Chip selection

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Actually I own four LM3875T. After changing pcb, I realized, that one of them produces TIM with a square wave signal 1kHz.

I remarked the problem some time ago and changed many things (gain, pcb, cap's, groundig scheme etc).

So, the problem exists just with one of them, I am sure.

What are your experiences? How big is the failure rate, when you buy 10 chips?

Is it possible, that I destroyed this chip by overheating (soldering) or some electrostatic sensation or some other mistake?

Here is the actual version of the cathode follower implementation. I am very happy with this version now! Next step: regulated PSU for the LM3875...

http://mypage.bluewin.ch/gysiaudiophil/Schemakathodenfolger.gif

Franz

P.S.
No one could tell me just by listening, wich channel has the problem. But on the scope, it looks very nasty.
 
> Actually I own four LM3875T. After changing pcb, I realized, that > one of them produces TIM with a square wave signal 1kHz.

Can you tell us how do you measure TIM?
Or, otherwise, can you specify what do you mean with "TIM"?

IMHO it is not possible to measure TIM with a simple square wave signal and a 'scope.

Please post the picture you get on the 'scope

It is no clear to me if you get TIM with the LM3875T alone or with the cathode follower in front of it.


Regards
 
I will send you a foto, but not today.

It looks like this (and happens just with one chip, independent of the buffer or line stage or without):

Input 1kHz square wave, in the ascending line I remark five spots.

When I tune the scope to a higher frequency, so that I see just the ascending line, you can see five "steps".

Franz
 
> Input 1kHz square wave, in the ascending line I remark five
> spots.

> When I tune the scope to a higher frequency, so that I see just > the ascending line, you can see five "steps".


The circuit seems that is partially oscillating or has a low phase margin. The oscillation is partially damped by the 220pf capacitor between the + and - inputs.

There are several things you can do:

1) The four grounds in the circuit (the two bypass capacitors, the output and the input) are star ground, isn't it?

2) Is the input of the amplifier driven by a low impedance source <= 1kOhm? (due the inverting nature of the amplifier the closed loop gain is source impedance dependent)

3) Put a Zobel network on the output (although National rarely suggest it (see for example the LM3875 datasheet - single supply application)). You can try 2.7 Ohm in series with 100nF

4) Put a 5pF capacitor in parallel with the 270K resistor (roll off frequency about 100KHz)

5) If the load is capacitive add in series with the output an inductance of 0.7 uH in parallel with 10 Ohm as stated in LM3875 datasheet.

Try one step at time and test the result.

All of the above are standard "good practices" in building power amplifier

In my experience it is unlikely to damage a power amplifier handling or soldering it.

Regards
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.