Vintage amp (HarmanKardon HK6500) mystery hiss - after full recap

[ExtraThiccBoi]

Hi,
just fixed this amp, changed all caps, some resistors for metal films for lower noise, better therm stability and audiophoolery.
I have a specific strange issue with a hiss that goes up as i turn up the volume potentiometer, but there is no active stage before the volume control.
There is a preamp with tone controls (tied in feedback loop) after the volume control, then it goes to the main amplifier board. I have tested the main amplifier board separately and that has very low noise floor, doesnt put out any of this mystery hiss. The tone control is quite a complicated circuit for me (beginner here). From what i understand, as i turn the volume control up, i raise the line impedance from 0 (shorted to ground, no audio signal) to whatever the value is (the pot is 100k log stereo). This introduces noise to the line. Could this be caused by a noisy transistor on the tone control board? I have couple of those 2SC2320 and 2SA999 transistors with matching beta (salvaged from the phono amplifier of this integrated amp). Anyone got any ideas on what it could be? I don't have a scope around unfortunately
Thanks for any help

[mzacharias]

Does it do this with shorted inputs?

[xwarp]

Both channels?

[shakalnokturn]

I'd answer the above questions first...

If it affects a single channel and still has as much or almost as much hiss with selected input grounded (line not phono input), I'd be going straight to the differential pair on the tone control board.

Sometimes it is also worth putting the cover back on and testing with lights off...

[ExtraThiccBoi]

Does it do this with shorted inputs?

Will test, i have my source connected all the time.

Both channels?

Yep both channels.

I'd answer the above questions first...

If it affects a single channel and still has as much or almost as much hiss with selected input grounded (line not phono input), I'd be going straight to the differential pair on the tone control board.

Sometimes it is also worth putting the cover back on and testing with lights off...

Ill see if it hisses with shorted inputs
But one thing i tried. If i disconnect the volume pot from the tone board input, the hiss gets super intense, that first transistor has 220K and 390K resistors to ground, so when it sees that, it hisses even more.
I have not changed those to metal film nor the 470R ones connecting the differential pair.

[LateLesley]

It could be that you actually introduced the noise just changing components, and using noisy ones, or putting capacitors in the wrong way around. I used to think it wasn't possible, but learned a LOT watching Mr Carlson's Lab on youtube. He's very methodical, and almost OCD about things, but he makes you think about stuff you''d normally not think about. So I think he's worth a watch, you can learn a lot from him.

https://youtu.be/uGXQc-zanWg?t=6310

[ExtraThiccBoi]

It could be that you actually introduced the noise just changing components, and using noisy ones, or putting capacitors in the wrong way around. I used to think it wasn't possible, but learned a LOT watching Mr Carlson's Lab on youtube. He's very methodical, and almost OCD about things, but he makes you think about stuff you''d normally not think about. So I think he's worth a watch, you can learn a lot from him.

https://youtu.be/uGXQc-zanWg?t=6310

I specifically watched the capacitor video and also the super probe, but i don't have a scope.
I made sure to mark foil ends on all film caps i replaced, i used a headphone amplifier to see when the hum is stronger.
I watch his channel, i like his OCD approach   
I can tell the hiss is a little small bit quieter (but still annoying) than before changing the film caps and carbon resistors.

[LateLesley]

Yeah, I like his OCD approach too, you learn so much from it.

What I was going to suggest, is why not get one of those cheap scope kits, DSO138?? It's not a real scope, it's only single channel, and compared to a real scope, its a toy. But compared to NO scope, it's surprisingly capable. It'll do audio stuff at least, and an inexpensive way to get some form of visual of what is going on. Is it accurate? No. But it is enough to give you an idea of stuff going on. I'm basically saying, if you have no scope, it's a cheap way into having something,and it isn't too big.

I got one of the kit DSO138's for about £15 I think it was, complete with a case, and actually had a lot of fun putting it together, and it gave me something to see signals.

