Overheating and voltage issues

[danielwritesbac]

Capacitor of 180p or so, from in+ to in- of the LM3886, does tend to dampen some of the clipping behavior during high power use.
Most would use even more than that; but, I thought it could be useful to mention the minimum.

 

[Hogwild]

AndrewT:

Is this true even if the "measure" function on the scope shows voltages which switch from -ve to +ve at times?

the readings do not swap polarity.
You must be using AC coupling to remove the DC bias.
You are displaying the AC VOLTAGES.

If you turn down the vertical scale so that you can display the DC+AC voltage and then attach the probes you will SEE that the polarities do not swap !

 

[AndrewT]

You need to look at the whole signal.
If you tell your machine to ONLY look at the AC portion of the signal and then ask it a silly question like "what is the DC polarity" of my AC signal, it will give you a silly answer.

 

[Hogwild]

AndrewT:

I think we may be crossing wires here. Doesn't AC coupling display BOTH AC and DC signals? Earlier on, you wrote that I should be using AC coupling.

 

[AndrewT]

AndrewT:

I think we may be crossing wires here. Doesn't AC coupling display BOTH AC and DC signals? Earlier on, you wrote that I should be using AC coupling.

I have looked through my posts of the last month and cannot see what you think you have read.

The scope picture shows the instantaeous voltage.
And holds it on screen long enough that you can see thet last few tens of milliseconds. Then repeats and overlays the new instantaneous readings on the previous to make it look like a continuous display.

It plots the voltage difference between the probe tip and the grounding clip.

If you read the whole voltage and want to see the whole voltage then you adjust the sensitivity to display the voltage limits of your signal.
Set to 10V/cm div and the 8 divisions allow a TOTAL signal of 80Vpp.
If your whole signal is less than that you will see it ALL on screen.
That whole signal could be 60Vdc with a superimposed ripple of 30mVpp
You will see a line at 60Vdc above the Zero volts line which you can set to be "on screen". You will not see the 30mVpp ripple because it's is only 0.03/10 = 0.003 of a division = 0.03mm from peak to peak. It won't even look like a widened line. The 60Vdc line will look flat.

Does that make any sense?

 

[Hogwild]

AndrewT:

Post 114. You wrote that if I used AC coupling, then I would likely have little chance of damaging scope/myself. Some of the post above this one, I'm not sure what message you're trying to get across. Though I thank you for your persistence.

I should've never taken on this project, I guess. It's turned into a nightmare.


Hogwild

 

[AndrewT]

The AC coupling cap on the input is usually rated for over 400Vdc.
I should open mine up and look to see the rating.

If you use AC coupling and you never connect to mains, or similar high voltage, you cannot blow up your scope.

I don't know how they protect the amplifiers after the coupling cap, but they seem to be pretty robust and accept overload without any apparent damage. But just like a scale setting on a DMM, start at the least sensitive and work your way down the range till your trigger starts to detect a small signal.

AndrewT:

I think we may be crossing wires here. Doesn't AC coupling display BOTH AC and DC signals? Earlier on, you wrote that I should be using AC coupling.

Post114 does not say you must use AC coupling.

It tells you what the AC coupling capacitor does by way of protecting the scope input from excessive voltage. And tells you to start with a low sensitivity vertical scale and switch to a higher sensitivity scale after you see you have not overloaded the scope input.

 

[Hogwild]

Andrew:

Your exact words were:
"If you use AC coupling and you never connect to mains, or similar high voltage, you cannot blow up your scope."

I took that to mean that you were suggesting I stick to AC coupling, as we agreed I am a newbie and didn't want to blow up my scope. Did I misunderstand? Was it not a suggestion?

 

[AndrewT]

the scope input has a 3way switch.
You set that to suit what you want to examine.

The middle position ensures no volts across the input and lets you see the "spot" or trace and use the vertical set adjustment to place it on the screen where you want Zero to be. Zero does not have to be exactly in the middle of the screen.

Either side you have DC coupled and AC coupled.
You choose what you want to look at.

 

[Hogwild]

Andrew:

These explanations are much appreciated, but I'm afraid some of it is above my head. I sent you a PM which explains why.

For now, I think I'll just wait til my friend who is more knowledgeable returns to do more testing.

