• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

TSE - R33 and R29 blown

Hi All,

Looking for a little diagnostic advice. I got my TSE up and working about a year ago and it’s been absolutely flawless. I typically run it for hours on end, and while it does get uncomfortably hot, it’s performed perfectly. Until this morning. I was running it at about mid volume when one of the speakers popped/crackled, then no sound. I ran over to it, cut the power, and noticed some carbon burn marks on the base of the left channel output tube. Opened it up and saw R33 burnt to a crisp. Replaced it with another 1/4w 100 ohm resistor, reassembled everything, and turned it back on. I then noticed a puff of smoke (no sound from speakers, despite active input signal). Shut it down, opened it up and this time I noticed R29 burnt. The picture below is after I replaced R33, but before doing anything with R29.

An externally hosted image should be here but it was not working when we last tested it.


I’m wondering if anyone had seen this pattern of failures? Or if anyone can offer any hypotheses? I wonder if R33 failed for whatever reason and perhaps ruined my left 300b tube? Is that likely or possible? If so, would a bad tube cause R29 to blow?

I suppose my next step is to replace R29, remove tubes, and go through the checkout process again. Unfortunately I don’t have a way to test tubes (short of using a multimeter, if that’s even possible). Any thoughts or advice? Thanks much!
 

Attachments

  • 23E847C6-7809-423A-B7BA-2176597BFF14.jpeg
    23E847C6-7809-423A-B7BA-2176597BFF14.jpeg
    32.8 KB · Views: 77
Last edited:
Failures of R29 and/or R33 would be symptoms of another problem,
probably a bad output tube, but the fet could also be damaged.
The input tube is not likely to be part of the problem.

Use your other channel's output tube for testing after the repair/checkout,
if you don't have a spare.
 
Last edited:
I have never seen an R29 failure. That would take a lot of current through a 300B to blow it. The first thought would be to measure the filament continuity in the tube with an ohmmeter. It's the two fat pins and they should measure 1 ohm or less. If the filament burned out and the filament wire touched the plate or grid these two parts could get burnt.

The more likely failure would be a blown mosfet. R33 is the 'fuse' that protects the tube if the mosfet shorts out. This can be tested with a voltmeter on the grid pin (or R33) with NO tube in the socket. The voltage should be variable by turning the bias pot R23. It should go from a large negative voltage near -100 volts, to a lower negative voltage, maybe -20 volts. A positive voltage that does not adjust, or only moves a little, indicates a blown mosfet.
 
George / Rayma - thanks. Very helpful.

George, the tubes checked out - about 2 ohms across the fat pins. I checked the mosfets as you described. Voltage at the right channel grid pin registered fine - started very negative then stabilized around -45v. Voltage at the left channel grid pin started the same, then shot up to +450v. I ordered a few replacement mosfets this morning. I’ll report back after I replace them. Really appreciate the help!
 
Opened it up and saw R33 burnt to a crisp. Replaced it with another 1/4w 100 ohm resistor, reassembled everything, and turned it back on. I then noticed a puff of smoke (no sound from speakers, despite active input signal). Shut it down, opened it up and this time I noticed R29 burnt.
Whenever such parts failure occur and replaced, the turn on process should revert back to the initial "check out" steps with all the tubes removed so that any other failed parts can be revealed.
 
The more likely failure would be a blown mosfet. R33 is the 'fuse' that protects the tube if the mosfet shorts out. This can be tested with a voltmeter on the grid pin (or R33) with NO tube in the socket. The voltage should be variable by turning the bias pot R23. It should go from a large negative voltage near -100 volts, to a lower negative voltage, maybe -20 volts. A positive voltage that does not adjust, or only moves a little, indicates a blown mosfet.

I finally received replacement mosfets. I replaced Q2 and unfortunately I’m still getting a high positive voltage reading at the left channel grid pin (no output tube installed). I even tried yet another new mosfet and still saw a high voltage. I’d think three consecutive bad mosfets is a fairly low probability and so I’m pretty sure my issue is somewhere else. Any other ideas? I’m wondering if either U3 is faulty, or perhaps a bad C11 cap, but I don’t know if either of these would result in +450 volts at R33. I’m truly at a loss. Any thoughts?
 
If the voltage at the plate of the 5842 is normal and still adjusts by turning R20, the U3 chip is good.

A shorted coupling cap (C11) could put the plate voltage of the 5842 into the mosfet circuit, but it shouldn't be in the 400 volt range. You could lift one end of the cap to eliminate this possibility.