I know the feeling of not having a scope. I just scored a cheap one last night on our favourite auction site, but as a for parts and not working. It's from a house clearance, VERY dirty, and apparently no power. I thought it would be a good project one to strip apart, clean everything, and change caps while its in bits, and test all the circuitry. I'm not expecting to have it running quickly, and be learning LOTS here testing and troubleshooting it, especially safe ways to check the high voltage parts. (I think from the service manual that there is almost 2KV (!!!) in parts of it, excluding the HT of course, which I know to be wary around.)

Anyway, I hope you find the source of your noise, and can quieten it down. Good luck and happy hunting.

[magic]

The noise is obviously Johnson noise of the Rs or current noise of the Qs producing voltage across the Rs. I'm not really familiar with audio amps, particularly vintage, but the volume pot and the input resistors do seem rather high value to me, compared to the stuff I have seen in newer audio designs.

Given that both channels are identical and the noise is probably insignificant compared to normal signal (isn't it?), I would guess the amp came that way from the factory.
I doubt that swapping the Qs for identical ones harvested from the same unit is going to change much.

I would start with disconnecting the volume pot and checking just how low resistance is needed from preamp input to ground to make it shut up. Then multiply that by 4 and see if it produces a reasonable value for a replacement volume pot. If so, problem solved.

Also, use this as an opportunity to compare carbon and metal film and see that they probably don't make much practical difference.

[ExtraThiccBoi]

The noise is obviously Johnson noise of the Rs or current noise of the Qs producing voltage across the Rs. I'm not really familiar with audio amps, particularly vintage, but the volume pot and the input resistors do seem rather high value to me, compared to the stuff I have seen in newer audio designs.

Given that both channels are identical and the noise is probably insignificant compared to normal signal (isn't it?), I would guess the amp came that way from the factory.
I doubt that swapping the Qs for identical ones harvested from the same unit is going to change much.

I would start with disconnecting the volume pot and checking just how low resistance is needed from preamp input to ground to make it shut up. Then multiply that by 4 and see if it produces a reasonable value for a replacement volume pot. If so, problem solved.

Also, use this as an opportunity to compare carbon and metal film and see that they probably don't make much practical difference.

By resistance to "make it shut up" you mean the max resistance producing tolerable noise floor?
As you said, if i feed it from my DAC which is quite a high output source, i have to move the knob just the slightest bit to get enough volume, so the noise is very low. But if i was to feed it from something like a phono preamp or another low output device, i would most likely go into annoying noise range.
I have the amp apart again, so i'm going off my memory here, i have one mm probe on the preamp gnd, one on the left channel, measuring the pot resistance as i turn the knob.
At the positions when i get loud enough volume, it's between 6 and 18K.
I think i could go with a 50K pot, those alps direct fit in replacements cost eyewatering 10 euros a piece though.
There are 4.7K resistors in parallel to gnd with the volume pot, i don't know if i would have to change their value as well.
Or maybe changing just the input resistors on the base of Q501 & Q502 would help?
The preamp sees 67K as max resistance when the pot is all the way up.

[magic]

By resistance to "make it shut up" you mean the max resistance producing tolerable noise floor?

Exactly.

I think i could go with a 50K pot, those alps direct fit in replacements cost eyewatering 10 euros a piece though.

50k would only give 6dB noise reduction I think, possibly not enough.

There are 4.7K resistors in parallel to gnd with the volume pot, i don't know if i would have to change their value as well.

I have no idea what they are. How are they connected to the pot?

The preamp sees 67K as max resistance when the pot is all the way up.

Not really, because the top of the pot is zero ohm again, although at a different voltage (whatever the source outputs). The worst case is 6dB below max where both halves of the pot are 50k. Try it - short the inputs to ground or connect some source which plays silence and see which volume setting makes the worst noise.

Or maybe changing just the input resistors on the base of Q501 & Q502 would help?

Surely, but not without side effects.
They form a divider with the output impedance of the pot which is two 50k halves in parallel worst-case, i.e. 25k. A tiny bit less because there is also R505 in parallel with it on the preamp board. So if you make R507 25k, signal at this volume position will be reduced by half and then it will skyrocket as the pot is turned to 100% and its output impedance drops back to zero. I hope you can see how it happens.
Also, they form a high-pass filter with C507 so C507 might need to be adjusted to retain the same cutoff frequency.
But you can try it, connect something in parallel with those Rs already there and see what happens and how bad it is.

