As per https://audiokarma.org/forums/index...pa74-75-2sa958-2sc2168.1050368/#post-16883500 these can be replaced with thermally-conjoined KSC1945 and KSA992 pairs.
Harman Kardon HK770 Thread
- Thread starter Connor G
- Start date
As per https://audiokarma.org/forums/index...pa74-75-2sa958-2sc2168.1050368/#post-16883500 these can be replaced with thermally-conjoined KSC1945 and KSA992 pairs.
@Kevviek you may want to also review the SMD-based dual transistor options and use an adapter board like referred to here: https://audiokarma.org/forums/index...c559-bc560-low-noise-bjt-transistors.1067614/
[relevant snippet]
[relevant snippet]
SMD-based PCB boards with 2SA798 and similar:
@redk9258 and @Warren-Oates have shared that we have AK members contributing to, designing, building and offering PCB boards and fully assembled replacements for original devices like 2sa798 and 2sc992/ksa992 here (in Bartertown Parts forum) and places like eBay. Read further below for details.
2SA798 dual transistor SMD replacement below
- 2SA798 --> Toshiba HN4A06J(TE85L,F)
- 2SA798 --> 2SA1873 on a SOT23 breakout board
- Gerber files for the 2SA798 adapter board and photos in use here: https://audiokarma.org/forums/index...-bjt-transistors.1067614/page-4#post-17547231
Other dual transistor SMD replacements:
- 2Sc1583 --> HN4C06J-BL(TE85L,F (slightly lower max dissipation)
The Plumber
New Member
Quick question: before you consider replacing transistors, have you checked DC levels / offsets in and around the input pairs?
No - but I have a chart with some levels so might have a look. The challenge with this amp is that it’s so compact it’s hard to get to the right places to measure. I bought a ribbon cable and some connectors so that I could take the preamp boards out of the chassis, but haven’t made them up yet.
The Plumber
New Member
Good point - access is not as friendly as the neat schematic might suggest ;-) An extender ribbon is a great idea.
Well this is a bit embarrassing….I built the ribbon extender cable and went about measuring all the voltages at different transistor junctions. They were all generally ok except for one (Q415 collector 1.3v instead of 0.3v), but I was using the HK770 table rather than the SA5050 one and I have not yet looked to see if there’s any circuit differences there that could explain it.
Once I measured everything I went to plug in the speakers, and noticed that the plastic covers on the banana lugs were very cold. Hmm…could these be aluminium and could they be conductive, because with the modified speaker input plugs the surrounds actually touch each other….
As it turns out, the speaker leads *were* shorting against each other because of the covers. Since removing them I have not been able to reproduce the distortion :-(
Measuring across the back gives open circuit but for some reason there is conductivity to the brass thread.
It explains why the protection circuit was kicking in.
The only thing is that I’m sure I tested this with no speakers connected before, when I did the first traces.
The Plumber
New Member
SA5050 collectors of Q413 (-1.2) and Q415 (0.3) look the same as for HK770.
Because you mentioned it, I became curious about the differences myself a while ago. Blue components are HK770, red ones are SA5050:
Some subtle differences and a big difference in feedback from V gain output stage.
Question though: If Q415(c) is 1.3V instead of 0.3V, how do Q421(e) and Q425(c) (and then Q431/433(b) ) look?
Removing the variable introduced by the conductive banana barrels should help diagnosis...
Because you mentioned it, I became curious about the differences myself a while ago. Blue components are HK770, red ones are SA5050:
Some subtle differences and a big difference in feedback from V gain output stage.
Question though: If Q415(c) is 1.3V instead of 0.3V, how do Q421(e) and Q425(c) (and then Q431/433(b) ) look?
Removing the variable introduced by the conductive banana barrels should help diagnosis...
I was unable to reproduce the clipping so I put it back into service, which means it's in a hard to get access to cabinet. I'll come back and measure the other points during the week and let you know what I find. I really should have checked the other channel to compare.
After watching some videos I took, I've confirmed that the clipping was taking place with no speakers connected, so removing the plug covers would help but it wasn't the culprit.
The only other thing I changed is to move the location of D418 (bias compensation diode on a spade connector at the end of two wires) for that channel. Previously I had snapped off the connector in the heatsink so it was temporarily connected to the screw that holds the pre amp board in place. While I had it open this time I made a clip and put it back in the original spot so both channels were the same. I can't think of a reason for this to cause clipping, but that original move is also around the time I started to have the distortion and protection cutout issue.
In mine I've also changed the values of C402 and 404 based on some other posts somewhere.
After watching some videos I took, I've confirmed that the clipping was taking place with no speakers connected, so removing the plug covers would help but it wasn't the culprit.
The only other thing I changed is to move the location of D418 (bias compensation diode on a spade connector at the end of two wires) for that channel. Previously I had snapped off the connector in the heatsink so it was temporarily connected to the screw that holds the pre amp board in place. While I had it open this time I made a clip and put it back in the original spot so both channels were the same. I can't think of a reason for this to cause clipping, but that original move is also around the time I started to have the distortion and protection cutout issue.
