what is a chip amp?

[tiltedhalo]

Burst power

I am sorry for using "burst" or PMPO, the term was twisted from its original context years ago to sell BS rated home and car amps to kids and the unwise. I test via a 1Khz signal 5ms and 20ms recovery and could hold this for around 5 minutes before the chips spike kicks in, note that at this power the amps distortion rises greatly to around 5%.
My amplifier uses dual 650VA transformers and 50,000mfd of capacitance to reduce rail sag at saturation

 

[FastEddy]

" .. a 1Khz signal 5ms and 20ms recovery and could hold this for around 5 minutes before the chips spike kicks in ... distortion rises greatly to around 5%. ..."

I would say that this is more of a square wave test signal of around a 25% duty cycle (5ms on, 20ms off), rather than a "burst" ... and it is a good test as far as it goes ... for a motor controller. I would suggest something similar in a sign wave rather than a square wave for music / audio amps ...

Nice looking amp, BTW ... yes, very nice ...

 

[jleaman]

Re: Burst power

I am sorry for using "burst" or PMPO, the term was twisted from its original context years ago to sell BS rated home and car amps to kids and the unwise. I test via a 1Khz signal 5ms and 20ms recovery and could hold this for around 5 minutes before the chips spike kicks in, note that at this power the amps distortion rises greatly to around 5%.
My amplifier uses dual 650VA transformers and 50,000mfd of capacitance to reduce rail sag at saturation

Good Detailed work

 

[cadbury8]

very nice amp. I really like the cap job.

All you guys have like evil skills. Its wild.

anyone mod one with tubes yet? hehehe.

 

[Arx]

" .. a 1Khz signal 5ms and 20ms recovery
I would say that this is more of a square wave test signal of around a 25% duty cycle (5ms on, 20ms off), rather than a "burst"

something similar in a sign wave rather than a square wave for music / audio amps ...

Nice looking amp, BTW ... yes, very nice ...

Ummm.. It's not really a square wave, as he said it's a 1kHz signal. That'll be 5 cycles at 1kHz, then 20 cycles worth of silence.

I think that is probably too short a burst to be useful for much. most anything you're going to hear at very high power in music is going to be a much higher duration.

a single burst test with a longer duration, (I haven't done any calculations, but I'm thinking at least 10 times that) might be useful in that it would tell you how much power the chip can supply while mostly hiding the influence of the heatsink.

of course this says nothing about your amplifier, but it would give an indication of the chip's capabilities given enough cooling, which might be useful in the design stages.

I'm guessing with a 5 cycle burst at 1kHz, you could probably draw some conclusions about the impedance, but not much else. You're effectively dividing down the level of heat the amp must deal with, so you may as well go all the way and just measure the maximum impulse power. anything else will just give you a muddled result somewhere between that value and the true continuous power.

It is a very nice amp BTW.

-Nick

 

[cadbury8]

, but it would give an indication of the chip's capabilities given enough cooling

So Arx are you saying the more heat you remove from the chip and the faster you can remove that heat the harder you can drive the chip?

Is this true?

 

[Arx]

So Arx are you saying the more heat you remove from the chip and the faster you can remove that heat the harder you can drive the chip?

Is this true?

I don't see why not.

My bet is that most of our amps will be limited by heat before anything else. Eventually the package's own thermal resistance will cause a high enough temperature gradient that even with the outside of the chip kept perfectly at room temperature, the inside will reach its thermal limits.

You'll pass the point of diminishing returns long before you get that far. At some point it's going to be much cheaper and easier to just go with a different design, rather than designing the necessary cooling to tweak a little bit more out of a chip which is already running beyond spec.

That said, it sounds like fun, and I'm going to experiment with it a little myself. See how much power I can really get from a single LM3886TF.

-Nick

 

[AndrewT]

So Arx are you saying the more heat you remove from the chip and the faster you can remove that heat the harder you can drive the chip?

Is this true?

I don't see why not.

My bet is that most of our amps will be limited by heat before anything else. Eventually the package's own thermal resistance will cause a high enough temperature gradient that even with the outside of the chip kept perfectly at room temperature, the inside will reach its thermal limits.

You'll pass the point of diminishing returns long before you get that far. At some point it's going to be much cheaper and easier to just go with a different design, rather than designing the necessary cooling to tweak a little bit more out of a chip which is already running beyond spec.

That said, it sounds like fun, and I'm going to experiment with it a little myself. See how much power I can really get from a single LM3886TF.

-Nick

Hi,
most datasheets quote parameters when the chip package is held at 25degC.
The LM3886 from National does exactly the same.
They are in effect using an infinite heatsink held at 25degC (probably watercooled) to generate the specs they quote.
They like all others will be using pulse testing to keep the chip at the lowest operating temp.
Ta=25degC is on most pages unless noted otherwise.
This means that the performance quoted by National (and probably all others) has already been stretched to the theoretical maximum.
Typical performance will be limited by the operating case temperature and National do play fair, they tell you in detail how to ensure adequate cooling.