[xwarp]

You know, the problem I have with this post about this amp in question is that you started it off with "just fixed this amp, changed all caps, some resistors for metal films for lower noise, better therm stability and audiophoolery."

You never stated the fault that you fixed and then you claim to be a "beginner" without an oscilloscope.

In my opinion, you aren't doing yourself any favors by trying to fix a problem that you could have very well introduced yourself.

Not trying to be a jerk here, but when I work over vintage audio equipment, I diagnose the failure, fix it, then proceed to correct marginal issues that affect the base performance of the unit before doing any mods.

[ExtraThiccBoi]

You know, the problem I have with this post about this amp in question is that you started it off with "just fixed this amp, changed all caps, some resistors for metal films for lower noise, better therm stability and audiophoolery."

You never stated the fault that you fixed and then you claim to be a "beginner" without an oscilloscope.

In my opinion, you aren't doing yourself any favors by trying to fix a problem that you could have very well introduced yourself.

Not trying to be a jerk here, but when I work over vintage audio equipment, I diagnose the failure, fix it, then proceed to correct marginal issues that affect the base performance of the unit before doing any mods.

When first got it, the amp was messed up (bad ouput Qs in Rch) and Rch idling current couldnt be set to spec. So far i have fixed all of that, didnt bother to mention that since it doesnt have any relation to this problem. I had these noise issues before, they were even worse before replacing Cs and Rs. The resistors, well, i was recommended to changed them to metal films by a friend who does vintage restorations.
I hoped the things above would get rid of the hiss, but as it turns out, it's probably a design flaw. I did not know that because as a beginner i don't know a whole lot about these circuits. I started this project to learn about this stuff originally

[elecdonia]

When first got it, the amp was messed up (bad ouput Qs in Rch) and Rch idling current couldnt be set to spec.

I have often discovered damaged input transistors in audio power amplifiers where the output stage transistors had previously failed. By "input transistors" I am referring to the differential pair of small transistors located at the input side of the power amplifier stage on the schematic.  Sometimes this pair of differential input transistors are packaged together in an IC chip type case, or they may be in the form of two separate transistors which are usually mounted very close to each other.

Excessive noise is one symptom of a damaged differential input transistor pair.  It may be heard as a steady white noise or as "popcorn" noise which occasionally stops and starts. For the common case of a stereo 2-channel amplifier repair where only one channel failed and the other still worked, there will be a considerable difference in noise level between the "repaired" channel and the other channel that didn't need repair because it was initially still working.  The "repaired channel" will be noisier.

Another common symptom  of a damaged differential input pair is excessive DC output offset voltage at the speaker terminals. Some amplifiers contain offset adjustment trimpots. Others do not. In every case there should be less than +/- 50 mV of DC offset at the speaker terminal. If the amplifier has an offset adjustment trimpot then it should be possible to reduce the DC offset < +/- 10 mV without needing to move the trimpot very far from its middle.

When replacing the input differential pair, always replace BOTH transistors at the same time. If the PC board layout has 2 separate transistors, it can be useful to position the replacement transistors so that they touch each other, and perhaps add a tiny lump of white heatsink grease bewteen the 2 transistors. Keeping both input transistors at the same temperature reduces output offset voltage drift caused by temperature changes.

In some Japanese amplifiers, the tone control circuits are integrated with the power amplifier stage. Basically they treat the power amplifier as a "large op-amp."  The architecture in this case has all the line-level RCA inputs switched into the volume control/balance control section, and this then goes straight into the power amplifier section without having any other "active" circuitry.

Other amplifier designs contain an active gain stage in the '"tone control" section. Often the architecture resembles a "discrete op-amp" containing several transistors. There might be a "differential pair" of transistors in the input of the tone control amplifier.  Other (older) circuit designs may have only 2 transistors in the tone control stage. The first transistor after the volume control (e.g. the input transistor) is the one to check in case of excessive noise, hiss, or popcorn noise. The best way to ID a "noisy" transistor is by replacing it. And if the input state for the tone control amplifier section is a differential pair of transistors, then they should both be replaced.

