Power Amp Load Box

[alexc]

I've built a couple of tube amps and am going to have a go at some discrete solid state amps with crossovers soon.

One of the pieces of test kit one needs is a good load box.

Not a good idea to power up a tube amp with no load. And it's very nice to measure amp performance with your real time audio analyzer system (RTAS).

Basically it is a high power dummy resistance connected across some input speaker terminals with an additional parallel high-resistance voltage divider. The RTAS output is taken across one leg of the divider.

So, I'm  starting this thread to gather opinions on how to make one better than the one I made 

Please feel free to provide your inputs.

----

The load box needs to have speaker level input connectors and should also have an output which allows you to connect a suitably attenuated version of the input signal to an RTAS for detailed observation of the amp under test performance.

The attenuation of the RTAS output should be variable so that different input power levels can be 'mapped' to the available RTAS input voltage range in sensible ways.

So features which can be provided are :

- aluminium enclosure for heat dissipation
- input speaker terminals, output speaker terminals, balanced RTAS output connector

- non-inductive power resistances for dummy loads

- switch to select dummy resistive or speaker 'thru' as the high-power load
- switch to select different dummy resistances  ie.  2ohms, 4ohms, 8ohms, 16 ohms

- switched variable RTAS output level with mute

And  something I'd like to try as well is a switched defeatable filter for removing the excitation signal from the RTAS output.  So that when you observe the amp signal on the RTAS, the test tone is removed. Not sure how this is properly done so that one can see the aggregated distortion components on the RTAS.

I'll put up my primitive schematic shortly 

 

[Brian Roth]

What sort of power requirements for the dummy load(s)?  Four of these:

http://www.mouser.com/ProductDetail/Vishay-Dale/NH0508R000FC02/?qs=sGAEpiMZZMvNd0dY0Kymzl%252bbo6xW4jofjpDy5L8Y8tM%3d

...will give you mono 200 W at 8 Ohms series-parallel... OR....... 100W 16 Ohms series or 100 W 4 Ohms parallel in stereo.

25+ years ago, I bought four of these:

http://www.mouser.com/ProductDetail/Vishay-Dale/NH2508R000FJ01/?qs=sGAEpiMZZMvNd0dY0Kymznz%2fzbWwLvuhFqEVb%252bALMAU%3d

...but OUCH...I sure didn't pay anywhere near that current price.

Bri

 

[emrr]

One of the Daven or General Radio power output meters would be useful. I know there's a 100W Daven version.  You get variable load that can be used to determine matching impedance at various frequency.  The smallest versions are good for determining output Z of small signal tube amps.

 

[alexc]

Thanks for the links - I'm interested in around 50W capacity.

Those high power resistors are just the thing.  I have used cheap ones so far - probably are not non-inductive, but seem to work OK. I'm not quite in the top end Vishay-Dale region yet 

I will do some searching on the power output meters you suggest, emrr - looks interesting for use in this area of measurements.

I am yet to successfully measure the output Z and damping factor of my diy amps. A means of facilitating thus would certainly be a useful addition to the load box feature set. I have the procedure somewhere - I'll dig it up and post.
And then actually do it!

Cheers

 

[emrr]

I see a 100W Daven OP-962 on ebay at the moment, price is high and it's in the US.  Marconi is probably a brand to seek in the UK. 

These meters are inductive load, a big multi-tap coil along with resistive elements. 

 

[ruffrecords]

One thing you do need to do is seriously over-rate the load resistors even if they are bolted to a substantial heat sink. I once made a 12V heater load to test a 5A supply i.e. 60W load. I over-rated the resistors 100% but they got so hot that the soldered connection between a pair of them melted.

Cheers

Ian

 

[joaquins]

Maybe for avoid melted soldered conections you could use long heavy solid copper with screw and solder away from hot resistors, so temperature go down all the long of the copper and don't melt solder.

If my fisics are right heat is a function of the power dissipated and temperature could go down with bigger heat sink, not with higher rated resistors. Though over-rating is a good and safe choise for obvious reasosn. Also is true that higher rated resistors have bigger self heat sinks but if you are going to attach them like external heat-sink with chasis will make things cooler and a fan also would help.

JS

 

[Matador]

I've had good luck using power resistors in TO-220 packages, to take advantage of the large number of heatsinks available.  They can be found in the 25-50W range.  Aavid sells a TO-220 heatsink that is in the 3.7 degrees C/W range, should be able to sustain around 10W indefinitely (and more like 20W if used with a fan).

 

[JohnRoberts]

One popular feature is a 100W (?) incandescent light bulb in series with the line cord. This way when you a first bringing up an amplifier, if it tries to draw too much current the light bulb lights up and limits the current. A bypass switch can short out the bulb after the amp is working right.

Another consideration is that for lowest distortion bench testing, high linearity power resistors are sometimes used. For very high power amplifiers, water heater elements can provide lots of power sinking and are easily cooled in a bucket of water. Sorry I do not have more specific info about those resistors, never had to source them myself. (they were available inside Peavey from the testing group). 

