The ultimate DIY? A Stax SRM-T2!

[audiostar]

10 hours ago, demonkuro said:

Thank you for your concern, before connecting the PSU to AMP, I let it work independently for several hours, all the output voltages are right, the drift is very small, and it looks very stable. So I confidently linked it to AMP, but nothing good happened... Just like the carbon I made 8 years ago, the PSU also exploded, and that time I replaced the damaged part and it worked fine. Thanks to your advice, I will go find an insulation tester to test each power supply. I should be more cautious.The PSU was tested today and luckily all worked fine except for -500V and +500V.Waiting for new parts and replace the wire with a higher withstand voltage rating.

Before you go deep into the rabbit hole with insulation testing, first you check extremely careful all connections through the umbilicals to the amp boards by continuity. Then you load test the PSU with some resistors, do insulation testing and all the other fancy things. 

[jamesmking]

1 hour ago, audiostar said:

Before you go deep into the rabbit hole with insulation testing, first you check extremely careful all connections through the umbilicals to the amp boards by continuity. Then you load test the PSU with some resistors, do insulation testing and all the other fancy things. 

yep I highly agree..  do the simple things first.... miss-wired psu to amp lead or mis-wired chassis sockets is far more likely than leaking insulation.

 

[demonkuro]

On 5/22/2023 at 10:59 PM, jamesmking said:

yep I highly agree..  do the simple things first.... miss-wired psu to amp lead or mis-wired chassis sockets is far more likely than leaking insulation.

 

Finally found the cause of the explosion.The cutouts in the radiator are slightly crooked, and despite the fact that I used insulation pads and insulation gaskets, they unbelievably caused a short circuit. Silicone pads have no problem with their insulating properties, but they are too thin and can create physical contact in some extreme cases. I have replaced them with ceramic insulation pads and what are indeed much safer. This is an important lesson for me. However, it is good to find the cause of the problem.

 

[jamesmking]

1 hour ago, demonkuro said:

Finally found the cause of the explosion.The cutouts in the radiator are slightly crooked, and despite the fact that I used insulation pads and insulation gaskets, they unbelievably caused a short circuit. Silicone pads have no problem with their insulating properties, but they are too thin and can create physical contact in some extreme cases. I have replaced them with ceramic insulation pads and what are indeed much safer. This is an important lesson for me. However, it is good to find the cause of the problem.

im glad you are making progress. checking for shorts from metal tabs on transistors to L bracket/heatsink should be part of a standard test procedure. I personally like aluminium oxide insulation pads... not as cheap as silicon but much more resistance to puncture.. .https://www.mouser.co.uk/ProductDetail/532-4171G

 

 

[audiostar]

Exactly because of their physical properties and being so thin, I wouldn‘t even think in using them in HV environments with naked tabs and inox bolts. Heck, I am not using them at all, especially if space permits it. 
 

Frst comments were already regarding those silicon insulators! Just no way one can tighten up the sands to specs at around 8-9 Nm with those without some damage .

Edited May 23, 2023 by audiostar

[demonkuro]

1 hour ago, audiostar said:

Exactly because of their physical properties and being so thin, I wouldn‘t even think in using them in HV environments with naked tabs and inox bolts. Heck, I am not using them at all, especially if space permits it. 
 

Frst comments were already regarding those silicon insulators! Just no way one can tighten up the sands to specs at around 8-9 Nm with those without some damage .

You are right. If I had noticed this in the first place, it wouldn't have exploded. I spent the whole day checking which parts were damaged. Some parts were damaged but could not be seen from the outside, and it took a lot of time to check all the parts one by one.

[jokerman777]

Hi, back with a load of dummy questions in the midst of building it 🙂 (Again I got the boards and chassis from GeorgeP which I believe are same as Kevin's original design).

PSU Front Panel Power Indicator: There is a circular cutout in the center of the PSU front panel, I measured and diameter is roughly 19mm, any recommendation on what part to use here? I suppose it will be a switched light, although the IEC inlet is already switched?  I think I found some push button with LED on parts express that should fit.

Grounding: If I'm understanding the layout correctly, I believe mains earth/XLR input ground/Amp chassis are connected together as chassis ground, and can be connected to the circuit ground via the on board switch. It seems like the PSU chassis is not grounded?

Also there are two GND output pads (electrically connected on board) on the PSU board and multiple GND pads on the AMP board, is it recommended to or not to pass both GND output points from PSU to the amp? don't how much it matters but passing both sounds like ground loop to me.

