Eden WT-800C Clipping problem

[nevetslab]

I have an Eden WT-800C on the bench, with one of the two power amp modules having two pairs of MJL21193/MJL21194 pwr xstrs in place of the three pairs of 2SA1492/2SC3856 pwr xstrs. Under 4 ohm load, the positive half of the output clips very prematurely, while the negative half gets full output swing. Under no load, or higher load values (8 ohm +), I don’t have the premature clipping.

I swapped out the voltage gain stage xstrs Q7 & Q8, which made no difference. Also changed out the upper driver Q10, which also made no difference.

I measured the voltage across the collector resistors of the input stage R8, R11, finding 700uA current flowing there. Measured the voltage across the emitter resistors R19 & R22 of the Voltage Gain stage, finding 12.5mA flowing thru that stage. Checked the voltage across the emitter resistors of the two drivers Q10 & Q11, finding only 5.3mA flowing there (all these readings under idle, 120VAC AC mains. Power supply rails are about +/-80VDC. I had set the bias per recommendation of 7.5mV across TP A & B.

The Driver circuit in this amp, with the collectors of Q10 & Q11 is fed thru 6.8k 2W resistors off the main supplies, and have 5.6V zener diodes at that junction of the R24/Q10-collector and R25/Q11-collector. I’ve never stopped to understand just how this works, as I’ve never had any issues with it, apart from having the zeners fail short, drivers & outputs fail, emitter resistors in the output stage fail….typical power amp failure mode stuff.

I decided to increase the collector current of the drivers by lowering the emitter resistors R26 & R28 to 47 ohms, now running 11.3mA. That made the clipping level far worse. When I calculated the voltage across those 6.8k collector resistors, the original value with 5.3mA flowing showed 36V across them, and with 11.3mA, now 76.8V. I ran out of time at this point, needing to go catch my bus home. I obviously don’t have a good understanding of this driver circuit, nor have I made comparative measurements against the other amp module in the chassis, it having the 3 pairs of 2SA1492/2SC3856 pwr xstrs. No hint of this clipping behavior.

I’d think at idle, with those 5.6V zeners (D50, D3, both 1n4734) connected to the output bus, I’d see 5.6VDC there. I was guessing maybe that zener voltage would end up being 5.6VDC less than the rail voltage at full output, but, I’ve never checked.

Anyone understand this circuit, or have a clue why I’d be seeing such severe positive clipping under high current output loads?

[glebert]

I’d think at idle, with those 5.6V zeners (D50, D3, both 1n4734) connected to the output bus, I’d see 5.6VDC there. I was guessing maybe that zener voltage would end up being 5.6VDC less than the rail voltage at full output, but, I’ve never checked.

My understanding of the driver circuit on this is that the "rail" voltage for the drivers floats 5.6V above (for the positive side) the output signal, so with no output signal the TIP41 collector should be at 5.6V for example. This seems like it was meant to prevent the driver transistors from seeing full rail-to-rail voltage swings, but I have two Eden modules that having clipping somewhat similar to yours due to some issue with this mechanism. I haven't gotten around to just replacing all the suspect transistors yet.

[J M Fahey]

I have an Eden WT-800C on the bench, with one of the two power amp modules having a pair of MJL21193/MJL21194 pwr xstrs in place of the three pairs of 2SA1492/2SC3856 pwr xstrs.

You are using a single 16A transistor instead of 3 x 15A ones
You are severely underpowered.
First and foremost add the missing transistors, not only for current purposes but for dissipation and Soar.
I am amazed that single transistor die didn´t explode spreading plastic shrapnell all over amp guts.

Under 4 ohm load, the positive half of the output clips very prematurely, while the negative half gets full output swing. Under no load, or higher load values (8 ohm +), I don’t have the premature clipping.

Proof the amp works, it´s just severely underrated as is.

I swapped out the voltage gain stage xstrs Q7 & Q8, which made no difference. Also changed out the upper driver Q10, which also made no difference.

I measured the voltage across the collector resistors of the input stage R8, R11, finding 700uA current flowing there. Measured the voltage across the emitter resistors R19 & R22 of the Voltage Gain stage, finding 12.5mA flowing thru that stage. Checked the voltage across the emitter resistors of the two drivers Q10 & Q11, finding only 5.3mA flowing there (all these readings under idle, 120VAC AC mains. Power supply rails are about +/-80VDC. I had set the bias per recommendation of 7.5mV across TP A & B.

The Driver circuit in this amp, with the collectors of Q10 & Q11 is fed thru 6.8k 2W resistors off the main supplies, and have 5.6V zener diodes at that junction of the R24/Q10-collector and R25/Q11-collector. I’ve never stopped to understand just how this works, as I’ve never had any issues with it, apart from having the zeners fail short, drivers & outputs fail, emitter resistors in the output stage fail….typical power amp failure mode stuff.

I decided to increase the collector current of the drivers by lowering the emitter resistors R26 & R28 to 47 ohms, now running 11.3mA. That made the clipping level far worse. When I calculated the voltage across those 6.8k collector resistors, the original value with 5.3mA flowing showed 36V across them, and with 11.3mA, now 76.8V. I ran out of time at this point, needing to go catch my bus home. I obviously don’t have a good understanding of this driver circuit, nor have I made comparative measurements against the other amp module in the chassis, it having the 3 pairs of 2SA1492/2SC3856 pwr xstrs. No hint of this clipping behavior.