[elecdonia]

The input transistors are Q501 and Q503. Both are 2SC2320. I recommend replacing both transistors on the "noisy" channel.



This circuit is a "discrete op-amp" with a differential input stage. This is a very sophisticated tone control amplifier circuit design.  The circuit design resembles high-performance discrete op-amps used in recording studio mixing desks since the early 1970's, such as the legendary API 2520 modular op-amp.

Is your tone control stage passing the audio signal OK (except for the noise) ?
If so, the cause of the noise is likely to be Q501 or Q503 (the input transistors)

The following voltage checks will confirm that all other parts of the tone control stage are working correctly:

DC voltage measured from ground (earth) to the junction between Q515 and Q517 collectors should be less than +/- 50 mV DC.
Differential voltage between emitter and base of Q515 should be in the range of 500-650 mV.
A similar differential voltage should be observed between E and B of Q517. 500-650 mV.  Note: Be very cautious to avoid any short-circuits when making these tests because E and B of Q515/Q517 are at 20V or -20V compared to ground (earth).
Finally, the DC supply voltage rails supplying power to the tone control section should be +20 and -20 V, within a tolerance of +/- 1.5 volts. Also these power supply rails should be pure DC, no noise or AC ripple.



I have one additional recommendation:
Because this tone control stage is supplied with +20V and -20V, the DC voltages at both audio input and audio output are very close to 0V.  There is no DC voltage present across the audio coupling capacitors, C507 at the input and C519 at the output. For best audio quality, it is best to use "non-polarized" aka "bi-polar" BP or bp electrolytic capacitors which are specially designed for high quality audio. I have had great results with the Nichicon MUSE. These are packaged in a bright green case. They are available from many vendors.

[schmitt trigger]

Many moons ago I finally repaired an HH Scott integrated receiver, which had a hissing sound, after a leeeeeeeengthy troubleshooting process, by replacing a transistor.

That puzzled me a lot, because on a simple transistor tester, the device tested just fine.

So I saved the faulty device, until I could travel to another city where I had access to a Tektronix curve tracer.

Comparing the transistor, vs. a known good device, immediately showed the problem. At a Vce setting of about half the maximum rating, the traces on a defective device would  start to wiggle. The transistor was going into avalanche!
The amplifier circuit operation would actually bias the transistor around half its rated voltage, so the symptoms matched.

And before someone asks, No, I don't remember the transistor part number.

[magic]

There is no DC voltage present across the audio coupling capacitors, C507 at the input and C519 at the output. For best audio quality, it is best to use "non-polarized" aka "bi-polar" BP or bp electrolytic capacitors which are specially designed for high quality audio.

Nope, you forgot about the base current of Q501 which flows through R507 and, by my quick calculation, drops about 2V across it. Therefore there is -2V on the right side of C507 and that's why it is polarized the way it is.
But you have a point. If the input pairs are broken and, for example, have lousy beta, their base current may increase unreasonably and together with it the current noise. I guess it costs nothing to check voltage across R507/R508 and R525/R526, calculate base and emitter currents and check if beta is within spec. If it isn't, they certainly are bad.

Still, I think tens of kΩ of input impedance may simply be too much for bipolar inputs, no matter how good. In my limited experience, numbers like hundreds or low thousands of ohms are thrown around as recommended input impedance for line level audio inputs.

Related to the above:

DC voltage measured from ground (earth) to the junction between Q515 and Q517 collectors should be less than +/- 50 mV DC.

Not really. Firstly, the diffpair will try to follow the -2V rather than 0V DC offset at Q501 base. Secondly, there is R535 and Q503 base current going through it, which adds another offset, hardly predictable to mV accuracy IMO.

[David Hess]

The noise is obviously Johnson noise of the Rs or current noise of the Qs producing voltage across the Rs.

Yep, the AC current noise into the differential pair multiplies across the source impedance to act as another voltage noise source at the input.  With the input volume control potentiometer at minimum, the input is shorted so this current noise has no effect.

