Luxman L-80V DC Offset a bit too high ~150mV

[A-Billi-Tea]

Hi all,

I have a Luxman L-80V late 1970's Stereo Integrated Amplifier.

The amplifier appears to work just fine but it is showing 174mV on the LEFT channel speaker terminals and 158mV on the RIGHT speaker terminals.

This is obviously not ideal! It can be better.

This amplifier has a DC Offset adjustment, but my problem is it isn't working!
If I turn the adjustment all the way clockwise it will shave about 10mV (at BEST) off from the DC voltage but the OP transistors get HOT! So of course, I wasn't going to leave it like for more than a few moments.

Has anyone experienced this problem before, DC Offset adjustment not working? It's quite frustrating, I'm no engineer just a lowly technician lol, so my knowledge on the design of a DC Offset adjustment circuit is limited.

To give extra details, I have repaired this amp from bad solder joints, replaced the trimmer resistors for both bias and DC offset with good multi-turn ones, replaced old VD-1221 diodes with 2 x 1N4148's in series and replaced the tantalums. Someone has worked on this before and replaced all the electrolytic capacitors and they replaced the OP transistors to ON Semi MJ15024/MJ15023 in the LEFT channel only so I replaced the RIGHT channel ones with matching ones for good measure. AFTER all of this is when I proceeded to do the adjustments, bias adjustment worked fine (adjusted to approx 12mV across emitter resistors to get approx 35mA current on the OP transistors) but DC Offset adjustment is no go.

I'm at a bit of a loss as to why this is happening lol.
I'd rather send this back to the owner in 100% working order

I will attach a picture with the circuit around the DC Offset adjustment. The adjustment is VR102 4.7K (towards the centre of the image).

Where should I look first?

[redwire]

Re: Luxman L-80V DC Offset a bit too high ~150mV

The amp might be oscillating, which is the only way I can see offset and bias interacting together. Scope the output and look for squiggles. The replacement transistors might have upset the amp's stability, as MJ15024/MJ15023 are 1/2 the speed of the original parts.

The VD-1221 (D102) is the bias temperature-sense part, so your 1N4148 pair substitute must somehow be in contact with the output transistor's heatsink, to properly track their temperature and maintain the bias level.
Doing this without shorting leads to the heatsink and blowing up things, is very difficult, so I'm wondering how you did it.
Otherwise the bias would not track O/P transistors' temperature and it could go into thermal runaway.

For the offset problem, ensure caps C101, C103, C107 are OK. Both channels are acting similar, so it's likely some mod/upgrade is the problem.

[R_J]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Put the VD-1221 diodes back and see if that fixes it, silicone diodes are not good replacements for temperature compesating varistor diodes
The nte601 is a sub for a VD1121 which has a similar function

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Hi rewire, thanks for the suggestions! Woops I forgot to attach the image in the initial post.

1st of all regarding the replacement transistors, I thought the same as you also but I put the originals back in which were still in the RIGHT channel (what I THOUGHT were the originals), 2SD388A/2SB541A and I got the exact same result. Looking more closely at the schematic now I see that they are meant to be 2SD-188A/2SB-545A does look right to you? I didn't see this earlier because I was only looking at the parts list section and I didn't notice the part listed at the bottom of the schematic.

Regarding the VD-1221 (D102) is definitely not touching the Output transistors, they are far from it! D102 is on the PCB but the Output transistors are chassis mounted away from the PCB. I actually didn't replace those ones, the previous person that touched this replaced D102 with 2x series 1N4148's. The ones which I replaced were D103. I also tried putting the original VD-1221 back in on D103 but same result.

Thanks for the capacitor suggestion, I will check those too.

I think at this stage, now that I think about it more, it is screaming obvious that the incorrect output transistors are causing an issue!
Like you said, because it's in both channels, it is likely a fault that has been caused by technician intervention. Someone worked on this in the past who didn't know what they were doing.

I will post an update when I source proper replacement OP transistors!
If you have a good suggestion for a replacement feel free to chime in
In the meantime, I'm just going to look up the data sheets and see what I can get.

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Put the VD-1221 diodes back and see if that fixes it, silicone diodes are not good replacements for temperature compesating varistor diodes
The nte601 is a sub for a VD1121 which has a similar function

Thanks RJ, I'll try that next if putting in the proper OP Transistors doesn't work!

