PSU Current Requirements

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ej_whyte

Well-known member
Joined
Nov 12, 2010
Messages
263
Location
Cambridge, UK
Im in the process of selecting a PSU for a build I am doing, but could do with a bit of guidance on the current requirements. I have tried searching but couldn't find any definitive numbers or methods.

The project includes the following opamps:
22x THAT 1200 inputs
12x THAT 1646 outputs
8x 5532's
4x 2520/990c's
2x BUF634

To start with the 1200's. The datasheet quotes supply current as 8mA max, so to add a 25% margin i'm allowing 10mA, however, is this 8mA per rail or for the whole IC? I would guess the whole IC as there is no indication otherwise. The other question is that that figure is with no signal conditions, so under signal conditions would I just calculate the max current from max output voltage and load and use that as a PSU requirement?

The 1646's are much the same story, with the datasheet quoting 5.75mA quiescent current, unloaded. Same questions as above.

Again, 5532's have 16mA total supply current, which is slightly clearer as it is total as opposed to per rail (reinforcing my guess for the previous IC's that the spec is per package). However, the conditions are still noted as Vo = 0 and no load, so the same questions as the previous ICs. I will again apply the 25% comfort margin and bump the 16mA up to 20mA.

The DOA's are am going to allow 100mA just to play on the safe side and keep flexibility open for swapping to more hungry opamps.

The BUF634's I am less clear on. The datasheet puts max quiescent at 20mA, but the seems low to me when I see mention of the 250mA output capabilities. My inexperience shining blindingly here.

Anyway for the meantime, taking the 1200's as 10mA, giving the 1646's a generous margin of 10mA, the 5532's as 20mA, DOA's as 100mA and BUF634's as 25mA that gives me the following:

22x THAT 1200 @ 10mA = 220mA
12x THAT 1646 @ 10mA = 120mA
8x 5532 @20mA = 160mA
4x DOA @ 100mA = 400mA
2x BUF634 @ 25mA = 50mA

Total 950mA

Would a 1A per rail supply suffice bearing in mind that I have used a 25% margin on most opamps and around 74% on the 1646's?

Thanks

Elliott

Datasheets:

1200: http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf
1646: http://www.thatcorp.com/datashts/THAT_1606-1646_Datasheet.pdf
5532: http://www.ti.com/lit/ds/symlink/ne5532.pdf
BUF634: http://www.ti.com/lit/ds/symlink/buf634.pdf
 
> is this 8mA per rail or for the whole IC?

The chip has no ground. The power can only flow from Vcc to Vee (what archaic names!) +/- signal current. For light loads, signal current is negligible. 30V 8mA. No matter that it is divided +15V and -15V.

The dynamic load: Assume 10K pulled to 10V peak. That's another 1mA. Since this is audio wiggling back and forth faster than the rail-caps will sag, it's 0.5mA for square and 0.35mA for sine, even less on unclipped speech/music.

Ah, but the THAT1200 has internal 6K resistor on output. It's all loop-de-loop connected so the actual internal load is unclear, but probably under 10K even before you drive your real load.

1K loading makes 5mA or 3mA or less dynamic current. 500 ohm makes 10mA/6mA, etc.

So yeah, if driving 600 ohms the dynamic current could matter. However the THAT1200 is only rated for loads down to 2K, so 2mA extra should more than cover sine-tests.

Alternatively: model the dynamic current as a rail-rail resistor 6 times the load. For 8 ohm speaker driven up to clipping, pretend there is a 48 ohm resistor on the power supply. For 600r use 3600r, for 2K use 12K. This is realistic for Power amps which are often run to clipping. It is generous for small-stuff which is normally nowhere near clipping, but OTOH you may have to pass a bench-test with sines run to the max.

I bet your DOA number is way-excessive (3 Watts in a 2" square gets hot!) so the 1Amp figure will be adequate; however I would look one line down and see if a 1.5A or 2A supply was not much more bucks.

(If you have to shave every last penny for a million-console profit production, you ought to mock-up and measure with chips from several lots before you sub-contract a 0.950,00-Amp supply.)
 
Thanks for the good clear explanation. So as the 8mA example figure is for the whole 30v range that means the total 950mA is for both rails together, and each rail only theoretically needs to be 475mA (ignoring any extra safety margins for now). That makes it a lot more convenient, as most PSUs I see round here are 1A per rail.

How about the BUF634's, is my 25mA allowance adequate?

I will probably go with the JLM Powerstation as it can supply 1.5A per rail, although I need to clarify whether that is 1.5A between both sets of +V rails or each, as I need a 12v rail for relays as well.

Cheers
 
Back
Top