In mine I've also changed the values of C402 and 404 based on some other posts somewhere.
The Plumber
New Member
A working channel is a valuable reference that can't be ignored
I think as long as D418 is making good thermal contact with the heat sink, things should be OK. My preamp board does heat up a fair bit too, but I don't know what kind of difference mounting D148 there might make. If the main output transistors aren't being temperature compensated, I'd expect more pronounced crossover distortion, not necessarily clipping. Unless the bias is way out of whack.
And if the bias is way out, that might result in enough DC offset that could trigger the protection circuit. Just guessing, of course.
Sounds reasonable. Stock 100uF seems overkill for input DC isolation
If and when you feel the need to open it back up, please post a picture or two of your ribbon standoff arrangement!
Last edited:
I had an hour spare today so I opened it up again.
I started by checking the DC offset, the left channel was at ~50mV, so I set that back, to zero and then checked around the transistors, using my extender cable to get better access.
There are a few voltage differences, on the whole they are consistent between channels so I assume maybe the manual is not completely correct.
I don’t have a distortion meter so couldn’t do idle current as per the Silver manual, but adjusted by measuring across the emitter resistors.
On the left channel there’s a difference between TP1-3 and TP2-3, one is double the other so I adjusted to 9.5mV/19mV for ~30mV across both (noting the 24mV value in the HK manual gets debated). The right channel I set to 14.5/15.5.
It’s now all back together and sounds great. Heat shrink is over the banana plugs just in case, and I’ll try not to wonder too much about why it suddenly started working. I’m assuming it may have been something mechanical. Time will tell
I started by checking the DC offset, the left channel was at ~50mV, so I set that back, to zero and then checked around the transistors, using my extender cable to get better access.
There are a few voltage differences, on the whole they are consistent between channels so I assume maybe the manual is not completely correct.
I don’t have a distortion meter so couldn’t do idle current as per the Silver manual, but adjusted by measuring across the emitter resistors.
On the left channel there’s a difference between TP1-3 and TP2-3, one is double the other so I adjusted to 9.5mV/19mV for ~30mV across both (noting the 24mV value in the HK manual gets debated). The right channel I set to 14.5/15.5.
It’s now all back together and sounds great. Heat shrink is over the banana plugs just in case, and I’ll try not to wonder too much about why it suddenly started working. I’m assuming it may have been something mechanical. Time will tell
The difference in the mV readings on the different left channel test points does bother me and I'm kicking myself for not doing more measurements.
It would be helpful if someone could explain a bit about how the circuits works, what the impact of the difference might be and where I should look as I probably should prepare to replace some parts when I next open it up. I'm guessing it's somewhere around Q420-426.
It would be helpful if someone could explain a bit about how the circuits works, what the impact of the difference might be and where I should look as I probably should prepare to replace some parts when I next open it up. I'm guessing it's somewhere around Q420-426.
The Plumber
New Member
I can only offer a high-level overview and some simple observations:
I was looking through your measurements near the bias spreader.
- Q406/408 and Q402/404 are the dual differential inputs using the rare and exotic uPA74V/uPA75V differential transistor pairs.
- Q410/412 buffer the input signal and set the bias for:
- Q414/416, who provide the bulk of the voltage gain.
- Q418/D418 and VR404, of course, establish temperature sensitive bias separation for the main output transistors downstream.
- Complementary Feedback pairs Q420/424 and Q422/426 provide current gain to drive the main output transistors with only one junction drop (compared to a Darlington arrangement) and, reputedly, O(10) less variation with temperature.
- Our main output pairs, Q428/430 and Q432/434, of course, are biased away from each other to reduce zero crossing distortion.
I was looking through your measurements near the bias spreader.
- One of your measurements might be off: Q414(c) reads -1.36 (close to spec) while Q420(b) reads +1.45. Those are the same node; I'm guessing the negative value is more correct.
- Q416(c) and Q422(e) are reading about a volt high, but Q426(c), which establishes the high-going main output bias, is pretty close to spec. This may not even have much of an effect, since Q422(c) and Q426(e) voltages are reading nominal, which means they aren't drawing worryingly more or less current than they should.
Thanks, I’ll absorb that with another look at the schematic.
Both 1.36 and 1.45 are correct. They were measured on different days and there’s that much variation. Same for all the readings, they move up and down about 0.1v for the lower readings. For the higher voltages it starts at 45v and can take 2-3 minutes to get down to 40v, some take longer than others to reach a steady state.
I’m turning it off between readings to avoid shorting anything out, and then waiting for things to settle, so it’s quite time consuming. Even with the card on an extender cable there’s not a lot of space.
JimL1
Active Member
Yes there are potentiometers behind the front cover. I would play a mono signal and only adjust one up or down. It looks like a challenge to adjust this unless you have an All plastic screw driver. I would be careful since steel screwdriver will touch something and you will blow it up all for the sake of a small meter adjustment. HK didn't do a good job of designing an easy to adjust meter but it is under the front panel which is a pain to take off be careful because some lights are hooked to it with Short wires.
t