Transistors Mounting: With AAVID 4171G thermaloy can I get away with not applying thermal grease? Anyone has the success of using nylon screws for mounting the transistors? I got a bunch of them and thought they will give me a peace of mind when it comes to insulation but second thought they are not as mechanically stable as metal ones.

Components placement: On the amp board I figured these are probably better to be soldered on the back of the board (facing top): LEDs, power resistors, trimmers and test points, are there any other components that might run warm which are better to be soldered on the back?

Thanks in advance for your opinions!

[Craig Sawyers]

Regarding transistor mounting. I'm thinking back a looong time to when I built mine. I used steel screws, because I could not get enough torque on plastic screws (I use a torque screwdriver and read the data here https://www.onsemi.com/pub/Collateral/AN1040-D.PDF ).

But the trick is to use the AAVID long penetration washers.

You need enough length to go through the TO220 tab, and penetrate most of the way through the 4171G insulator.

Now the problem is that the thickness of the tab is extremely widely specified - somewhere between 0.5mm and 1.39mm. I've done a straw poll of TO220 devices I have, and the vast majority are between 1.25mm and 1.3mm thick. My 4171G insulators are actually 2mm thick (at the top end of the tolerance; they are nominally 1.778mm (0.07") +/- 0.254mm (0.1"). So mine are towards the top end of the tolerance band.

Anyhow, it looks like the AAVID 7721-13NG (Mouser has 8,000 of them) is the man for the job. I can't find the remainder of the ones I bought to check, but I think I used those. The extension is between 2.71mm and 3.26mm. You need to check your tab and insulator dimensions, but under most circumstances the shoulder washer will go through the TO220 tab and most of the way through the 4171G.

If the tolerance stack goes against you, that shoulder washer might actually protrude from the bottom of the 4171G. If that is the case you absolutely must trim the end shorter. Otherwise you won't make thermal contact between the transistor and the heatsink.

That will enable you to use steel screws and nuts (don't forget the lockwasher!).

Oh - and you absolutely must use heatsink compound!

Craig

Edited May 30, 2023 by Craig Sawyers

[jokerman777]

Thanks for the tips Craig! Metal screws + shoulder washers + thermal compound it is then, no more fooling around. Will for sure check if the length of the shoulder washer from the other side for each installation. 🙂

[Craig Sawyers]

You have that right - the T2 has some fearsome voltages knocking around - and detail for everything is the name of the game.

To make the umbilicals connecting the power supply to the amp chassis took 8 hours solid work. That is typical of the length of time needed to not only make it work, but continue working as the months and years go by.

[Craig Sawyers]

Found the spec sheet in my T2 parts folder - looks like I used 7721-3PPSG . That has an extension of 3.81mm (1/8") and that clearly worked. Again Mouser - they have 11,500 of them in stock.

[jokerman777]

Thanks! I got the ones with 0.79mm extension according to a bom posted in an earlier thread here, I guess that's too short to cover TO-220 pads with varies thickness. I will get another batch with longer extension.

Actually why is it bad for the extension to protrude the isolation pad? Assuming it's not so long that it will also protrude the metal bracket and board underneath. With that in mind seems like the 7721-10PPSG from the datasheet is a good fit with 2.41mm extension going thru entire TO-220 pads and about half way into the isolation pad.

[Craig Sawyers]

12 minutes ago, jokerman777 said:

Actually why is it bad for the extension to protrude the isolation pad? Assuming it's not so long that it will also protrude the metal bracket and board underneath. With that in mind seems like the 7721-10PPSG from the datasheet is a good fit with 2.41mm extension going thru entire TO-220 pads and about half way into the isolation pad.

because the OD of the washer extension is larger in diameter than the hole in the bracket and PCB, assuming they are nominal clearance holes for the fixing screws.

[Craig Sawyers]

Here is an interesting figure

 

The bottom axis is in kPa-cm. Atmospheric pressure is 100kPa. Let's say that we use a shoulder washer that goes into the ceramic insulator 0.5mm, so the overall arcing distance is 1mm, or 0.1cm. So we are interested in 100 x 0.1 = 10 on the horizontal axis. That says that the breakdown voltage is about 5kV - so we should be OK?

Well no. Those curves (Townshend curves) are taken between two parallel plates, so an ideal measurement arrangement - the result of a very careful experiment.

The case of a small diameter screw thread, which has voltage stresses from each thread peak reduces the breakdown voltage significantly. And it is uncontrolled - tiny bits of conducting swarf and other grot possible too. So an 0.5mm insertion into the ceramic is getting marginal for 500V.