JR

 

[alexc]

Another consideration is that for lowest distortion bench testing, high linearity power resistors are sometimes used

Yes - I wondered about that : how much the precision of the power resistors contributes to effects in distortion testing.

Not that it's a limiting factor in my case, even though I use the cheap orange aluminium clad power resistors. I'm still figuring whether the distortion I see on my RTAS is the amp or the RTAS.

I do know that approaching the clip point (about 2dB or so before) of the Motu interface I use, distortion goes up dramatically - hence the need to easily 'map' the output of the amp to the useful input range of the sound interface.

Preferrably in a relationship which can preserve the information relating to noise levels and distortion so that measuring this way is still meaningful.

 

[alexc]

For very high power amplifiers, water heater elements can provide lots of power sinking and are easily cooled in a bucket of water.

Last time I worked on a water cooled load box was in my days as a trainee elec eng with the local state rail corporation. Back in the early 80s - I would work with a team measuring the power characteristics of the diesel-elec locomotives. 3000hp supercharged diesel feeding a huge dc generator with giant cables connecting it to huge steel tanks full of water with giant plates that were raised in and out to vary the load.

Those things would steam! Nothing like revving the hell out of a 3000hp engine

 

[emrr]

Now that's a power show I'd like to see. 

 

[pacemaker]

regarding 250W resistors,
i found these , cheaper than the 300$ ones

http://fr.mouser.com/ProductDetail/Arcol/HS250-8R0-1/?qs=sGAEpiMZZMvNd0dY0KymzvyeJONiK2j5%252bk6V1VO%252bBjk%3d

 

[zayance]

There are choices on evilbay as well if you search for "High Power non inductive resistors" 1/2/4/8/10/16 ohms and 100/200/300/500 Watts.
Big green sausage with mounting brackets.

EDIT: sorry read that too fast, you want to fit it in a box, might get too much space, the 100W seems to be ok, but anyway just sayin....

Well seems like somebody did just that 8)

 

[alexc]

I think that is the last word on  quality and capacity for a load box 

Thanks for the pic.

 

[sawyer]

Hi - This thread sounds like what I am looking for.

I just gearing up this past year getting some used audio gear and am wanting to take the next step with measurements at all points source to the accoustical domain. Capture to a PC and analyze with software.

I the measurement location identified in this thread is perhaps the most challenging.

As I read this this thread the problem statement is:

Since the soundcard or whatever capture device is used needs a specific voltage level input and the power is variable it involves obtaining/building a power amp load box which contains high power non inductive resistors. This needs to drop the voltage and it needs to do so without creating noise

A quick check of what was cited Daven OP-962 - looks like an off the shelf alternative to building your own load box.  I haven't filled in the gaps in the setup yet and would like to hear someone elaborate on the "noise" issues since that can make the exercise from a qualititive standpoint moot.  Hopefully this is still on-topic for folks.

 

[alexc]

Yes - whatever load box you use needs to have a resistor network to output a signal to your  soundcard/RTA application which is a fraction of the power the amp is dumping into the load.

The calculation becomes something like this :

Amp is 30Wrms into 8ohms. So, I start with 8ohms of high power dummy loading.

I want to test the amp over it's full range, so 30Wrms to the dummy load means 15.5V rms or around 47Vpp for a sine wave.

OK so I have 47Vpp in the load at max power. My Motu interface starts to distort before clipping at around 10Vpp for a sine wave.

So - I need to divide the 47Vpp down to say 9Vpp to make best use of my soundcard's available input range.

You need to do it without drawing much current away from the dummy load, so it's a parallel resistor branch across the load.
The load is small resistance, so you need a high resistance so as not to messup the load.

So you calculate a pair of resistors in series and put across the load. Then you do a voltage division of the two resistors 9/47Vpp

There it is - you can now display safely on your RTA the output of the amp at full belt without significantly altering the load presented to the amp.

Now - that's fine for relative measurements and also qualitative measures.

For absolute measurements it's not accurate - you would need to reasonably relate the dummy load voltages to the RTA amplitudes and all that.

But for basic amp amp testing of noise, distortion, spectral characterisation, feedback ratios and all that , it works well.

 

[MagnetoSound]

You're on the right track, but 9Vpp into your MOTU seems like a lot.

0dBFS is old thinking, why bang it all the way to the top?

You could aim for a nice +6dBu or so max (call it 1.5V RMS to make it easy, since you have 15V RMS to begin. You want a 10:1 divider.)

I usually find a 1k shunt works well, since it offers a suitable source impedance to pretty much anything that follows. So you want a 10k/1k divider. The 1k can easily be a pot.  8)

If you do want to increase the upper range, you can reduce the 10k accordingly.

 

[alexc]

Fair enough - I like to use the avaible input voltage range as fully as I can.

At 9Vpp the distortion at the motu remains low so why not use it. And that is the max case.
A lot of the testing happens at lower power levels, in which case I'm sending a much lower voltage to the Motu.

If I divide too far, I simply am reducing the vertical sensitivity of my RTA. Don't see any point in that.

Actually, the way I go is to have a variable sensitivity send to the Motu.
I find it convenient that way.

 

[MagnetoSound]

Fair enough backatcha, mate!