Please wake me up from this PCB nightmare [Krell KSP-7B Power Supply]

Agree with Naresh, you could use the phone or REW with a soundcard or usb dac. If it is 1V out, then change my divider for 50/5K to 10 ohms/10K to get 1000:1 reduction in voltage. You will need to get the scope working though to probe around.
Real waveform generators are surprisingly expensive.
 
The DIY shifts to CYDIFM 🙂
Never heard of CYDIFM so I googled it. This diyaudio thread turned up, in which you expound at length.

I have volumes to say but I'll just make two brief comments:
1. For better or for worse (but mostly the latter), I can completely empathize with the things you say in that thread.
2. And it is completely wrong.

That said, there is I think a good discussion to be had over what constitutes acceptable requests for help in forums while being respectful of people's time, and I am grateful for the generosity with their time and expertise of the folks who answer questions, not just here but in every forum.

I never ask others to "do it for me" because I won't learn that way. If I'm stumped, like with my Krell's transistors that overheat despite the fact that absolutely nothing downstream overheats, then a nudge in the right direction is always appreciated.

My idea of forum use is to (a) learn, and (b) leave detailed documentation and resolution of the issue I'm working on for the benefit of the next guy who has the same problem, tomorrow or in 5 or 10 years.

I will update the thread.
 
Hi @mkc

Here it is. Sorry for the low production quality, I can redo it the way I did the linestage if this one is too hard to read. Let me know. Thank you!

1000003296.jpg
 
Hi @MFRD

This is perfectly fine. I do have some difficulties in tracing the tracks for the MC/MM switch. It appears that there might be some gain switch involved. I do not know if you can amend my schematic. Also, the PSU regulators. I can't determine if the MJF150xx are in parallel or as darlington. Perhaps you can determine? The feedback network of the second-stage in the phonp-stage is a litlebit difficult to trace from the pictures. Can you help?

I have attached what I have at the moment.
 

Attachments

  • Thank You
Reactions: MFRD
@mkc

Thank you, and I will let you know the answers to the questions you asked.

Interesting you mention the MC/MM switch. I was testing it for continuity and it looks like it's shot. Pin pairs read open when they should be closed, and at times I observed resistance across pins located in opposite rows.

The faults seemed intermittent and kind of random, but it was 2 am therefore more testing is in order

I was testing in circuit, but that can't explain individual switches reading open when they should be closed.

The switch is a 4PDT toggle with two six-pin rows (Alcoswitch TT41).
 
@MFRD

The diodes in the regulator seems to be "constant current diodes". This i a guess, but something else doesn't make sense. Can you see the type number?

Also, the PSU part with the most discoloring is the part that supplies the line stage, as far as I can determine. It somehow makes sense given what you have observed. The "upper" section supplies the phono stage.
 
I can't determine if the MJF150xx are in parallel or as darlington. Perhaps you can determine?

Each two MJF1503x back to back on a single heatsink are a Darlington pair.

Also, the PSU part with the most discoloring is the part that supplies the line stage, as far as I can determine. It somehow makes sense given what you have observed. The "upper" section supplies the phono stage.
I re-read your post, and it finally clicked! Better late than never, I guess.

So the "lower" half of the voltage regulation section (the "hot" half) has nothing to do with the phono section, it supplies the line stages and the load on it is determined by the draw from the line stage transistors, and therefore by their bias settings. Suddenly it all makes sense! Thank you for that, I was fixated on that whole regulation section serving the phono stage.

So back to bias setting. I originally dialed 57 mV across emitters (86 mA).

Today, first I dialed it down to 40 mV (61 mA), then to 33 mV (50 mA). Here is a temperatures chart with the different biases:

main board temps.jpg


I also double-checked DC offset: still .01 mV on both channels.

Next step, listen to music!
 

Attachments

  • main board temps.jpg
    main board temps.jpg
    66.6 KB · Views: 16
Last edited:
  • Like
Reactions: Boydk
I have been listening to the Krell in my system a few hours every day. It sounds glorious! I am very pleased with the sound quality. The new TKD volume control is excellent, I recommend it.

Operating temperature and current draw are moderate and stable, and I am not hearing a difference in sound quality for having cut the bias current almost in half.

Some trim pots are cooked so I ordered new ones and I will replace them throughout. They're sealed multi-turn cermet Bourns 3296W series.

I sprayed a bunch of deoxit into the MC / MM selection switch, but that did not fix the issue where one of its four switches is correctly closed when the toggle is in one position, but is open in the other position whereas it should be closed.

It's a discontinued Alcoswitch with 2 rows of 6 pins (2x6 pattern). There are no replacements out there except for a couple NOS ones asking 50+ euro. All new 4PDT toggles on the market today are 3x4 pattern. 2x6 sliders are available, but the PCB hole patterns don't match.

I may have lucked out on ebay though. A guy is selling a bunch of NOS Alcoswitch, exact same ones as mine except they have 90 deg angled pins instead of straight pins. It's not a drop-in replacement (again the PCB hole pattern doesn't match), but I'm hoping they are internally identical so I can use parts from the angled one to repair mine. Since he's asking $3.75 each I ordered two. We will see.

I also ordered parts for my overheating Carver tuners and my nonworking oscilloscope. I sure could use a scope.

I think that's it for now. I'll update the thread after I install the new stuff.
 
  • Like
Reactions: Boydk