Amp in Parallel Resistors: Wirewound vs Metal Film

[TheLaw117]

Hello,

I'm just about done putting my big old amp project together. I'm a newbie, but thanks to a lot of you, I've got through it well.

For the subwoofer section of the amp, I'm using an LM4780 in parallel mode which will be driving a 4ohm subwoofer. I'm trying parallel to reduce stress on the chip, and because I do not need much output in the first place. (~50W is plenty).

I do not know exactly the name for the high power resistors you use on the outputs to make sure that either channel doesn't "fight" eachother, but that's what I am refering to. I have 0.5Ohm 10W Metal Film Vishays (Sandstone kind)...but I can get much higher wattages in wirewound, which might be better? (I'm not sure).

Any difference. I think the metal films will suffice?

Thanks.

[indianajo]

Some BJT Output Transistor designs use the inductance of the wirewound to help prevent oscillation on capacitative loads. Don't know now prone LM4780 is to oscillation. The Peavey PV1.3k amp I'm working on now has two 5W 1 ohm resistors in parallel on the emiter of each transistor. I don't know why this is better than one .5 ohm 10 W resistor, but I replaced all the burned out ones with the same thing. Some calculation about 360 deg phase shift of the load or something, way too tedious for me to fool with. See the John Curl book, I'm sure it is much more to the point than the math intensive book I tried to learn transistors with forty years ago.

[TheLaw117]

Hmmm. I'm not sure then.

Thanks. Is the John Curl book online? Or do I have to buy it?

Anyone else?

Thanks.

[Michael Bean]

Here's what I use in my system, output balancing resistors to speaker are .15 ohm/3 watt metal oxide. Also note that input and feedback resistors must be matched to at least .1% or better to keep off-set voltages low, I bought 10 of each and measured for best match, absolute value isn't important.

WOOFER AMP.JPG

Mike

[Michael Bean]

I should also add, because I didn't use DC bloking cap for feed back network, that the relative values for all of the precision resistors is important. So in this design, the resitance seen at inverting and non-inverting inputs of chip equal 1k ohm, but agian, absloute value can be whatever you want (within reason), just keep relative values matched and input resistance equal.

Mike

[TheLaw117]

Well I'm new at this so please excuse me. I measured just about every like resistor on the board. I built it with 1% Xicon metal film. Just about all of them seem to be dead on with each other according to my DMM. I only really had one that was askew from the rest.

In your schematic, (I think the values on my board are different, but the location in the circuit should be the same), what is input and what is feedback? Is it the 1K and the 100K?

Thanks. And is the DC blocking cap a big polypropylene film? I have a big fat 2.2uF Wima MKP4. Is that the blocking cap? What does this mean in relation to the resistors?

[Michael Bean]

No problem being a noob, we were all there at first.
On my schematic the input resistor is the 1k, and the feed back network consists of the 1.05k and the 21k resistors. By Norton's theorem, the feedback resistors are fuctionally in parallel, so are equal to a single 1k resistor. Therefore, both inverting and non-inverting inputs see the same resistance.
Here's how it all works. First, all amplifiers using an OpAmp topology will have a DC offset (or non-zero) voltage at the output which is determined by the chips'(or OpAmp built with discrete components) inherent input bias current and voltage, the resistance difference seen at the inverting and non-inverting inputs, and the amount of gain of the amp circuit. The LM4780 (dual version of the LM3886) has an inherent offset of 10 milivolts maximum, and an input bias current of 1 microamp maximum. If the inputs have different resistances at the inputs, there will be different currents into each input, which causes slightly different voltages at each input that is amplified by the gain of the amp. So to keep offset low, resistance seen at both inputs should be the same, or as close as possible. Generally speaking, it's good to keep output offset under 100 milivolts, many believe it should be less. The DC blocking cap I referred to would be between the 1.05k resistor and ground, and creates a high pass filter function which would cause the amp to have unity gain for frequencies below the corner frequency. So to keep the corner frequency as low as possible, either the feedback network has to have a high resistance or the cap has to be really big, that's why I prefer to leave it out of amps that will be driving woofers. And one more thing, you might not get an accurate resistance measurment if the resistors are in-circuit, it's better to lift one lead before measuring.
I hope this helps.

Mike

[TheLaw117]

Thanks Michael. It's all not entirely clear yet, but I'm trying my best.

If the offset voltages are too high (if the resistors are not perfectly matched), what does that do to the amp? I mean...if I just left it as it is...would it damage it?

I'll compare your schematic to mine and see what I can measure.

[Michael Bean]

No damage to the amplifiers per say, but it's best to keep offset as low as possible or you'll have power loss and additional heating of amp chip, and you get excessive power dissipation in the current balancing resistors on the outputs(the .15 ohm ones). For example, lets assume that there is 100 milivolts of positive offset voltage at both amp outputs. In that case, the offsets are equal and cancel out, there won't be any current flow from one amp output to the other, and power dissipation in output resistors is minimal, almost all of the power is delivered to the speaker, but it will still cause power loss and chip heating as I stated before. But if one amp has 100 milivols positive and the other 100 milivols negative for a difference of 200 milivolts, you'll end up dissapating more power in the resistors and the amp outputs, and deliver less power to the speaker, and agian, more power loss and chip heating. But even worse, and this is why I don't use the DC blocking cap for the feedback network, if the caps aren't very well matched, you'll get another source of offset that is much harder to correct. Remember I mentioned needing a high value cap in the feedback network in my pevious post? well a high value cap will usually mean an electrolytic type, and those are notorious for having loose tolerance and value drifting with use and/or aging. Please understand, that none of this is etched in stone, there is no such thing as a "perfect" amp design that works in all cases. That's why it's necessary to figure out what your design goals are, and what compromises(there are always copromises) will have the least impact on the final results. If you want to try it, I recommend checking the amp outputs for for offset before putting in the current balancing resistors. And once everything seems OK, put in the resistors and test with a dummy load, like a high wattage 8 ohm resistor, or if you don't have or can't get a dummy load, use a speaker you're not afraid to sacrifice if something goes wrong.

Mike

[TheLaw117]

Thanks again for dealing with me. Your explanations are outstanding.

I do not think that my board has a DC blocking cap on it. I got it from a Chinese seller with negative documentation, so I'm going by the traces on the board...

As far as I can see, I only have a (de?)coupling cap on the positive of the RCA input.

LM4780 stereo/parallel power amplifier PCB ! | eBay

By reading the title, it advertises parallel operation, so they might have put that into consideration.

Also....again sorry to be a pestilence, but if you look at the eBay page...

I have a few things bugging me. I really wished that I had atleast got a bit of an instruction manual, but of course not.

If you look at the underside of the board..In the picture, at the top of the board, right in the middle. Do you see the two square pads that are pretty close to each other? It seems like the almost want to get connected to get soldered to each other...It seems that THAT is how they want me to make a parallel amplifier, but meanwhile they forget about the whole resistor thing. Err. I can't think of anything else.

In the same manner, if you look at the lower right part of the board, there are two more pads which are very close to each other. They are much smaller, but if they were linked, it would connect 2x 22K resistors together. Going by your schematic, I do not think that would comply with the whole parallel thing...

Thanks to your schematic, I've ruled out a couple of other things I was concerned with, but I'm still lingering on a few unknowns.

So I'm not sure exactly what those pads are doing.

Thank you for your help.