The source resistor from the mosfet to the negative supply, R25 could be open allowing the voltage to float high.

There is a voltage divider made up of the pot (R23), R22 and R24. These provide the adjustable negative gate voltage for the mosfet. Any of these could cause issues as could R26. You can probe the center pin of the pot or the end of R26 closest to the pot to verify that the negative voltage is present and adjustable.
 
Thanks George. Really helpful. Here’s what I’ve found so far ...

If the voltage at the plate of the 5842 is normal and still adjusts by turning R20, the U3 chip is good.

I confirmed 5842 plate voltage is good.

A shorted coupling cap (C11) ... You could lift one end of the cap to eliminate this possibility.

Lifted one end, still saw a high voltage. Moving on ...

You can probe the center pin of the pot or the end of R26 closest to the pot to verify that the negative voltage is present and adjustable.

I checked this with 3 meters. One on B+, one on R26 and one on R15 (both on leads closest to pot). I checked R15 as a reference since the problem seems to be only on the left channel.

B+ measured at +452 volts. R26 measured at +57 volts and R15 measured -42 volts. I would have expected a lower (more negative) voltage off of R15. Could there be some interplay between channels that result in a less negative R15 voltage?

I’m leaning toward just replacing R22 - R26, including the pot, unless there’s a better solution at this point? Also, is there anything on the power side that’s worth checking? Thanks again for all the help!
 
I would have expected a lower (more negative) voltage off of R15.

-42 volts on R15 is probably about right. That would translate to about -45 to -50 volts on the grid of the tube in the good channel.

+57 volts on R26 is very wrong since all voltages on the pot should be negative. This could be caused by an open R25. Measure it with an ohmmeter with board power shut off for a few minutes to allow the caps to discharge. You can compare it to R14. I have seen these resistors burn, or silently go open without obvious visual damage when the mosfet blows.
 
Would the deviation from 36k on R25 be enough to cause a positive voltage at the pot?

If it's conducting current there should be no problem. The absolute value od the resistor is not important for circuit function.

Something is either causing the mosfet to conduct excessive current, or there is an unintended path for current to flow around the mosfet (solder bridge).

What is the part number of the mosfet that you put into the board?
 
What is the part number of the mosfet that you put into the board?

iH9hp91.jpg


Here's the actual one (click picture for better image). The solder pads on the board look pretty torn up, but I've checked that there's no connectivity issues between these pads. The underside of the board looks much cleaner.

AHJVnH9.jpg


Here are some additional pictures of the board as it is currently. Tubelab SE - KHG album
 
Last edited:
Those are the exact same fets that are in the board that's in my hand now. Even the SPH30 (date code?) is the same.

Try resoldering the end of R25 closest to the pot from the top. The solder joint looks a little odd, at least in one of the pictures. I have seen some odd issues with solder joints that get thermal cycled a lot, and those resistors will get real hot when the fet shorts out.
 
Try resoldering the end of R25 closest to the pot from the top. The solder joint looks a little odd...

Thanks, tried that, but didn't help.

However, I did fix the problem! I broke my own rule and did two things at once, so I probably won't ever know exactly what the issue was. I replaced the fet with yet one more (my last new one) and I decided to remove the chance of a solder bridge. Because the pads on top looked so bad, I broke the path from the center fet pin and bent the center pin of the fet and bridged it directly to the B+ pad. Looks crude, but it works! Really appreciate all the help!
-KHG
QDPMnQc.jpg
 
Well, I finally got the left grid pin to measure a reasonable negative voltage. I completed the checkout process and adjusted the tube bias. All went well until I connected an input signal and listened. The left channel seems fine, but the right channel is acting up. One thing is it "pops" as it powers up (heard through the right speaker only). This has never happened, with either speaker, ever. The second issue, and more important, is that the right channel has no or very little high frequencies coming through. I swapped speakers (left / right) to be sure it wasn't that, and I also swapped the input and output tubes. It's something else. I was going to swap output transformers next, but before I go unsoldering stuff again, I was wondering if there's a more obvious solution - some other component (C11 cap, perhaps?) - that would be more likely to result in no high frequencies? I hope it's not the other mosfet ...
 
Last edited:
I switched the 300b tubes (left to right) and played the amp this morning for about 20 minutes - when the right channel popped and then completely cut out (no sound). I measured the voltage of the right channel grid pin and it's reading a high positive value, so same issue I had before on the left channel. I believe the right channel mosfet is toast.