I’d think at idle, with those 5.6V zeners (D50, D3, both 1n4734) connected to the output bus, I’d see 5.6VDC there. I was guessing maybe that zener voltage would end up being 5.6VDC less than the rail voltage at full output, but, I’ve never checked.

Those are band-aids,go for the realproblem.

Anyone understand this circuit, or have a clue why I’d be seeing such severe positive clipping under high current output loads?

They are bootstrapping TIP41/42 so they can fly beyond rail voltage and fully saturate power transistors.
They gain some 2V peak swing; main gain is not increased power which is minimal but better saturation of main power transistors, they will work colder.

Worst dissipation situation is reaching squarewave clipping and NOT reaching rail voltage, so you have maximum current for longest time and a still significant voltage drop.

Say, 10A through a transistor at 2V saturation is half dissipation as when transistor passes same current but dropping 4V.

I am aware of that since my earliest designs when testing and measuring *everything* to check them (OCD sometimes is a good thing ) and found that on full saturation heatsinks actually **cooled** compared to "normal" operation.

[nevetslab]

You are using a single 16A transistor instead of 3 x 15A ones

I found the typo in the explanation, and corrected that. I'm using 2 pairs of the MJL21193/MJL21194, not one pair. Much like the original version using the two pairs of Sanken 2SA1494/2SC3858. Their PCB solder pads/plate-thru holes self-destruct during careful dissection, so this one module has just the two pair. I'd rather be using 3 pairs of outputs from a pure safety standpoint.

[nevetslab]

Scope waveforms of positive drive problems

After removing the 47 ohm emitter resistors I tried earlier, restoring the 100 ohm values for R26/R28, I went to remove the miller cap C9 to replace it, just in case that was involved. Eden's PCB vendor is low quality, and in spite using well-maintained Pace Deslodering Iron/tips, the plate-thru's come out with the greatest of ease, solder pads pull off, ripping themselves from attached traces, etc. Makes you question your worthiness as a qualified electronics tech/engineer regardless of decades of experience.

Regardless of that POS attribute to their PCB's, I restored the original values, replacing C9 with another 1nF/400V cap and tried again. This time around, I captured some scope photos showing what still has me stumped as to the cause.



The 1st photo, showing the output from 0V thru full power, so you can see the sequence of clean waveform until hitting some limit at around 36VDC at the collector of Q10. The 2nd photo only shows the full power waveform. The 0V reference for the lower waveform is at the second gratical, while the 0 phase of the waveform begins at the 5.6V zener voltage in all of these scope photos. Looking at the peak of the lower trace, you can see the peak waveform, just before hard clip happens at progressively higher output levels. Then, 3rd photo is with 8 ohm load, 4th photo is with 16 ohm load. Open circuit, I don't see waveform clipping...like there may be a vactec hidding out someplace preventing that. The limiter is turned off.

[nevetslab]

Problem found.....solder joints again!

Twice now I both briefly saw full output at 4 ohm load, and heard a faint click, after which the upper half of the output waveform was severely clipped. So, pulled the module back out, and went hunting with the surgical loupes/headlight, re-soldering all of the emitter resistors, power xstr connections where traces were connected....most on the top side of the board, as the bottom side solder pads had one by one failed and came off. Not sure just which connection(s) was the culprit, but, once I powered the amp back up, I had solid output. It's now burning in using sine-random to drive the output level from low level thru hard clipping at 4 ohms.

[J M Fahey]

You might have been losing 1 of 2 power transistors when PCB heated up.

Remaining one was go0od enough for 8 ohm but not 4 ohm.

In fact it was probably triggering the short protection after surpassing its assigned current rating.

Does the amp have two fans?

[nevetslab]

You might have been losing 1 of 2 power transistors when PCB heated up.

Remaining one was go0od enough for 8 ohm but not 4 ohm.

In fact it was probably triggering the short protection after surpassing its assigned current rating.

Does the amp have two fans?

Single fan, not even high velocity from what is available in 80mm fans. Short protection? While I'm used to seeing that in most solid state amps, I haven't found any in the chassis nor in the schematics, unless it's in disguise. Now, I don't have any document that details the left Ch output PCB assy, which has a relay on it, so that may be some sort of protection circuit. Nothing like that in the earlier WT800's. No traditional short circuit protection to shunt off base drive, no fused outputs, only a mains fuse. The small AC switch on the front panel triggers a Triac on the floor of the chassis for turn-on. And, each channel has Fan control thermal switch and in-line thermal switch to open AC if overheating heat sink. That's the protection circuits on the Eden WT800, as far as I've seen.

[teemuk]

I could be off track here, but I think the driver arrangement is not only a bootstrap but a short circuit protection as well: Increased current draw of the finals will 'starve' the driver collector voltage and limit the driver output by clipping. Due to clipping the amplifier does not produce full output signal swing, which effectively limits short circuit current.
As you may note, the design does not incorporate a classic VI limiter. Never does in any implementation.