G8700DB driver board related DC offset problem

This G8700DB is driving me nuts. I got this thing in with both channels blown, one half blown, the other had all four outputs wrecked in spectacular fashion such that the mains fuse blew.

One channel and drive board have been rebuilt. It works fine with about 1.7mV of offset.

The other has also been rebuilt with all new semiconductors that have been verified as good but I cannot get rid of the offset. The offset adjustment provides a range 15-24V and the pot is new and tested. There appears to be no bias with 0mV between the emitter of TR2 and TR3 on the OPS board. The bias pot has no effect. Signal integrity is good otherwise. I get clean signals at the output, save for 15-24V of offset.

TR6 runs considerably warmer than TR5 on the drive board.

Outputs are good. The channel works with a known good drive board.

The drive board with the offset had a bad 2SK97. One FET was fine, the other was totally dead. This was replaced with a couple of reasonably matched OnSemi J110s for testing. My measured transconductances are 15.4 and 15.6 mS so they're pretty close. Do they need to be closer, is it that sensitive?

Fusible resistors have been checked and replaced as needed.

15 to 25 volts? or milli volts? If it's volts, something is driving the pre-drivers. Check for open resistors on the driver board for that channel.

- Pete

A 'short' problem (or low resistance) on the +ve side or an 'open' problem (or high resistance) on the -ve side of the driver assembly.

Hyperion said:

A 'short' problem (or low resistance) on the +ve side or an 'open' problem (or high resistance) on the -ve side of the driver assembly.

Click to expand...

The question is *where*? It's hiding in plain sight and driving me crazy!

The drivers are indeed getting signal. I have placed the oscilloscope probe between chassis and the output (before the protection of course!) and the signal out is very clean save for the dramatic offset.

How clean? 0.035% THD at 1kHz as measured by my Keitley 2015. In short, it would be fixed if not for this offset.

I agree you have done the right things, and those readings are puzzling considering the offset. Have you considered the possibility of the voltage not coming from the driver board output, but from somewhere else? - the TR5/6 heating issue seems to be a huge clue, what about a wrong value resistor? or possibly a capacitor back to front?

I would make comparative voltage readings around TR05 & TR06 for each channel - and even if nothing pops up (after checking all resistors) - change them both (in the defective channel), then see what happens.

There is negligible voltage at my test point without the bad driver board installed. If I install the good one from the other channel, operation is well within specification on all parameters we care about. (bias, offset, THD, etc)

Caps have been checked. They're in the right way round.

To answer LBpete: It's 15-25 or so volts. Massive but not full rail offset. As stated earlier, the drivers are being driven as I get output on the scope.

Game plan right now: Change ALL fusible resistors regardless of test result. I've heard suggestions that they can fail in such a manner as to present a DMM with the proper resistance but fail under load. I've never seen that in the shop but I don't deal with many of these G series Sansui receivers so maybe it's a thing with these. (I mainly deal with the old ones like the 2000A, different animals entirely.)

How about a cracked or poor solder joint at pins 11,12,13,14 of the defective driver board?

Get a powerful (watchmakers?) magnifying glass on those joints - then all the others if you find nothing.

All fusible resistors swapped, Q5, Q6 changed. No change in malfunction.

It may be time to hunt for a cracked trace.

Edit: All pins on the board to main unit interface have been resoldered. That was one of my first steps. We have Vcc and Vee.

Second edit: The G8700 has an error in the markings for the FET pinout. I was made ware of it by finding the 2SK97 datasheet and tracing the board to confirm. So I am aware of it and that's not the issue either unless when I got gate and source mixed up at the first it damaged something downstream. I'm going to check that now. All the big semiconductors are good and new. Open loop between heatsink and tab on both drivers. Tab is solidly connected to the board via the screw.

This isn't the first error I've found, that would be C616 on the main power supply board. Found that one by blowing fuses and ending up with a warm to the touch cap and a non powered tuner.

I had a G9000 which had a similar problem, which turned out to be a cracked trace on the driver board....
It was completely invisible to the naked eye....

it turned out to be the earth reference on the input molex connector on the top of the driver board, one of the solder pads had cracked and disconnected itself from the trace on the board...
So when it lost its earth, the DC offset would slowly rise up to about 60V......

So I would be looking at every connection and even using a plastic implement to push on a few things to see if you can spot a problem...
Time to get out the 5X Optivisor......:thmbsp:

kevzep said:

I had a G9000 which had a similar problem, which turned out to be a cracked trace on the driver board....
It was completely invisible to the naked eye....

it turned out to be the earth reference on the input molex connector on the top of the driver board, one of the solder pads had cracked and disconnected itself from the trace on the board...
So when it lost its earth, the DC offset would slowly rise up to about 60V......

So I would be looking at every connection and even using a plastic implement to push on a few things to see if you can spot a problem...
Time to get out the 5X Optivisor......:thmbsp:

Click to expand...

:dammit:

I've discovered something.

The gate pinch off voltage on the replacement transistors I used is MUCH higher than that of 2SK97. Oops.

So to test this hypothesis, I obtained two 2SK41...not right but closer. Circuit operation is much more in line with expectations now. Offset is still not good at 1.6V but that's certainly better than 16V!

But wait it gets better! I HAVE BIAS!

So its time to hunt down a 2SK97 or something else with a pinch off voltage of <2V and low noise.

hardwareguy said:

I've discovered something.

The gate pinch off voltage on the replacement transistors I used is MUCH higher than that of 2SK97. Oops.

So to test this hypothesis, I obtained two 2SK41...not right but closer. Circuit operation is much more in line with expectations now. Offset is still not good at 1.6V but that's certainly better than 16V!

But wait it gets better! I HAVE BIAS!

So its time to hunt down a 2SK97 or something else with a pinch off voltage of <2V and low noise.

Click to expand...

Good work......Looks like you'll get it sorted with the right JFET then.....:thmbsp:

I love living in Dallas! You know why? I found the elusive 2SK97 in a dusty bin at a local electronics shop. (I also found a rather elusive TC9212P that I need to get some rather large late 90s BPC up and running, but that's another job.)

I'll pop it in after I get back from the IMAX tonight. I think I'm gonna save this one! :yes:

2SK97 installed, driver board is operational as predicted.

Bias has stabilized to 4.97mV, offset is 1.2mV max.

This repair is almost complete, just some dial lamps, a meter calibration and a load test to go.

So for reference, in addition to needing the proper voltage, current, polarity and a gain matched pair, you must also check the pinch off voltage of your new FETs if you cannot obtain a 2SK97. The pinch off needs to be under 2V and preferably between 300mV to 1V.

well hunted down...

Well done indeed.

- Pete