George, is the mosfet you list on the TSE-II BOM (STF3LN80K5) a suitable replacement?

Is it possible that my tube is causing this? Perhaps an intermittent short (grid to plate) within the tube? I'm using genalex gold lions. Just seems strange that the blown mosfets are on the channel using one particular tube.
 
is the mosfet you list on the TSE-II BOM (STF3LN80K5) a suitable replacement?

I haven't actually tried them in an original TSE, but they are in my TSE-II's and the circuit is similar, so they should work.

Is it possible that my tube is causing this? Perhaps an intermittent short (grid to plate) within the tube?

A grid to plate short would take a gross failure in the tube since they should be spaced pretty far apart. It would put B+ on the source of the fet, which could cause damage, but I have never investigated this.

A grid to filament short would put ground on the source of the fet, which already has B+ on the drain, likely blowing the fet. I have not seen such a short in a 300B, but I HAVE SEEN it in an 811A, so it is possible.

Look carefully into the top of the tube and see it all of the filament wires are still in hooks in the support springs, and all 4 of the springs are the same height. I had several early vintage Sovtek 300B's fail with HALF of the filament going open. The tube still played reasonably well, but wouldn't make as much power as it should. This was a pretty common failure mode for the Sovteks about 15 years ago. Sovtek would not replace them, just told me, and other buyers to buy their new EH 300b's since they were better....I bought cheap Chinese tubes instead, and they are still good today.
 
Last night I went through a series of attempted fixes. I replaced the right channel mosfet (Q1) using the one recommended for the TSE II (STF3LN80K5). The left channel still has the one referenced above (NDF04N60ZG). I tested the voltage at the grid pin and experienced a high positive voltage. So, I skipped ahead and did the same thing I did on the left channel -- replaced the mosfet again, "wiring" the center pin above the board directly to B+, and severed the copper path on the board to try to eliminate any potential shorts between the mosfet pads. I re-tested the grid pin voltage and this time it held a lower negative value. Good. Put the output tubes in (with one new tube), powered it up, set the bias (without input or speakers connected). Still good. I let it run for about 30 minutes, then reassembled the amp back into the chassis and connected an input signal and speakers. I immediately heard a pop from the right and no music.

Okay, back to the bench. Replaced the fet again, replaced R18 for good measure, checked resistance of R34 and R14, checked grid pin voltage, commenced checkout procedure (no input or speakers connected), everything good. Reassembled chassis, connected input and speakers, pop / fizz from the right channel - this time R18 burnt.

One more time - replaced R18, checked grid pin, measured other resistors, checkout process was fine on bench. Reassemble, connected speaker and input, powered it up, heard an immediate pop from right channel.

I unplugged everything and went to bed.

As I slept, I started thinking of everything that could be causing this and potential troubleshooting next steps:
- Do the mosfets interact at all such that they need to have similar characteristics? In other word, do the mosfets need to match?
- Could there be a short somewhere on else on the board - perhaps at the line-in or left channel speaker terminal (or in the output transformer?!).
- Could there be a short in the right speaker itself (old Klipsch Heresys)
- Maybe the issue is still at the mosfet. Perhaps it's worth attempting to wire it point to point above the board and just sever all the paths on the board that lead to/from that mosfet?
- Maybe I just order a TSE II and rebuild everything re-using my transformers and tubes, this time with better ventilation (By the way, F-4V / F-6V work perfectly to power a small 40mm 12v fan).

Any thoughts / ideas / help are greatly appreciated!
 
Last edited:
Well, I'm back up and running (for a few hours now, anyway). Just wanted to bring this thread to a close in case others experience the same issues.

I replaced R34, since it had blown. I also replaced the right channel MOSFET again, using the one recommended for the TSE II. It seems to work fine. Upon replacing the fet, I did what I did on the left channel and bridged the center pin above the board to the B+ pad immediately in front of it and severed the path on the board leading to the center pin.

I also severed the board between the center fet pin and the left fet pin (the one closest to Q2 if looking down onto the board). I did this on the left channel as well. The board almost seemed hollow between these 2 pads. When I was scraping away all the crud, the knife blade sunk easily into the board at that point. I wonder if this small area is susceptible to shorts - perhaps promulgated by the heat buildup over time? The same happened between the same 2 pins at Q2.

I added a small fan and raised the chassis up to create a bit more airflow. Hopefully this will help keep the temperature down a bit.

Thanks again for everyone's help here.