When first got it, the amp was messed up (bad ouput Qs in Rch) and Rch idling current couldnt be set to spec.

I have often discovered damaged input transistors in audio power amplifiers where the output stage transistors had previously failed.

As is the case here, the maximum input differential voltage is about +/-12 volts.  If this is exceeded when the output stage shorts, then this may cause base-emitter breakdown of the input differential pair transistors damaging their hfe which has a direct effect on input current noise.  In applications where this is a possibility, the differential pair is protected.

Note however that the 470 ohm emitter degeneration resistors in series with the emitters means that this was not a particularly low noise design to start with which is usually the case for power amplifiers.

Excessive noise is one symptom of a damaged differential input transistor pair.  It may be heard as a steady white noise or as "popcorn" noise which occasionally stops and starts.

Popcorn noise is a different problem caused by bad semiconductor fabrication or contamination of the transistor.

[ExtraThiccBoi]

snip

Thank you very, very much for such a deep analysis!
Very interesting to know this is more of an advanced design, guess the "all discrete, no ICs could achieve this level of performance" in the manual isn't there just for marketing hotwords 
The noise is equal in both channels.
The differential pair on the main amp board you mentioned (unobtanium 2SC1775) is most likely OK. It is coupled with a heatshrink, but DC still drifts because its a small design flaw of this harmankardon design. I tested the main amplifier board separately (injected audio signal right in the power amp input), the main board is very quiet, there is a little bit of white noise but it's audible only when i put my ear against the speaker and even then, it's very quiet and from my memory equal in both channels. Actually, the channel that didnt blow up has a little bit more noise (hum-buzz) from i suppose being close to AC lines, still very quiet and inaudible from listening position even in death quiet room.
Now the tone control, it's passing audio OK and the EQ works great too, i replaced those film caps in there, i could try matching the foil ends but i won't bother since the factory config wasn't like that too. Also the resistors in the feedback circ. are metal films now.
I can't really check those low voltages since it's very finnicky to measure that, but i do know the power supply voltage should be ok. The raw voltage is 50V, filtered by 12000uF per rail, the stepped down using a zener-transistor (2SB941/2SD1266) regulator with 2 x 330uF per rail and then goes to the tone control with extra 2 x 220uF per rail. I replaced all caps for either Panasonic and Nippon Chemi-con with extra generous voltage ratings.
I will replace those two trannies with matching beta 2320s i have, see if it makes any difference.
I replaced the electrolytics in audio path with Nichicon UFG, i unfortunately wasnt aware of the UES (muse) at that time. Maybe in the future, my local supplier run by a friend has all of them available.
Thanks again for all the analysis and interesting information!

[ExtraThiccBoi]

Many moons ago I finally repaired an HH Scott integrated receiver, which had a hissing sound, after a leeeeeeeengthy troubleshooting process, by replacing a transistor.

That puzzled me a lot, because on a simple transistor tester, the device tested just fine.

So I saved the faulty device, until I could travel to another city where I had access to a Tektronix curve tracer.

Comparing the transistor, vs. a known good device, immediately showed the problem. At a Vce setting of about half the maximum rating, the traces on a defective device would  start to wiggle. The transistor was going into avalanche!
The amplifier circuit operation would actually bias the transistor around half its rated voltage, so the symptoms matched.

And before someone asks, No, I don't remember the transistor part number.

Well, ill grab some of my other 2320s and try it, see what happens, looks like they all come from one batch since they all have the same beta.
I measured all of those from the tone board and they seemed fine but as you said, transistor tester wont probably help me much.

As is the case here, the maximum input differential voltage is about +/-12 volts.  If this is exceeded when the output stage shorts, then this may cause base-emitter breakdown of the input differential pair transistors damaging their hfe which has a direct effect on input current noise.  In applications where this is a possibility, the differential pair is protected.

Note however that the 470 ohm emitter degeneration resistors in series with the emitters means that this was not a particularly low noise design to start with which is usually the case for power amplifiers.