 

[Hogwild]

Okay, my friend came over again and we retested using a homebrew power supply. This consisted of transformer, bridge and reservoir capacitors. There were no bleeder resistors or power supply snubbing.

-A different Toroidal transformer with 8.3V-CT-8.3V output (among others) was used. My friend chose the lower voltage to try to prevent the amp boards from overheating again. That way, we hoped we could test longer.
-A GBPC2504-1 bridge rectifier (Thank you DanielWritesBac)
-Two 10,000 uF capacitors

The wiring was done just the way it would normally be done for, say, a Chipamp.com PSU.

The results were interesting. We checked the homebrewpsu with a meter and scoped it at the transformer output, bridge output and capacitor output. We saw just the expected voltages and waveforms.

We connected amp board B1 to the homebrew PSU. We could not hear any hum. So we connected the input to a CD player, which played classical music. The heatsink ran cold. The sound seemed very good.

We then connected amp board B2 to the homebrew PSU. Not so good. We heard the same intensity and sound quality of hum aswe did when we had the Chipamp.com PSUs connected earlier. We tried to scope noise at speaker output of chipamp pc board but the DSO seemed to be unable to provide a waveform (we are guessing due to DSO sample and resolution).

We will try other scope settings and/or an analog scope with this same goal when we can.

One curious finding...when we measured the two sides of the homebrew PSU with a meter, the voltage was dropping faster on one capacitor than on the other. Is this common/expected?

BTW, in October, I had ordered some of the Homemart.usa amp boards from Ebay discussed in another thread. I thought I would use those to quickly isolate which boards were the cause of the problems. Unfortunately, they never arrived.

Thanks for sticking with us, all of you.

Happy holidays

 

[AndrewT]

We connected amp board B1 to the homebrew PSU. We could not hear any hum. So we connected the input to a CD player, which played classical music. The heatsink ran cold. The sound seemed very good.

We then connected amp board B2 to the homebrew PSU. Not so good. We heard the same intensity and sound quality of hum aswe did when we had the Chipamp.com PSUs connected earlier.

remove the good amplifier and connect the second amplifier alone.
Does this hum, or not?
measure the output voltage both mVdc and mVac.

 

[Hogwild]

Thanks Andrew. Will do, as soon as I get a little time. I'm just finishing up soldering and testing another Chipamp.com PSU board channel. After I've done the test you recommended, I'll test that PSU3 for stability/cleanliness.

This must be the longest thread ever in history for such a "simple" project. I'll never look at lead-free solder again.

 

[Hogwild]

Hi Andrew/Anyone else still following:

We finally got around to some more testing with my helper friend. It turned out that the things
that were wrong were not quite what we'd thought.

First, it took us some fiddling, but we finally realized that one of the oscilloscope probes
was defective. We got bizarre readings and my friend finally narrowed it down to that second probe.
These are junky probes in our opinion, and Instek should have included something better
with the scope. That left us with only one probe for now.

On the two amp boards (B1 and B2), the coaxial cable I had been advised to use for input wires
was creating buzzing/noise. We replaced that with bog standard wire, and the buzzing died down
to a level you can only hear with your ear a few inches from the speaker and that's with everything
still ungrounded. Both amp boards B1 and B2 played music nicely, with less harshness now that I
was using the standard wire.

The first two PSUs I built (PS1 and PS2) tested OK again. The third (PS3) tested even better.

Then, when we used an inlet fuse with a higher (4A) rating, PS1 smoked. Well, one side of it did.
There was a bunch of burning plastic smell, and the terminal block for AC input on one side
of the PSU seemed to be burning. After some brief testing, we believe the cap on that side
is now shorted, and possibly some of the diodes along with it. This part was puzzling, since the board
seemed to be working fine for a while (though we never tested it with a 4A fuse before).
I guess I'll have to build another.

The amp boards soldered by someone else both buzzed heavily. Is that typical when the Input RC
network is not installed? Is the Chipamp.com LM3886 circuit sensitive enough to noise that without
input the RC network you could get that much hum or noise? I'll try to solder in the input RC components
today or tomorrow to see what happens.

So at the moment, it seems that we have at least 2 good amp boards, and two good PSU boards.

Thanks for your sticking with me on this.