[R_J]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Just to clarify there are two VD-1221 per side, someone replaced D102 with diodes, what about D103?

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Just to clarify there are two VD-1221 per side, someone replaced D102 with diodes, what about D103?

Yes, I know. What I am saying is that someone replaced both D102's with dual series 1N4148's and I replaced both D103's with dual series 1N4148's.

Perhaps D102 is a problem with the 1N4148's? I have exactly 2 VD-1221's at my disposal so I guess I can try putting the VD-1221's into D102 an D103 on one channel.

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Update: I have tried putting VD-1221's into D102 and D103 of the RIGHT channel but no change!

[redwire]

Re: Luxman L-80V DC Offset a bit too high ~150mV

I think you have two problems.

Out of curiosity, can you (out of circuit) measure a VD-1221's voltage on multimeter diode-test.
Rain City Audio said you can replace them with two 1N4148's and I am sceptical, as his was a VFET amp, not BJT.
Japanese semi manufacturers would not make these special varistor bias diodes to be replaced by $0.04 parts.

D103 is for bias only and cannot affect offset much. It is supposed to monitor heatsink temperature, and looking at other L-80 pics, Luxman did not bother putting the part on any heatsink, so this amp will have bias current all over the place.
There should be small heatsinks on the driver transistors. Some pictures have them, some do not. These would help both the bias and offset.
IMHO I think properly the driver transistors are on a small heatsink with varistor diode glued to it, to track temperature. You can try run much lower bias and see how it behaves. Luxman moved to DC servo in later years, perhaps to correct for this drama.

This amp might have had delicate bias issues and the new parts have upset the applecart. What transistors were changed?

D102 is for the Q105 current source and depending on your V readings (a VD-1221 verses 2 of 1N4148) probably nothing there to worry about.

People replace the electrolytics in vintage power amps, have some other drama around soldering/transistors etc. which damages the new capacitors. Don't assume they are OK.

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Hi redwire,

You say "D103 is for bias only and cannot affect offset much". So, I'm thinking why is VR102 (which is SUPPOSED to be for DC OFFSET according to the service manual) directly in series with D103?

Answer = V102 IS NOT FOR DC OFFSET! The service manual is incorrect, V102 is for IDLE ADJUSTMENT and VR101 is for DC OFFSET! Gees thanks Luxman for trolling me lol.
VR101 did appear to in fact affect the Idle Current, so that mislead me a bit!

So basically, my problem is solved!

I will also check the diode test readings on the VD-1221's vs dual 1N4148's out of curiosity.

As far as heatsinks go, I don't know about the driver transistors which are TO-220 like packages that could easily have heatsinks installed (which I think I might do), but it does look like Q106 is indeed missing its heatsink, you can see from the silkscreen on the PCB that it's clearly meant to have one. The Q106 part number is the correct 2SC945 but it's entirely possible that it's been replaced by someone in the past.

D103 is right next to Q104, do you suggest it might be a good improvement to put a heatsink on Q104 and attach the case of D103 to it (perhaps with a thermal adhesive)?

I'm not clear about what transistors were changed, I can guess that maybe Q106 was. As I mentioned already earlier, as I found it the OP Transistors Q110/Q109 in the LEFT Channel has MJ15024/MJ15023 which are obviously not the originals but the RIGHT channel has 2SD388A/2SB541A which is entirely possible that they are the ones that were installed in the factory, even though it's not the parts specified in the SM. I did find pictures of other L-80V's online that had 2SD388A/2SB541A (I did change the RIGHT channel to match the left with MJ15024/MJ15023 but I put the 2SD388A/2SB541A back in for troubleshooting purposes).

[redwire]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Japanese service manuals in the 1970-80's were full of mistakes so I am used to ignoring the wrong sentences or part number typos etc. The engineer and translator were on different pages.

That makes sense now, the VR102 bias and VR101 DC offset trimpot's functions were mixed up.

I see L-80 pics with small heatsinks for driver transistors Q107, Q108.
L-81 looks the same, but no little heatsinks. Note your transistor tabs are discoloured and oxidized so they do run hot. I would add cheapola eBay TO-220 heatsinks if they can fit and not short to other parts.

