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

@rhthatcher

The dual-rail SLB is 142.5 mm x 100 mm. The Krell case is 200 mm x 156 mm ID, so the SLB fits crosswise. Leaving 7-8 mm around the PCB for ventilation leaves approx 90 mm for the transformer, whose diameter is... 90 mm.

Standoffs would need relocated. Also, not sure if the populated board would fit heightwise. I have to read their BOM

Dual rail SLBs are expected within a few days it seems.

Hmm
 
Back to the main preamp, I recorded trim pot settings as found before adjusting bias. The one trim pot associated with the overheating transistor pair kept giving me fluctuating readings, whereas the other 3 were stable. I gave it some attention until it gave me a stable reading too, and I set all 4 pots within a 510-530 ohm baseline.

Time to power up. I immediately noticed power draw was only 750 mA per channel, down from 1.55A (at startup) last time!

After a 10-15 minutes warm-up time, I recorded voltages across the 4 emitter pairs: 51, 51, 57 and 70 mV respectively, the latter being the previously overheating pair. So I think that pot was in a state that caused trouble leading to overheating, yet it looks like that one pair is still a little "hot" compared to the other three.

I scientifically picked 57 mV as my target voltage, based on (a) the inferred bias current being 86 mA which didn't seem completely out of whack for a Class A preamplifier, and (b) numbers superstition 🙂

With that, I observed power draw stabilized at 640 mA per channel (half as much as last time!)

adjusted bias.jpg


Transistor temperatures are now between 40 and 47 C. The previously-hottest pair (113 C!) is now in the mid-range.

I did notice the bias voltages fluctuate constantly within a +/- 1 or 2 mV range.

All in all I'm pretty stoked with this round of testing!
 
Last edited:
  • Like
Reactions: Simon1972 and Boydk
I am attaching two documents that might be interesting if you're into vintage Krell:
  • KSP-7B datasheet
  • KSP-7B user's manual

From the datasheet we learn this

All amplifier stages are high-bias Class A with individual tracking regulators.

Would have been nice if they'd attached a number to that statement...

And this:

Additional voltage regulation for the phono stage increases isolation from the line stage and enhances its superb low frequency definition by effectively minimizing 60, 120, 180, 360Hz residual power line components from blurring bass reproduction.

So I was correct that the left side of the top section in the pic in post #141 is in fact a final power supply regulation stage, although it only serves the phono stage.

And the user's manual confirms the gory truth:

The KSP-7B is designed to be "on" at all times. In other words, it is not intended to be switched on and off on a routine basis. The only times it is recommended to turn the unit off is when your system will be left unattended for a long period of time, such as during a vacation or a
business trip. In these cases all other equipment should be turned off, and the KSP-7B power supply should be disconnected from the AC
mains. When you return, reconnect the power supply to the AC mains before turning on other equipment.
This procedure should also be followed if you reorganize your system and need to change wiring of the supply to the preamplifier.

So there 🙂

Would still be great if a service manual magically turned up 🙂
 

Attachments

  • Like
Reactions: Boydk
Nice job, one step closer.

I'd advise against Krells leave it on advice. Never leave anything on that runs hot. 20 minutes warm up is more than enough for everything to stabilise. My ksp needed no more, same for my KSA200.
Especially given power prices. Intel used to say a dollar a watt in the early 2000's. Now I'd say it is closer to 2 dollars a watt. Cal might even be 3 or 4. Power strips make it easy to turn off always on.

And enjoy the new bench supply as an added bonus of the fix!
 
Exactly. Intel was saying for machines that run all the time a machine that draws 1KW will cost you around 1K/year to run. I still run my main machine 24x7, but with better components of today versus 2 decades ago, it only pulls around 150W. 2 decades ago I ran 3-5 machines continuously and I did notice it in the bill. I probably was pulling around 2KW continuously back then.
 
I checked DC offset, there was 0.1 mV on the right channel and 19.1 mV on the left. I adjusted it, it shows 0.1 mV on both channels now.

However I totally forgot about the voltage regulators in the phono stage. As noted before, half of them are running twice as hot as the others. There are four pairs, each pair on a heatsink. Two pairs are running at 57 C, the other two at 28 C. As you can see on the main board, temps in the line stages look great but those ones in the phono stage at top left are out of whack. Here is the main board:

ir main.jpg


And here is the phono stage:

ir phono.jpg


Nothing appears to be drawing abnormal current elsewhere in the phono stage. I checked voltages at the 8 transistors and they are within these values:

Base to ground: +/- 17.75 to 18.59 V
Collector to ground: All 8 exactly +/- 20.91 V
Emitter to ground: +/- 17.14 to 18.05 V

That's all the troubleshooting I've done so far I'm afraid. I haven't seen any obvious cause for the huge temperature imbalance yet.
 
  • Like
Reactions: STOXX
Most likely resistors and capacitors out of whack due to long term heat exposure.
Check those first, if in doubt remove from board and check.

Use a good quality meter, the graphic display types for capacitance are decent for the money.

Comparison between same points on different channels helps find defects, the same parts are rarely defective on both channels.
I would ask if all the electrolytics are due for replacement due to end of life, leaky capacitors can be a source of trouble.
 
@NareshBrd you are correct of course. At the same time, I trust you empathize with my utter reluctance to recap this PCB, considering it is of the same dodgy manufacture as the power supply PCB, thus removing components without ripping off a bunch of too-tight stick-on eyelets and traces involves quasi-magical skills I am not sure I possess.

If I have to do it, naturally I will, but I need some rest first 🙂
 
I would build a new PCB, with thicker foil.
And mostly new parts, the semis are degraded as well.
New capacitors, resistors and so on.

Possibly converted to SMD and assembled in China, minimis rules still apply I think, so any shipment below US$800 will not attract tariffs.

I have used cyanoacrylate adhesives to stick tracks that came off back to PCB, and also to secure eyelets...the smoke is an irritant.
Not recommended unless you have no choice.