LM4780 - Heatsink and isolation

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've built a dual LM4780 amplifier, and I have a quick question before I power it up. I have both LM4780s bolted to the same heatsink. I don't think this is a problem since the datasheet shows that the case is attached to the V- rail and both chips are powered from the same supply. Can anyone let me know if I've caused myself a problem doing it this way? Thanks!
 
It's no problem as long as there is a physical barrier between the heatsinks and the outside world. If your heatsink is exposed and the chassis is grounded (probably) it is very easy to short out the power supply by any random piece of metal falling on the amp (a pen, a paper clip, etc.). In a higher powered amp the voltage difference could be dangerous to people touching the chassis and heatsink, and it is still not a good idea to allow people to come into contact with the 30V (or so) supply in your amp.

If the heatsink is inside the chassis, and you have some screens covering the openings in the chassis, there should be no problem at all. Many commercial amps are made this way.

I_F
 
homer09 said:
It doesnt sound like a good idea to me. But you may be able to get away with it... Everyone i've seen using uninsulated chips ALWAYS isolates chip from heatsink. Your asking for trouble with this kind of setup. :hot:

The LM4780 only comes in the non-insulated version.

If you look at the various thermal impedances there is a lot to be gained by attaching the LM4780 directly to the heatsink. An insulating washer adds a bit to the thermal impedance and if you are going to run the LM4780 hot, every 0.1 C/W that you have to compensate for means a much larger heat sink (or more lineal feet of air per minute) -- (if I had the time I would work out the math.)

My bridged amp has the heat sink attached to a sheet of birch ply via nylon screws -- i tapped the heat sink so as to not worry about the 6-32nd nuts.
 
I have used the plastic from antistatic bags with good success, but I only used them as a temporary solution until my isolator sheets arived from digikey. Just be sure to use silicon grease or some suitable alternative.

As long as you are sure that your heastink has no connection to ground it is safe to attach the LM4780s to it directly, but you had better be sure. I have blown up chips this way.

I have used 2 LM4780s unisolated on a single heatsink and they ran fine, but sound quality was not as good as after I isolated them.
 
The only part of the LM4780 package (besides the pin) that is connected to V- is the metal part on the back. You can use normal metal screws as they will only contact the non-conductive plastic case. I have actually tested this.

The test is simple put the negative probe of your DMM on the plastic part of the case at the ear and put the other lead to ground. You will get 0V. Now move it to the metal part of the back and you will get the negative rail voltage.

So you see, there really is no need for nylon screws or any such thing. Just be sure the metal part of the back is isolated from the HS.

Of course using them will cause no harm either. Its just a false sense of accomplishment as it has no effect.

As for the part number of the isolating material I don't recall right now. It was a 8 x 8 sheet. I think someone mentioned the part number is a prior thread.
 
The reason they left it exposed was to give the implementor a choice. Certainly leaving it unisolated is one of those choices, but sound quality will suffer if you mount more than one chip unisolated per heatsink. The difference is easily measurable. Also there is a good selection of isolating material with extremely low thermal resistance such as mica.

If you could do one chip per sink and keep the sink completely isolated it would be a great solution. I am not how practical that is though.

Do A/B measurements of THD with the chips unisolated and isolated and have your eyes opened. That is exactly what I did. Actually I am pretty much a noob, but my bro-in-law is an EE and he brought over his O scope for tests of 3 different GCs I built(2 stereo LM4780s and stero LM3886). 2 had THD of less than .02% at 1K into 4ohms but the unisolated LM4780 version had THD of more than 3%. Ick... That was my direct experience. We spent 20 minute adding mica isolators and violla THD as good as the other two.

In any case, have fun.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.