I'm not planning to sell this one.
 


Inside view of the same area. We note that the PCB is lightly soiled in the positions of the caps which gave high ripple on measurement. Hmmm. I don't think that was my desoldering, not because that's implausible but because I was roughly equally cack-handed on all four and the pattern here is not the order in which I took them off.
 


After soldering. Enormous amounts of solder were needed to fill the large pads.
 
Oh and detail view of the small mod to the PCB. If you look at the component side photo above, there's a chassis connection there which is cut and replaced by only this capacitive connection. (This is the star point of the supplies on the PCB.) Supply star connection is great, but mainly for DC. If there's any other electrical connection between the chassis and a "ground" point (say, signal cold from the input terminals), AC and DC can circulate quite happily and make trouble. A small ceramic cap simply makes RF noise go away and breaks any DC loops.
 

New caps in place. The spacing is not brilliant, since the leads come out in a different place relative to the centre compared to the old caps. Oh well.
 


NOS japanese valves. Nice. The ones I took out where from the USA.
 


All done, only the trimming to complete.
 
And that was that.
 
Best Regards
 
   Jens
 
[1] I learnt at an early age that it's a good idea to always be very careful around a bench with open equipment that's powered. I was told this by the electronics classes I took in the evening in primary school and the point was driven home when I was fiddling with the supply to an amp for electrostatic headphones - the Wireless World design. I was done with measuring and proceeded to take the (non-safety) crocodile clip off the 400V AC. Ouch! Happily just a little burn which didn't even scar permanently. If my other hand had not been in my lap, I would have been looking at possibly a career ending event. As it was, the adage of keeping one hand off the bench stuck very nicely in my mind. Can you tell?

 

[davidsh]

Great writeup, it's appreciated!