White noise and slight crackles in one amp channel

Jump to Latest
#41
In the screenshots attached, the crackles show as high spikes. What can generate that kind of noise? Is it an electrolytic capacitor, like C3 and C4? The suspect transistors are the differential pair, Q5 and Q6. These are shielded by Q20 and Q18 which act as a cascode. The input pair and cascode feed into a current mirror comprised of Q9 and Q7.
 
#44
Replacing is what I meant. 🙂
I still think Q1 and Q4 are most likely to be the cause. Or call them the primary suspects if you like.
This is purely from experience with many Japanese transistors that have this problem.
Just saying...

Hugo
 
#45
edbarx said:
In the screenshots attached, the crackles show as high spikes. What can generate that kind of noise? Is it an electrolytic capacitor, like C3 and C4? The suspect transistors are the differential pair, Q5 and Q6. These are shielded by Q20 and Q18 which act as a cascode. The input pair and cascode feed into a current mirror comprised of Q9 and Q7.
Click to expand...
The culprit may be an intermediate stage--- i.e. the defective device introduces an error voltage, but global feedback attenuates the glitches. The loop gain tends to differentiate the error mechanism.
 
#46
The cascode and current sources have electrolytic capacitors smoothing their base voltages. Not shown on the schematic. A leaky capacitor can affect both of them. The effect on the input stage may cancel out, but the effect on the VAS has nothing to block it. The current source will have the signal from the leaky capacitor superposed on its output.

When the ambient temperture cools off, I will do the testing. I can use a hot air soldering station to heat the capacitors. The hot air is temperature and volume controlled.

Apart from this issue, the amplifier is a pleasure to use.
 
#48
There are two bias voltages which have the potential to inject noise in the signal. These are the current sources bias and the cascode bias. The biases should be DC and stable. Any departure from DC is a signal injection which is unwanted.

My limitation is the oscilloscope sensitivity.
 
#49
A dogy capacitor can be flashed out with a good laboratory power supply. Set the required voltage to power up the amplifier, and then start cranking the current up and monitor the crackle. Hopefully, the culprit capacitor will pop out/explode. Sounds scary but it is not... It is actually harmless... the amplifier will not suffer any damage, but a shorting cap will overheat and explode. I used this technique widely, during the good old tantalum cap epidemic days when they were popular...

The oscilloscope should definitely be used. Place the probe set to 1:1, along the signal path (the first oscilloscope channel). When you reach a potential transistor/diode that is breaking up (or a leaky capacitor...), the slight capacitance from the probe will flush out the problem... the crackle should become louder. Also, as I said previously, you have a good AMP PCB there, so you can compare. The second oscilloscope channel should be monitoring the output; load the output with something like an 8ohm 50W resistor on a heatsink, and a 0.1uF capacitor in parallel. Input 1kHz sinewave, output around 10W. Keep comparing both channels for a clue...

I had a fair amount of small signal diodes that were causing the crackle you explained... they started to break and cause issues. 1N4148 are quite notorious... replace them all. The PN / NP transistor junctions that are starting to fail could do the same... flush them out with an oscilloscope probe... as per above. NOTE: these components (diodes and transistors) would test okay with a diode tester! But, they were starting to fail only just, and only just enough to cause the issue you are experiencing.

You could also see if the increase in tone generator frequency will flush out the issue.

I also had trim pots failing (intermittent wiper contact, especially with multiturn trim pots)

I asked a few questions and provided a few suggestions in post 32... care to post some photos? Are there any trim pots? if yes, where exactly??
 
Last edited:
#50
I have an apt amp that the diff pair front end went out something like yours. Offset voltage went up as well. Very difficult to track down as like others have said, with negative feedback, almost impossible to isolate. For me in the end it was the excessive offset voltage that led me to replace the diff pair. After I pulled them, I noticed beta was down so I replaced them and it fixed both the noise and offset problems. Front end used ksc1845's, so cheap to replace.
 
  • Like
Reactions: pwdiya12
#51
Extreme_Boky said:
I asked a few questions and provided a few suggestions in post 32... care to post some photos? Are there any trim pots? if yes, where exactly??
Click to expand...
There is one multiturn trimpot per amplifier channel. This is used to set the bias. The variable resistor is a kermet 10 turn pot. Presently, I am using one channel only. I will post pictures when I open the amplifier box.
 
Last edited:
#52
Thanks goes to all who replied. I am attaching a picture showing the blue multiturn pot.
 

Attachments

  • kermet001.jpg
    kermet001.jpg
    82.8 KB · Views: 71
#54
I don't understand the physics behind it, but apparently emitter-base avalanche breakthrough causes some sort of hot electron damage, resulting in crystal defects. That increases the non-ideal base current component that causes hFE to drop at low collector currents. The non-ideal base current has a large amount of 1/f noise associated with it. If the transistors were not packaged, you could solve it with thermal annealing.

A colleague of mine wrote an article about it. I'll try to post a reference to it next week.
 
Last edited:
#56
I will use a hot air soldering station with a very narrow air nozzle to aim hot air on transistors selectively. The soldering station is temperature controlled and the lowest temperature is 100C. I am suspecting the issue is caused by either a capacitor or a transistor. Noise signals can be injected into the base of the inverting input by a leaky capacitor. The capacitor is quite large, 2200uF which implies the plate area is also large. A leaky dielectric gives rise to random voltages which are fed into the inverting input.
 
  • Like
Reactions: oggioffshore
#57
#58
edbarx said:
I will use a hot air soldering station with a very narrow air nozzle to aim hot air on transistors selectively. The soldering station is temperature controlled and the lowest temperature is 100C. I am suspecting the issue is caused by either a capacitor or a transistor. Noise signals can be injected into the base of the inverting input by a leaky capacitor. The capacitor is quite large, 2200uF which implies the plate area is also large. A leaky dielectric gives rise to random voltages which are fed into the inverting input.
Click to expand...
As I am experiencing similar problems with my 11-12 year old amp, I am wondering if you ever managed to find the cause of the crackle. Did you, and if so, what was the cause?
 
You must log in or register to reply here.
Top Bottom