Okay, interesting discovery tonight. It's only when both the TDA2040 heat sink (AGND1) and the input jack ground (AGND) connect with the chassis that the signal drops out. So that would seem to short the two grounds together, which I assume are suppose to be isolated?

This would mean that the thermal paste between the TDA2040 and its heat sink was the only thing preventing this from happening previously until it wore away?

AGND1 has a circle around it in the circuit diagram AGND does not (and input jack ground actually says "CHAS", what does that indicate?

Update: Circled Ground No. 5018 Noiseless (clean) earth (ground): To identify a noiseless (clean) earth (ground) terminal, e.g. of a specially designed earthing (grounding) system to avoid causing malfunction of the equipment.

Quote from: g1 on August 06, 2019, 01:28:39 PM
I'm not sure your test set-up, but it's critical that you do not ground the speaker (-) with any test gear, grounded load, etc.
If you do so you short out R43, which can cause all manner of weirdness.

Hrm, I may have inadvertently reverse the speaker polarity, but I don't believe I've grounded it out, something to check. Also that the speaker leads are short-free too I guess.

The grounding is intentional, there's a conductive plate that is soldered to the jack's ground lugs and folds in front of the jack to ground with the chassis. Both for the instrument input and the CD input. If it didn't seem intentional, yeah, I'd just isolate and close it up, but something's off.

The picture is a switching socket, inserting the 1/4" plug lifts the contacts opening the switches.

Upsetting the bias sounds like it may be the case because the max volume seems to fluctuate if the ground contact is intermittent.

I replaced C26 which is listed as 100nf, but the actual component was 200nf.

The board can ground to the chassis and everything works -- however if the ground of the input jack is connected to the chassis (as it is when assembled as a plate on the jack tightens against the chassis) the signal drops out.

Thoughts? Summarizes as the signal lost when input ground connects to chassis.

Quote from: psafloyd on July 04, 2019, 02:10:56 PM
There are at least two editions of the 158 Rage. The later/latest one had a switch to switch between modern and vintage voicings. If your one has this and the schematic doesn't, it could be more than the switch got altered.

Mine does have this switch, but the schematic seems pretty bang-on, I'll have to look closely for differences.

Quote from: g1 on June 20, 2019, 01:46:19 PM
How about resistance of R44 to ground?  (try both sides)

2.5m on both sides. Resistor reads proper at 4.7r across.

Quote from: g1 on June 20, 2019, 12:10:20 AM
How about checking resistance from heatsink to each of the IC pins.  The only one that should be low resistance is pin 3.

Pin 3 is a dead short, as expected. The other pins are 250k or over 1m.

Quote from: g1 on June 19, 2019, 04:24:53 PM
Can you check same with the board fully mounted?

Basically the same, once side climbs up to 260k, the other climbs down from 1.2m until about 300k then drops to 0 and starts climbing up.

Quote from: g1 on June 17, 2019, 12:09:52 PM
Check with meter set to ohms, lowest range.  Across D10, with it in circuit.

Seems to be all over the place. Either climbing to around 250k Ohms one polarity, other polarity drops from from 1m Ohm. But I also get all sorts of other readings across it.

Quote from: g1 on June 16, 2019, 04:29:48 PM
Schematic shows pin3 connected to ground, so no insulation needed for the IC.
Check D10 in circuit with your meter.  Check for any solder bridges at the IC pins.

D10 and D9 I get a beep both polarities with the diode test on my meter. I pulled them out before and they tested fine out of circuit.

None of the photos of this amp that I've found (three or four examples) have any wafer under the IC, just some thermal compound. So something else must have gone wrong.

Quote from: Enzo on June 15, 2019, 08:39:53 PM
The thermal compound grease is NOT an insulator.  There should be a mica or silicone wafer between the power amp IC and the heat sink.   You are shorting the power supply to ground.

There is/was no wafer -- should I add one? Is that the solution here to prevent the board from grounding on the chassis? Or would this be happening due to a component failure?

Yeah it seems like if I ground the board to the chassis anywhere then I lose signal.

It seems the power transistor is grounding through the heat sink to the chassis and that causes the signal to cut out. So by taking the board off everything worked, putting it back the transistor grounds and the signal cuts out.

I guess previously there was enough heat sink compound to prevent the short, and when it got hot enough it conducted, when I replaced the board the first time I must have put it on more secure and therefore made the short permanent (until I took the board off again).

It seems, by the layout that it's suppose to ground here anyway. Could they really have been depending on the heat sink compound to prevent this?

Unless the TDA2040 has an internal short.