The ultimate DIY? A Stax SRM-T2!

[s_r]

Yea, I wouldn't dare move the probe while the amp is still on. The problem is I have no idea where/how to start looking. What parts would be a likely culprit?

[mypasswordis]

What exactly have you tested so far in terms of trying to swap channel noise? Or are both channels exhibiting the same amount of noise now?

[s_r]

It's been the about the same amount of noise on both to start with. Swapping tubes, power cables, or powering up one channel at a time didn't change whether a particular channel had noise or not. Earlier one channel alone would be much quieter than the other powered up alone. Haven't tried that again after doing some part replacements though.

[mypasswordis]

Wait, one channel alone is much quieter in terms of total volume or noise volume? I believe I read that you replaced all the resistors in the batteries with Xicons (and everything can be adjusted correctly) and added gate stopper resistors to the FETs... have you done the other mods? Power supply mods/fixes and added 5pF 1kV caps across 100k feedback resistors? It's better if you say everything you've tried already in one easy to read post so people aren't just shooting in the dark.

[luvdunhill]

OK - decent scope, but be careful what and where you probe.  My tendency would be not to probe with the T2 turned on - one slip with the ground lead, probe tip or your finger could spoil your day big time.  I would turn the T2 off, attache probe connections and turn on again.  But bear in mind that any complex feedback amplifier can oscillate - IIRC there are the odd single figure pF capacitors in there to stabilise the amp.  Also probe impedance can modify the effect - either stopping the oscillation, making it worse, or changing the frequency.

It would be easier if he could tie the clip to the "bottom" negative rail though - which he can't - unless he floats the scope or uses a battery supply.

[s_r]

19 minutes ago, mypasswordis said:

Wait, one channel alone is much quieter in terms of total volume or noise volume? I believe I read that you replaced all the resistors in the batteries with Xicons (and everything can be adjusted correctly) and added gate stopper resistors to the FETs... have you done the other mods? Power supply mods/fixes and added 5pF 1kV caps across 100k feedback resistors? It's better if you say everything you've tried already in one easy to read post so people aren't just shooting in the dark.

As in total volume. The relative amount of noise is probably unchanged, but it's too quiet to notice. The board I have has spots for the 5pf caps in parallel with the 100k resistors. As far as I can remember this is everything I've tried:

-swapping tubes, power cables/plugs on the PSU to see if the noise would swap channels

-trying different sets of input & output tubes

-powering up one channel at a time

-replacing the takman resistors in the battery with xicons

-replacing the 10k pots in the battery with PTF56 resistors

-adding a 1k resistor to the gate of the K216s

-lowered the resistors in the PSU for the -500V supply

-touch an insulated screwdriver to the trace between the ptf56 & k246 per 8street's suggestion

[Craig Sawyers]

4 hours ago, luvdunhill said:

It would be easier if he could tie the clip to the "bottom" negative rail though - which he can't - unless he floats the scope or uses a battery supply.

That is where your active diff probe comes in handy.  But interestingly the really old generation Tek probes, I guess because of their physical size, can handle much higher voltages.  http://www.reprise.com/host/tektronix/reference/voltage_probes.asp

Although CMRR would not be as good as the current product, using two P6000's say, inverting one channel on the scope and adding might skin that cat.

There are two NOS P6007 on eBay at the moment for about $70 each BIN/ONO, x100 and handles 1.5kV AC and 2.4kV DC.

[Kerry]

So here's a progress update on my T2 mini build.

I started off with the a golden reference version of the T2 supply.  Had a couple of issues.  There was some 60Hz noise I was hearing (I'm guessing from the integration of the negative rail supplies), which I can probably solve for.  Also, the current limiting, which works, wasn't fast enough when there was a shorting issue with the amp on the negative rail (FETs in the output section shorted to the heat sinks).  It would pretty much blow out several components in the supply.

I decided to go back to the original design, but as a shrunken version (206mm x 200mm).  I did some small mods (added cascaded current source for the 30V supply to the opamp and added some additional spots for compensation caps.

I finished this over the weekend and have spent some time listening.  I've still got to tweak the PS (some small oscillations - probably because I was experimenting with the .1uF cap to the negative input of the op amp and added a .1uF cap across the voltage divider).  My BK multi-meter is showing between 1mV and 0 on the AC setting.  I've had some of the supplies on the scope but will need to go through all of them.  This causes some very faint background noise.