A perfect design has the drivers and output transistors on one heatsink with the bias-sense diode bolted to it. That way those transistors are all at the same temperature and the diode can accurately track temperature and keep bias correct for no crossover distortion (bias too low) or meltdown (bias too high).

But most amps are a kluge and here the output transistors, driver transistors, VD-1221 diode are all at different temperatures so bias is going to drift.

If the bias is undercompensating (amp gets warm, bias goes up, amp gets hotter, bias goes up...) and thermal runaway happens, I would consider improving the original design.

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

I guess the D103 VD-1221's are just picking up the temperature of the air inside the unit!!! Doesn't sound like a great design...

It couldn't be too hard to modify it so that the driver transistors are mounted on the OP transistor heatsink (and both of D103 too). The problem is replacing VD-1221, I don't trust them for long-term reliability.

R_J mentioned NTE601 as a replacement, but they're practically just as obsolete now and way over-priced on eBay.

There must be another substitute/replacement? Could I not just find a modern varactor diode with the same basic specs?

[kaboom]

Re: Luxman L-80V DC Offset a bit too high ~150mV

There must be another substitute/replacement? Could I not just find a modern varactor diode with the same basic specs?

DO NOT USE varactor diodes! Those are for RF tuned circuits which have ZERO resemblance to an audio amp.

Two 1N4004s will work for bias temperature compensation.

Change these three electrolytics in BOTH channels:
Input coupling cap
Feedback return cap
DC offset adjust bypass cap

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

DO NOT USE varactor diodes! Those are for RF tuned circuits which have ZERO resemblance to an audio amp.

Two 1N4004s will work for bias temperature compensation.

Change these three electrolytics in BOTH channels:
Input coupling cap
Feedback return cap
DC offset adjust bypass cap

Thanks for the tip. I guess I'm getting my V's mixed up lol. I read somewhere that VD-1221 is a Varactor of some sort but that's obviously wrong info? I cannot find the data sheet for it.
Even "Silicon Varistors" are a pain to find, the only Varistors readily available from my usual go-to suppliers (RS, Farnells, Mouser or Digikey) are MOV's which are obviously not what is wanted!
So, do you think 1N4004's will suit better than 1N4148's?

Are you shure I need to replace those caps that you mention? This amp has had all of it's electrolytic's replaced by someone recently enough. Not my personal choice of electrolytics but they're Panasonic so probably not bad.

[kaboom]

Re: Luxman L-80V DC Offset a bit too high ~150mV

You can use a pair 4148s for D102. There should be approx .6V across R120. Q105 is a constant current source for the voltage amp stage Q103.

The CCS sets the current thru the bias string and driver, which is 5mA (.6/120), so another pair of 4148s is sufficient for D103.

Lift one end of R105, 106, 108, 109 107, 113, and check em. R108/R109 (3.3k) and R107/R113 (47k) should be pretty close, perhaps within 3%. This affects the balance of the input stage, which if disturbed, gives DC offset on the output node.

What caps are in C101 and 107? If leaky, they'll let base current thru, which also upsets the balance of the input stage.

Suitable replacements for Q101/102 are KSA992- if you see corrosion on their leads that appears to enter the body, think about replacing them.
Matched pairs are not needed, you've got the offset pot to correct this.
Don't forget to clean the offset and bias pots while you're in there.

[A-Billi-Tea]

Re: Luxman L-80V DC Offset a bit too high ~150mV

Thanks, kaboom. I might just go for 4148 pairs for both D102 and D103.

I can usually get away with most in-circuit resistance measurements because my multimeter (Fluke 79-II) has pretty low voltage output on resistance mode. I'll lift my legs when I need to
EEV Blog has a good video on this, I don't remember the video number.

C101 is a Tantalum, originally 10μF/10V and I've replaced with a fresh 10μ/16V Tantalum. C107 is an electrolytic 33μ/10V. All of the Aluminium electrolytic caps have been replaced by a previous tech with those metallic blue Panasonic ones, possibly M or AD series. Good chance they're fine.

I don't think Q101/102 need replacing but I might order some KSA992 anyway, I've got buttloads of KSC1845 so why not.

No need to clean the bias and DC offset trim pots, I replaced that crap with better ones