KLH 20 Weird Behavior - Repair Project

[kf7lze]

It's been a very long time since I did any repair projects, but I've had some time on my hands lately and got inspired to spend some of my hobby time on this again. Starting with a KLH Model 20.

The nicest-looking one was totally dead when I found it, it'd power on and draw some power, but no output or response anywhere. I replaced the electrolytic capacitors and cleaned the controls, adjusted the bias per service manual, cleaned the function switch, and now got a test signal through the unit but with a bad channel imbalance when hooked up to a dummy load.. The volume control ended up having cracked leads at the board, so I replaced it with a donor (I have a few parts units I got at the same time) and checked the other controls which looked good but they got fresh solder for good measure.

Now, with the unit connected to a test signal (1 kHz tone) going into the Aux input, and the amp connected to a dummy load, the channels are beautifully matched and show perfect sine waves on the scope. But if I hook it up to my speakers...it's bad. Wild siren noises at maximum volume until I quickly power the unit off. I torqued down the chassis bolts which hold the halves together, and this didn't change anything.

Okay, not great, but I figured it might be a grounding issue - with the unit on an isolation transformer, and an isolation transformer on the audio signal, I get a moderately loud feed-through tone even at zero volume from the left channel, and never manage to get any sound out of the right channel at all. There's a lot of noise as well.

Voltages look pretty close, -36V was a -39V but the others were within 10% (edit: 39 vs3 6 is....also within 10%. Math.) Wire routing is all original. This is a later production one, it looks like, the power amp board has 1 fewer capacitor populated and it has silicon output transistors. It also has an MX-3 (or maybe the MX-6) MPX board, although the power supply has a 4th capacitor populated which is labeled in the manual "not provided with MX-3 board" so I'm not positive, but I already know KLH has played mix-and-match on different production revisions without great documentation from past experience years ago.

I used to consider myself an advanced tech (although not quite an expert) but it's been long enough of not doing it or thinking about it that loads of the knowledge has been paged out. I'd love some guidance and suggestions for where to go next. I have detailed photos of the work and some scope waveforms saved I can share if that's helpful at this stage. I appreciate anyone's input here!

 

[kf7lze]

Updating here while I keep at it:

Pin connectors had oxizided a bit - removed, cleaned, and reseated all of them; this helped with the zero-volume feedthrough, hum, and general noise.

The amp now plays well, one channel at a time:

Left channel input only, left speaker only: good
Right channel input only, right speaker only: good
Both L+R inputs, left speaker only: good
Both L+R inputs, right speaker only: good

Both inputs and both speakers connected, bad - no output from R-Ch in this arrangement. L-Ch sounds OK if the balance control goes full left; with it centered, it starts to have some noise and distortion. With the balance full-RIGHT, I still get nothing from the right channel, and I get badly distorted feedthrough on the left channel.

Left input only, and switching to MONO, gives the same as L+R inputs in STEREO - good left with balance set left, bad right. Using AUX2 for testing, which based on the schematic looks like it comes in after the phono preamp transistors. I'm going to hook up the signal to the Control Board's inputs and bypass the entire PAB to eliminate it for sure as the next test. I'm able to get perfect audio from each channel separately now (was not the case before cleaning the pins) so this seems to me to point towards something earlier in the signal chain.

 

[gadget73]

sounds like an open ground somewhere maybe? The un-used channel may be causing the whole thing to act sort-of like a differential amplifier, input on one side, nothing on the other = output. Input from both and you just get the differences between them. I've run into this when the speaker negative was open, but only if both speakers were connected and both - were connected together.

 

[kf7lze]

Yeah it does kind of seem that way...but I'm having a heck of a time finding it. Since last post:

1. Disconnected front end from power amp boards and fed in a signal directly. No change in behavior.
2. Re-tightened chassis screws for grounding (per service manual recommendation), removed, tightened connectors, and reattached all connections.
3. Re-soldered output transistor socket connections
4. Pulled a Power Amp Board from one of my parts units (luckily enough, same revision!), replaced the 4 caps and swapped the whole board. No change.
5. Ran extra grounding from the power supply to the power amp board grounds.
6. Traced all the wiring around the transistors, speaker terminals, output caps, etc. and confirmed everything appears wired correctly and resistors are in range.

Attached the schematic for this one here. (Revision K)

 

[elvi]

What would a shorted can in the signal path do? I also think it's an open or nullified signal ground.

 

[gadget73]

so here's a bit of super retro funkiness



That resistor that sits between speaker negative and ground actually produces current feedback rather than the usual voltage feedback. In the mid 50s, a pot to control how much feedback was voltage vs current was a "thing", and called a variable damping control. If you got the balance just right, instead of a constant voltage or constant current amp, what you got was a constant power amp that had output Z equal to the speaker Z. I actually don't even see a voltage feedback path, typically that goes from speaker + back to the input. I guess it may have a partial loop feedback through R304?

anyway if one of those 0.27 ohm resistors is bad, or the ground to it is bad, it will cause some strange problems.

 

[kf7lze]

I did actually measure both of those resistors, and confirm the resistance to ground, and it all checks out -- but at this point, I might just replace them anyway, and run fresh grounds while I'm at it.

The facts that this weirdness persisted (1) with the front end totally disconnected, and (2) with a totally different driver board for the impacted channel at points the way (at least in my mind) at the very few components which stay. That's really just the output transistors and sockets, and the 0.27 Ohm resistors.

This is the late revision with Silicon output transistors (2N3055s), not the earlier Germanium ones. But I think everything else is PNP transistors. Maybe that accounts for feedback coming off the - rather than the +?

 

[gadget73]

That shouldn't affect it, though I admit to not really being a solid state guy. I mostly recognize that because its a solidly mid 1950s tube thing that I don't believe I've ever seen in a solid state amp before.

it does point to some problem external to the driver, but there just aren't that many parts left.

 

[kf7lze]

Okay, a conclusion! It was actually a broken ground wire at the bottom of that channel's 0.27 Ohm resistor to ground. I lifted the low end of the resistor and ran a new ground right back to the power supply, and it was all perfect.

The ground for the speakers is actually wired as separate grounds wires going to a few - an isolated lug on a terminal strip, then a jumper to the bottom side of Q325, then back to a dedicated ground terminal on the power supply board. (A different version of the schematic has that actually labeled "+", and then the power supply schematic also has a connection between "+" and the ground symbol, so it's at the same potential but run separately.)

The first segment of wire, from the low side of the resistor to the lug at the socket for Q325, seemed to be broken inside of the insulation. And it looks like the orientation I had the amp out on the bench when connected to the actual speakers, vs. to my Sencore PA-81 dummy load, was just enough that the wire was flexing...

Strange problems indeed. I've actually run into this exactly once before, a LONG time ago, on a 1934 tube radio I was fixing up.

Problem solved, and thank you very much for your input @gadget73 - really appreciate you chiming in.

 

[elvi]

Thank you for closing this thread.