The amp has had it's fair share of issues, primarily with the output FETs shorting to the heat sink.  I've had this once on the left channel and three times on the right.  I believe it was that some thermal compound found it's way inside the collar of the thermal washers (7721-3PPS).  I've cleaned these up and have had the amp on for 12 hours now with no issues.

Also, the right channel DC test points are all pretty much spot on, but the left channel has an issue with the -543V test point.  It is sitting at -502V (the rest of the DC test points are pretty close except for the top of the batteries that sit at +238V).  I've been looking at this issue for a while, but haven't figured it out yet. I've got to focus here a bit.  Any thoughts would be appreciated.

I also have to finesse the battery adjustments to get rid of some occasional noise coming through the right channel as well.

I have had my 404's and then later O2's on this beast.  Wow, how nice    It's been too long...

[JoaMat]

42 minutes ago, Kerry said:

I decided to go back to the original design, but as a shrunken version (206mm x 200mm).  I did some small mods (added cascaded current source for the 30V supply to the opamp and added some additional spots for compensation caps.

Great!

Personally I think the original T2 PSU is a very nice design. A modern SMD version is nothing but awesome.

[joehpj]

what's the el34 you are using?

[Kerry]

They're just the new Mullard tubes.  Nothing fancy yet.

[purk]

That gotta be the coolest DIY T2 ever.

[chinsettawong]

Nicely done!  Really like how you use all those SMD parts.

[iwik]

Nice work Kerry, what else can one say.

[Craig Sawyers]

I take my hat off - that is a thing of beauty.

[johnwmclean]

That is very impressive work Kerry.

Are they Conrad heatsinks? Have you got dimensions, just curious as to the total finished size.

[bozebuttons]

Very Nice Kerry,

So when are you going to lend it to me when you are done

[palchiu]

Beautiful!

[Kerry]

Thanks everyone.  I really appreciate the comments.

Yes, these are Conrad heat sinks.  The PS uses 200mm x 75mm and the amp is using 350mm x 75mm.  I'm seeing about 110 deg. F on the heat sinks at the hottest point of the amp. 

The amp and PS will be identical in size at 12" (W) x 14.27" (L) x 3.33" (H - not including feet).

[Craig Sawyers]

1 minute ago, Kerry said:

I'm seeing about 110 deg. F on the heat sinks at the hottest point of the amp. 

 

IIRC my original T2 clone build gets to about that.

[Pars]

Yeah, that is one sweet build Kerry! Congrats! Hope you get the noise issues solved, but I have no doubt that you will in short order.

[gepardcv]

Incredible work. Any chance you might release Gerbers for these T2 boards?

[iwik]

im a bit curios re the Aavid    7721-3PPS shoulder washers.  About to finish my build and noted Craigs reference to these being the best to use with

Stainless screws and getting the required Torque of 1.1nm. I see from the Datasheet that it says recommended force be .5 to .7. So no one has had

any problems by cranking them up to 1.1 for T0220.

Thanks

les

[mwl168]

Maybe Kerry can help verify.

I have both Aavid 7721-3PPSG and 7721-10PPSG. The 7721-3PPSG has longer shaft than the 10PPSG (3.18mm vs. 2.41mm). But in my case, I was not able to fit the 3PPSG into the tab hole of the TO220 transistors. I ended up using the 10PPSG for all my builds.

There are others that have experienced similar issue and some had to enlarge the tab hole to make the 3PPSG fit.  

Edited June 3, 2020 by mwl168

[Kerry]

Noise issue solved on the right channel 

I'd like to think I'm brilliant, but in this case purely dumb luck.  I changed out one of the K216s coming out of the top of one of the batteries.  I needed to do this because of the dumb part of my luck.  I do love this power supply

Just listening now in all of this the baby's full grandeur

Still need to address the DC offset issue in the left channel and there seems to be some HF oscillation in both channels around 739KHz.   

@iwik I haven't gone through any final torqueing yet.  My method is to drill out the transistors using a #26 drill bit and then deburr.  This takes away any stress on the washers.  I don't see why there would be an issue with bringing these to spec though.  I bough a torqueing driver a while back.  There are still a number of mechanical steps left on the amp.

EDIT:  Yes, I'm using the 3PPSG washers.

Edited April 21, 2016 by Kerry