LM3875s peak currrent capability

[Dxvideo]

Hi all,

As discussed before, technical facts say (in fact AdrewT says) in real world, speakers may sink higher current than calculated values...
For example,
If I am making a 60W gainclone with LM3875 IC for my Wharfedale Diamond 9.1 speakers (6 ohm) then my calculations will be as follows;
Voltage swing: 6 x 60 = 360 ---> sqrt(360) = 18.97v RMS / 26,74v Peak
Current on load: 18.97 / 6 = 3,16A RMS / 4,47A Peak
So LM3875 seems suitable under that conditions. Its current limit is 5A..
And if we look at SOA side, IC is still safe.. Overture Design Gude says 36W will be dissipated at full power..

However, if we look at from the other side (not from hell!) a 6 ohm labelled inductive load may sink 16A peaks (dependent to signal frequency) then I must ask;
- How can we source this amount of current? From capacitors?
- If we could get this 16A anyway then our IC can source that much current to load?
- If that IC cannot source that much current, is it cheating to label it as "56W audio amplifier"? Audio amplifier means "You will use it with the speakers". Does anyone know a speaker without inductance?

Plase somebody tell me that matter clearly.. Will I need a 16A capable PSU and IC for my 60W amplifier or not?

Thanks in advance..

[peranders]

Measure the DC resistance, then you'll have a grip of max current. If you are worried about current capability, why don't you forget the LM3875 and look at LM3886 where you have nearly 12 A?

[dfdye]

With regards to the PSU, I had a similar question and there is a good discussion of "minimum" PSU design considerations in this thread

[sreten]

However, its impedance plot (fig.1), other than a dip to 3.9 ohms in
the lower midrange and a combination of 5.7 ohms impedance and
–36° capacitive phase angle at 120Hz, indicates that the speaker
should be fairly easy to drive.

Hi,

I'd say worst case loading is equivalent to ~ 4R resistive loading.
The power amplifier peak current should be optimised for a 4R resistive load.

Looking at the LM1875 data sheet it simply is not designed for 4R loads.
See Output Power vs Load Resistance graph.
Looking at the LM1886 data sheet it is designed for 4R loads.
See Output Power vs Load Resistance graph.

However music is not RMS sine waves and has a higher peak to
mean ratio, at full volume ~ 80% of the time is spent below ~ 20%
of full output for reasonably dynamic music, so continuous current
ratings for a power supply are fairly meaningless.

With the LM1886 (and its aggressive current protection) for stereo
anything above ~ 140VA for the transformer would be pointless IMO.
Peak currents the chip allows would be provided by the capacitors.

/sreten.

[tangmonster]

wouldn't the maximum current the chip can deliver be much higher due to the very short duration it needs to deliver it?

wouldn't your power supply caps be able to deliver that surge of high current for those short durations?

[peranders]

Originally posted by tangmonster
wouldn't your power supply caps be able to deliver that surge of high current for those short durations?

The chip has a current limited output.

[tangmonster]

so basically it has an inbuilt circuit that will CUT off output if current goes higher than a certain ampage?

[Dxvideo]

Originally posted by sreten




Hi,

I'd say worst case loading is equivalent to ~ 4R resistive loading.
The power amplifier peak current should be optimised for a 4R resistive load.

Looking at the LM1875 data sheet it simply is not designed for 4R loads.
See Output Power vs Load Resistance graph.
Looking at the LM1886 data sheet it is designed for 4R loads.
See Output Power vs Load Resistance graph.

However music is not RMS sine waves and has a higher peak to
mean ratio, at full volume ~ 80% of the time is spent below ~ 20%
of full output for reasonably dynamic music, so continuous current
ratings for a power supply are fairly meaningless.

With the LM1886 (and its aggressive current protection) for stereo
anything above ~ 140VA for the transformer would be pointless IMO.
Peak currents the chip allows would be provided by the capacitors.

/sreten.

Yesterday evening, I have measured the DC resistance from binding posts and saw 3,5ohms..
Anyway, the graphic says more..
It seems I cannot use LM3875 as you say. LM3886 is better to use, with 33v rails it can produce 100W (into 4R worst conditions!)..
In fact I was giving up hope from chipamps.. I think an amplifier without current limiting is a better solution. Its totally free and can give 10, 20, 30 or more ampere peaks..
So a LM4702 + lateral MOSFET amplifier seems a better solutions I guess. Of course without output protection (may be a DC offset protection will be enough).
What do you say?
By the way, are you sure about the 140VA matter? I was thinking to use two 225VA s for stereo...

[peranders]

Originally posted by tangmonster
so basically it has an inbuilt circuit that will CUT off output if current goes higher than a certain ampage?

I'll guess you don't have read anything about it. National has someting they call SPiKe protection
http://www.national.com/an/AN/AN-898.pdf

[dfdye]

If you really need more current, you can always use multiple chips. There are a ton of bridged/paralleled circuits rolling around the forum and on National's web pages. Page 12 of that app. note is pretty demonstrative of how beefy you can get these chipamps.