The main output fail was in the actual power amplifier (schematic included). I didnt experience any problems (ear on the speaker - very quiet hiss, not worth concerns) there after replacing them with NOS parts (matched within production tolerance of toshiba). I don't really want to start messing with transistors in the main amp, it's quite a complicated system. It works fine and hasnt released magic smoke yet.
By the way, should i try replacing the 470 ohm resistors (R525/527) with metal film?
 

[David Hess]

Very interesting to know this is more of an advanced design, guess the "all discrete, no ICs could achieve this level of performance" in the manual isn't there just for marketing hotwords 

It is just marketing.  An integrated design would be lower noise because the emitter degeneration resistors which add significant noise would be left out in favor of a better way to implement Gm reduction of the input stage without adding additional noise.

In this design if the resistors were left out, then the tail current to the differential pair would need to be reduced to reduce Gm and maintain stablity.  But with lower tail current, the full power bandwidth (slew rate) would be reduced which is a real problem with power amplifiers.  The degeneration resistors were added so that a higher tail current could be used for a higher slew rate (full power bandwidth) while maintaining closed loop stability.

Application Note A from National Semiconductors discusses the details and math of why the emitter degeneration resistors are included and the alternatives.

[duak]

I 've also encountered small signal bipolar transistors that are noisy and have significant leakage.  Two I'm thinking of were in the input diff stage of a power amp I built in 1981 so they are from before then.  If memory serves they are Motorola MPS8099.  I didn't do any AC tests on the parts by themselves on the bench so I don't know if they are avalanching.  It took about 10 years before the noise manifested.

I understand that the leakage and noise performance of a bipolar transistor can be permanently affected by reverse biasing the B-E junction to breakdown.  I wonder if this is what is happening here.  It's easy enough to overdrive the input of a power amp when it's off if driven from a preamp.

[schmitt trigger]

Application Note A from National Semiconductors discusses the details and math of why the emitter degeneration resistors are included and the alternatives.

Extraordinary paper,......everything you wanted to know about monolithic opamp architecture, but were afraid to ask.

Unfortunately, it does not list the author's name.
Although in the references, the persons listed were the who is who of linear IC design back in the day

[ExtraThiccBoi]

Very interesting to know this is more of an advanced design, guess the "all discrete, no ICs could achieve this level of performance" in the manual isn't there just for marketing hotwords 

It is just marketing.  An integrated design would be lower noise because the emitter degeneration resistors which add significant noise would be left out in favor of a better way to implement Gm reduction of the input stage without adding additional noise.

In this design if the resistors were left out, then the tail current to the differential pair would need to be reduced to reduce Gm and maintain stablity.  But with lower tail current, the full power bandwidth (slew rate) would be reduced which is a real problem with power amplifiers.  The degeneration resistors were added so that a higher tail current could be used for a higher slew rate (full power bandwidth) while maintaining closed loop stability.

Application Note A from National Semiconductors discusses the details and math of why the emitter degeneration resistors are included and the alternatives.

Very interesting! I will go through it though i am afraid it's a little bit too complex for me. Thanks!

[ExtraThiccBoi]

I 've also encountered small signal bipolar transistors that are noisy and have significant leakage.  Two I'm thinking of were in the input diff stage of a power amp I built in 1981 so they are from before then.  If memory serves they are Motorola MPS8099.  I didn't do any AC tests on the parts by themselves on the bench so I don't know if they are avalanching.  It took about 10 years before the noise manifested.

I understand that the leakage and noise performance of a bipolar transistor can be permanently affected by reverse biasing the B-E junction to breakdown.  I wonder if this is what is happening here.  It's easy enough to overdrive the input of a power amp when it's off if driven from a preamp.

Hmm, i have not overdriven the actual power amplifier at all. It's got quite a headroom (70w 0.09% rated, 120w 1% capable).
All of the noise issues are on the preamp/tone control board.
I replaced the differential pairs with my other 2sc2320s and will test it tomorrow, im currently working on a modification that will turn one of the old phono (salvaged the components since it was bad anyway) into a direct input right into the power amplifier, so i can see how much noise is the preamp adding.