Increasing the insertion to 1.5mm gives an arc gap of 3mm (0.3cm), giving a (ideal, parallel plate) breakdown voltage of about 10kV. Even if this was degraded by a somewhat outrageous factor of 10 (1kV), we are still safe at 500V.

Which is the basis for using shoulder washers that are as long as reasonably possible

Edited May 30, 2023 by Craig Sawyers

[audiostar]

2 hours ago, jokerman777 said:

Thanks for the tips Craig! Metal screws + shoulder washers + thermal compound it is then, no more fooling around. Will for sure check if the length of the shoulder washer from the other side for each installation. 🙂

Make sure the thermal compound paste you are using is non-conductive as well as not all of them are. 

[jokerman777]

hmm I've gotten ceramic straight pin tube sockets from thetubestore and turns out the 8pin octal ones don't fit the board.

Are these from diy hifi supply the right ones to get? https://diyhifisupply.com/tube-sockets/premium-teflon-sockets/

Searched older posts where Kevin mentioned which ones to get but links are all expired.

 

[audiostar]

Yes, ceramic sockets have bit of different pin legs which do not fit the board properly. Use the teflon sockets, you can get them from the diyhifisuppy you linked above or on ebay as well. 8 and 9 pin available, check out this seller https://www.ebay.com/itm/223597938937

Edited June 1, 2023 by audiostar

[demonkuro]

HI！My exploded T2 is back to life.

Today I finally connected the PSU to AMP after replacing all the damaged parts. Following the instructions in the forum, I set the R42 (22K) at 6.55V using RV2 (10K) and then set the battery at 740V (+190V/-547V) using RV1 (2K). When I was ready for success, I found that it was not so simple. The output R- to GND was -91V, R+ to GND was 0.05V, L+/L- to GND was -28.6V, and the BIAS (574V from PSU) was only 384V after 5M resistance.

I tweaked RV2 (10K) and RV1 (2K) to get a near-zero DC output, but I couldn't get a 740V battery at the same time and made the DC output zero.

Do I need anything else? Or are some parts damaged again? I checked, but nothing seemed to be found.

[jokerman777]

1 hour ago, demonkuro said:

HI！My exploded T2 is back to life.

Today I finally connected the PSU to AMP after replacing all the damaged parts. Following the instructions in the forum, I set the R42 (22K) at 6.55V using RV2 (10K) and then set the battery at 740V (+190V/-547V) using RV1 (2K). When I was ready for success, I found that it was not so simple. The output R- to GND was -91V, R+ to GND was 0.05V, L+/L- to GND was -28.6V, and the BIAS (574V from PSU) was only 384V after 5M resistance.

I tweaked RV2 (10K) and RV1 (2K) to get a near-zero DC output, but I couldn't get a 740V battery at the same time and made the DC output zero.

Do I need anything else? Or are some parts damaged again? I checked, but nothing seemed to be found.

don't know about anything else but the BIAS is probably fine, measuring after 5M resistor the finite impedance of the voltmeter forms a voltage divider with it.

[jamesmking]

2 hours ago, demonkuro said:

and the BIAS (574V from PSU) was only 384V after 5M resistance.

 

Measure the bias voltage before the 5M resistor. 

 

 

[GeorgeP]

3 hours ago, demonkuro said:

HI！My exploded T2 is back to life.

Today I finally connected the PSU to AMP after replacing all the damaged parts. Following the instructions in the forum, I set the R42 (22K) at 6.55V using RV2 (10K) and then set the battery at 740V (+190V/-547V) using RV1 (2K). When I was ready for success, I found that it was not so simple. The output R- to GND was -91V, R+ to GND was 0.05V, L+/L- to GND was -28.6V, and the BIAS (574V from PSU) was only 384V after 5M resistance.

I tweaked RV2 (10K) and RV1 (2K) to get a near-zero DC output, but I couldn't get a 740V battery at the same time and made the DC output zero.

Do I need anything else? Or are some parts damaged again? I checked, but nothing seemed to be found.

Try getting your 190v closer to 200v if possible. And then readjust after 30min or or as the value will change until the temps stabilize. If your offsets are still problematic then adjust as required (30ish volts or lower is good) but it might suggest an issue somewhere. 

[sorenb]

5 hours ago, demonkuro said:

... The output R- to GND was -91V, R+ to GND was 0.05V, L+/L- to GND was -28.6V, ...

