I have a Cyrus iii (HA7A) on my workbench with crackling noise in the right channel. It looks like this:
The lower trace is the KILLPRE line which goes high (+5 V) when there is DC on the output but that does not happen here. There are cases when it does happen though. I observed this with both channels going silent but only the right channel having spikes:
Note that this image is very stretched out because of the very long acquitision time (20 minutes). The input signal is 500 Hz, the aliasing prevents showing the correct waveform of the sine wave but does show the spikes.
There seem to be SMDs that can be suspect but before I start poking around attaching my probe to possibly suspect transistors, are there any known suspects? Right now I am at the start of a lengthy sleuthing effort so anything to cut to the chase and replace the bad transistor.
The lower trace is the KILLPRE line which goes high (+5 V) when there is DC on the output but that does not happen here. There are cases when it does happen though. I observed this with both channels going silent but only the right channel having spikes:
Note that this image is very stretched out because of the very long acquitision time (20 minutes). The input signal is 500 Hz, the aliasing prevents showing the correct waveform of the sine wave but does show the spikes.
There seem to be SMDs that can be suspect but before I start poking around attaching my probe to possibly suspect transistors, are there any known suspects? Right now I am at the start of a lengthy sleuthing effort so anything to cut to the chase and replace the bad transistor.
Definitely. I have it produce 2x 1 W and after an hour or so it will misbehave. The low power heats up the amp nicely to some 60°C. Right now I'm monitoring the emitter of T504 (in the manual wrongly referenced as T503) but I think I can save myself the trouble and just replace all four input stage transistors (BC846/56) and then put it on soak test again. I've had amps before with bad transistors of this type.
True. But to find an intermittent fault (kraakfout, if you will) a scope beats a DMM:
Top trace: output and trigger source. Any crackling will trigger the scope. This is just free-running.
Bottom trace: emitter of T504. You can see the tiniest bit of noise on the lower edge of the signal.
These faults are a bitch to fix. I tossed the amp in a corner and will look at it later. Or much later. Right now I'm watching a stupid movie (Space Cadet. Good grief...)
My point is that nobody here seems to think bout fault finding, by systematically tracing the fault back from output to where it occurs and fix it.
Instead people suggest random changes (replace all transistors) or just throw the unit aside expecting it to fix itself spontaneously?
As Ed Dell would say, "mute evidence of inability to confront the issue".
You say: "these faults are a bitch to fix". Well, you don't know that, because you don't know what the actual fault is, you only know how it manifests itselfs.
Jan
Instead people suggest random changes (replace all transistors) or just throw the unit aside expecting it to fix itself spontaneously?
As Ed Dell would say, "mute evidence of inability to confront the issue".
You say: "these faults are a bitch to fix". Well, you don't know that, because you don't know what the actual fault is, you only know how it manifests itselfs.
Jan
Last edited:
Well, that is exactly the reason why these faults are so hard to find. I left the Cyrus on the whole frikkin' day and I got nowhere. The alternative is to replace all suspect transistors but that is a last resort only. There are about ten SOT-23 SMD transistors in the right channel. Replacing those will take maybe half an hour. I am not a fan of blanket replacements but that certainly beats having it on soak test for a day.
You don't get the point. Why do you think leaving it on a whole freakin' day gets you anywhere??
You mentioned the option of replacing all transistors in a stage.
I suggested measuring the DC conditions on those transistors to see if there are any broken. Doesn't that sound like a logical step?
Scoping the output going towards the input to see where the 'crackling' starts, doesn't that seem like a logical step?
Does putting it on a shelf and watching a move sound like a more logical step to find and fix the problem?
Don't be afraid of the beast. There is a solution and you will find it.
If you are willing to look for it.
Jan
You mentioned the option of replacing all transistors in a stage.
I suggested measuring the DC conditions on those transistors to see if there are any broken. Doesn't that sound like a logical step?
Scoping the output going towards the input to see where the 'crackling' starts, doesn't that seem like a logical step?
Does putting it on a shelf and watching a move sound like a more logical step to find and fix the problem?
Don't be afraid of the beast. There is a solution and you will find it.
If you are willing to look for it.
Jan
Maybe this will help:
https://www.diyaudio.com/community/threads/a-manifesto-for-audio-diy-ers.298272/
Jan
https://www.diyaudio.com/community/threads/a-manifesto-for-audio-diy-ers.298272/
Jan
No, it doesn't. That is the problem with intermittent faults. Still, it may be an option. If I measure the V(be) of all suspect (SOT-23) transistors, and there is one outlier, it is likely to be the culprit. I don't expect miracles but it sure beats having it on soak test for a day without the issue manifesting itself. If the crackling was present all the time this thread wouldn't even have existed. And the Cyrus would have been happily sitting with its owner.
I'm no expert but:
Can you provoke the fault with targeted heat/cold or tapping/prodding components?
Could it be a dry joint?
Can you provoke the fault with targeted heat/cold or tapping/prodding components?
Could it be a dry joint?
Like I said get the IR camera on it, if that shows no culprits then it's time for the freeze spray, a bit of component tapping, then get the probes out.
If just fixed a Paradise that was crackling, negative rail had a cap with a dry joint to ground. I followed that exact procedure.
If just fixed a Paradise that was crackling, negative rail had a cap with a dry joint to ground. I followed that exact procedure.
Tapping does exactly nothing. Hot/cold is a bit tricky with SMDs. Freezer spray can has gone phut and my Weller is grounded so I cannot apply the tip to the collector pin of an SMD transistor without isolating it first. Right now it is off my workbench having been replaced by another repair.
Finally mustered the courage to delve into this amp again. No results whatsoever. The only variations in V(be) due to temperature rise. Even touching the SMDs with my trusty Weller (ungrounded for the occasion) did not provoke the fault. I refuse to do a blanket replacement of SMDs only to find out it did not fix the issue. If it gets worse the owner knows how to find me.
Epilogue. The owner did know how to find me and returned the amp. Having found a solution to screendump my Siglent scope every 14 seconds to my Linux laptop I eventually observed the issue again. And again I could not pinpoint the fault to a specific transistor. The freezer spray I purchased was fake and I had to clean waxy goo off the PCB. I finally resorted to do a blanket replacement after all for both channels. 12 SMD transistors. I used BC846B/56B, marking 1BW and 3BW. It's now under soak test and it behaves:
This is both channels driven with a square wave so I get a nice fat red line.
So what did I learn? Just don't shy away from blanket replacements. Saves a lot of time and frustration. Who cares that I replaced 12 transistors with a single bad one between them. And now I can screendump my scope screen to my laptop. Beats the flaky USB interface on the Sigslet.
This is both channels driven with a square wave so I get a nice fat red line.
So what did I learn? Just don't shy away from blanket replacements. Saves a lot of time and frustration. Who cares that I replaced 12 transistors with a single bad one between them. And now I can screendump my scope screen to my laptop. Beats the flaky USB interface on the Sigslet.