Hum to noise problem with class A chipamp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have built a chipamp (+20dB) based on the LM4562 driving a TIP142 single-ended on a constant current sink made with another TIP-142. The LM4562 operates in class A as it is always sourcing. The thing is biased at Vbe/0.5Ohm (1.3-1.4 A) from 13V rails (10Vrms - 0.7V) to get something arround 6-8W of power. The programming transistor of the ccs is far away from the output transistors (which are on different and big heatsinks) and it is the best sounding chipamp i have ever heard.

The problem i get is that the amplifier hums at start, after a couple of minutes the hum fades and it becomes noisy. Then the noise becomes louder and the radio starts to play from the loudspeakers at an annoying level. The amplifier has to be cooled for a while to return to the "humming" stage.

EDIT: I forgot to say that the radio sounds "extremely clipped but very audible" and that the amplifier sometimes seems to clip at quiet passages while it can sound unclipped "as loud as i feel enjoyable to listen to music"

Any idea about how something like this can happen?
 
ionomolo,
I would first check my solder joints. Sometimes a cold solder joint isn't visible, but can cause some strange problems. It might be best to re-solder all components. If that doesn't solve the problem, try to get a can of "freeze-spray" and after the amp warms up and the problem rears itself, try cooling each component to see if you can isolate the malfunction. If you can't find the problem after that, one of the more-knowledgeable-than-me people will have to help.
I hope this helps you solve the problem.
Mike
 
I still have some problems with that amp... While the hum/noise is gone, it has a curious "popcorn" sound that is very annoying.

The amp is really impressive, i love the way notes fade to nothing while keeping all the detail, it leaves me hearing everything on the room, but the popcorn appears to spoil the experience.

I have looked at it on the scope and it's very strange. The "out of current clipping" looks ugly, but it's what i expect, as the voltage on an inductive load drops when the current is constant, so the negative clipped region has a dip on the middle (the other side looks as common clipping), but while when clipping it also "popcorns", what i see on the scope is the wave moving back an forth in a weird manner.

Any idea about what is causing that problem? :'(

Another question, is there any problem in placing a low-bandwith darlington inside the feedback loop of a high-speed opamp?
 
I have been working arround that amp and, while the 700 KHz (this closely matches 1/(2*pi*Rz*Cz) ) oscillations went away when once the zobel network was removed, now i'm having problems at 3.75 MHz. These can be relatively small (0.5 to 0.2 V) but aren't something good to have in an amp, and are the most probable cause of the "popcorn" sound it occasionaly does. I've been claiming that it sounds really well. It does, but it has that superimposed effect that is present only sometimes and that while allowing to see how good it sounds (it does not happen with all songs) it's very annoying.

My idea is that there is either a layout problem or the LM4562 is going mad due to its output not responding as fast as it does.

The only solution i'm thinking about is replacing the output devices with mosfets. This would ruin the power output and add a bunch of distortion but the +90º phase shift will occur at some 20 MHz (assuming 0.2 nF gate to drain capacitance and Cgs completely bootstraped).

The other option seems to be compensating the LM4562 to make it become a uA741, but i'm not going to do this.

Anyways, 0.2V seems small enough to be solved by some other mean. Any idea?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.