Provided the readings are right, I'd look at the right channel LF353 and re-check all the diodes in the right channel servo are properly oriented
if everything seems right, pull the right channel LF353 and see what readings you get.

[simmconn]

The offset of the output stage is maintained by the Q33-Q32 servo and they'll fight against any adjustment you make on the active batteries in the attempt to bring down the offset. My preferred solution (which may not be popular here) is to change R73 (6.2k) to a 10k multi-turn trimmer so that offset and balance adjustments can be de-coupled.

In terms of balance, the balance servo around the LF353 has a relatively narrow compliance range. If you did not pay great attentions to the parts matching between the +/- arms of a channel, especially the triode-to-triode matching in the 6DJ8 tubes, chances are the LF353 opamps could bottom out trying to balance. Check the output voltage of the U7 and U8 opamps. The difference between them tells us how unmatched the two arms are. If any of them gets close to the rail (>10V or <-9V) I'd want to intervene. Adjusting the two active batteries' voltages in the same channel in opposite directions can bring the balance servo back to their comfort zone. Swapping the 6DJ8 tubes may also help, if one has better matched triodes than the other. Mismatching is not always a bad thing - slight mismatch adds some 2nd order harmonic which is pleasing to the ear. 740V on the active battery is a starting point/nominal, you may end up a couple of volts away from the nominal which would be fine.

[demonkuro]

On 6/2/2023 at 11:25 PM, jamesmking said:

 

Measure the bias voltage before the 5M resistor. 

 

 

Bias voltage before the 5M resistor is 574V.

[demonkuro]

18 hours ago, simmconn said:

The offset of the output stage is maintained by the Q33-Q32 servo and they'll fight against any adjustment you make on the active batteries in the attempt to bring down the offset. My preferred solution (which may not be popular here) is to change R73 (6.2k) to a 10k multi-turn trimmer so that offset and balance adjustments can be de-coupled.

In terms of balance, the balance servo around the LF353 has a relatively narrow compliance range. If you did not pay great attentions to the parts matching between the +/- arms of a channel, especially the triode-to-triode matching in the 6DJ8 tubes, chances are the LF353 opamps could bottom out trying to balance. Check the output voltage of the U7 and U8 opamps. The difference between them tells us how unmatched the two arms are. If any of them gets close to the rail (>10V or <-9V) I'd want to intervene. Adjusting the two active batteries' voltages in the same channel in opposite directions can bring the balance servo back to their comfort zone. Swapping the 6DJ8 tubes may also help, if one has better matched triodes than the other. Mismatching is not always a bad thing - slight mismatch adds some 2nd order harmonic which is pleasing to the ear. 740V on the active battery is a starting point/nominal, you may end up a couple of volts away from the nominal which would be fine.

Today I checked all the 2SK216 and 100V diodes and they are all good, and the 15V diodes and 1N914 I checked should also be good. Q33-Q32 is also good, HFE is at 50-52.

Then I recorded the voltage output to GND, @battery 740V (+193/-547).PSU：+251.3（+250V）+498.6（+500V）-503（-500V）-266.7（-260V）-564.5V（-560V）

Cold crank： L+ is -15V, L- is -15V, R+ is about 0V, and R- is -61V.

After 30 minutes: L+ is -13.4V, L- is -13.4V, R+ is about 0V, R- is -56V.

After 60 minutes: L+ is -12.3V, L- is -12.3V, R+ is about 0V, R- is -52V.

After 90 minutes: L+ is -11.6V, L- is -11.7V, R+ is about 0V, R- is -52V.

 

Then I also recorded both outputs of the LF353.

Cold crank: LF353(L), output 1 is 1.65V, output 2 is -6.65V, LF353(R), output 1 is -0.65V, output 2 is 9.66V.

After 30 minutes: LF353(L), output 1 is 6.2V, output 2 is -1.63V, LF353(R), output 1 is 0.96V, output 2 is 11.07V.

After 60 minutes: LF353(L), output 1 is 6.96V, output 2 is -1.63V, LF353(R), output 1 is 0.05V, output 2 is 11.07V.

After 90 minutes: LF353(L), output 1 is 11.09V, output 2 is -2.28V, LF353(R), output 1 is -0.15V, output 2 is 11.07V.

 

During the 90-minute test, the voltage of the 740V battery dropped by about 2V. All the LED lights are on, the AMP radiator is very hot, and the PSU radiator temperature is lower.

 

It seems that only R+ is always working fine？Over time, the voltages from the other outputs to GND are always drifting. This confused me and I didn't know what else I should check.