[kin0kin]

Ive built one for a friend and for some reasons the left channel has some odd problem. It would make really loud crackling sound after awhile (around 1 hour+/- of use). I've checked the bias and had purposely set it below 30mA (around 28mA) in case it was overheating but that wasnt the problem. I also checked the offset when the crack occurs and there was no offset. I also swapped the servo opamp with a 2227 and it didnt make a difference. What could be the problem? Im wondering if the BD139/140 on that channel maybe bad...

 

[amb]

If it works ok cold and then makes noise after some use, then most likely there is a transistor with a thermal-related problem. There are special purpose cooling sprays used for debugging electronic problems of this nature. Reproduce the problem, then chill each transistor, one at a time, to see if you could get the problem to go away. That would help you isolate the faulty device.

 

[kin0kin]

Does digikey/allied carry the spray? I couldn't find much info about it without knowing the term used for the spray. tia

 

[nysulli]

if you want to pick something up near you, i've used the electronic duster cans, and spraying them while holding them upsidedown

 

[amb]

Some examples of the cooling sprays can be found here:

http://www.mcminone.com/search.asp?k...3830011&MODE=1

You might find something locally in an electronics supplies store, and other online sources.

 

[kin0kin]

any chance of the goo shorting the parts when freezing them? these stuff are not really cheap, I can almsot change a whole set of transistors for the price of a single can

it seems the problem is getting more prominent. it now cracks without needing to warm up. I'm hoping that the bad transistor quickly dies off so that i can start measuring them....they seem to measure fine despite having the problem...

 

[amb]

Quote:

Originally Posted by kin0kin /img/forum/go_quote.gif any chance of the goo shorting the parts when freezing them? these stuff are not really cheap, I can almsot change a whole set of transistors for the price of a single can

The spray is non-conductive and evaporates completely.

Replacing all transistors is brute-force, but certainly an option considering the cost. The downside is more work desoldering and resoldering a bunch of parts, and the possibility of PCB damage.

Quote:

it seems the problem is getting more prominent. it now cracks without needing to warm up. I'm hoping that the bad transistor quickly dies off so that i can start measuring them....they seem to measure fine despite having the problem...

A noisy transistor may not ever measure "bad" as far as DC operating points are concerned.

 

[kin0kin]

it seems that the amp is working fine now. probably a very cranky transistor needed to run-in

 

[zer061zer0]

I also experience similr problem, but somehow i think i might have shorted the left channel.

I am not sure abt the actual situation though as it was my frd who is helping me out with it. But what i have read, i think he lowered the mv to about 4.5. Will that affect the sound in anyway.

 

[amb]

Quote:

Originally Posted by zer061zer0 /img/forum/go_quote.gif I am not sure abt the actual situation though as it was my frd who is helping me out with it. But what i have read, i think he lowered the mv to about 4.5. Will that affect the sound in anyway.

Yes, that's one third the recommended bias, which means that the output stage will easily fall into class AB.

 

[zer061zer0]

Amb,

Can i check with you whats the minimum Mv to reach class A.

 

[amb]

Quote:

Originally Posted by zer061zer0 /img/forum/go_quote.gif Can i check with you whats the minimum Mv to reach class A.

"Class A" is load-dependent. Basically, with a push-pull output topology like the CK²III, the output stage quiescent current must be at least one half the peak output current in order to remain in class A. Let's assume a 32Ω load and you want to deliver 100mW of peak power. Then calculate the peak output current: P = I² * R, (P is power, I is current, R is impedance), solving for I gives I = √(P / R) = 56mA, so if the output stage quiescent current is 28mA it will remain in class A up to that output power level. The default recommendation is 30mA, pretty close to that. When the volume is increased beyond that, then the amp will operate in class AB until the maximum voltage swing is reached (at which point the amp will begin to clip).

The default quiescent current of 30mA causes about 14mV drop on the 0.47Ω output resistors (simple Ohm's Law V = I * R).

The calculation above assumes a resistive load. Real "32Ω" headphones may have an impedance curve that drop below that at certain frequencies, so the current demand would be higher than what we assume.

Using the above example you can calculate what it takes to remain in class A. Obviously, the higher the quiescent current, the more "deeply" into class A the amp runs and less likely to drop out of class A from various load conditions, but then the output transistors will run hotter, requiring heatsinking.

 

[zer061zer0]

thanks amb,

what happened to my board was that it get so hot that it started smoking. in the process its seems that some parts are damage, the transistors and some other parts. and now there is a funny sound on the left channel.

 

[amb]

Btw the support thread for this amp is here, please use that instead of this thread so it would be easier for others with similar questions to find it.

The bias adjustment should be done is small incremental steps, because the output transistors are BJTs which have positive temperature coefficient. This allows the transistors to warm up and settle.