Solid State Testing- Tug of war

[sumaudioguy]

We actually test amps this way. Amps that seem like they will take it anyway and wondered if anyone has a valuable opinion for this test.

Take channel 1 of a stereo amp and hook it to channel 2 of the same amp with a resistor. This would be red to red binding post with the resistor in between. We start with a 4 ohm power resistor. Drive channel 1 and read all your favorite measurements on channel 1. Next look at the signal on channel 2 which is used as the virtual ground when driving channel 1 with your favorite test signals. This residual signal seems to speak volumes about the amplifier or in many cases the ampli-fire. Many amps do not like this test but the amps that do well here seem to usually sound pretty good in general and are often the most reliable. Of course this is if all normal testing goes well also.

I like this test as driving a speaker is kind of like pulling a trailer with a truck, robust is better.

Anyone have a thought about this? I really like this test!

[Jan Dupont]

This test doesn't say anything about how the amp handles inductive and capacative loads like real speakers though.....

[sumaudioguy]

Originally Posted by Jan Dupont

This test doesn't say anything about how the amp handles inductive and capacative loads like real speakers though.....

Yes you are correct.

We go on to add complex circuits there and resonators- L C in series with various resistors as a standard test method using a repeatable procedure. It is a virtual ground so in theory if it were perfect there would be no signal on channel 2. Always far from that in reality.

The straight resistor is very useful and valid for determining real damping factor at any frequency.

Speakers may be placed where the resistor goes. Does not get any more real than that.

For horrible load simulation a brush type low voltage AC motor is placed where the resistor goes.

I hope some of you try this with your favorite amps! Let us know what you find. This is also a great way to hook two amps together and learn which is the master... I mean for testing.- tug-o-war.

[llwhtt]

Sounds similar to a method explained by Nelson Pass back in the early '80s. He used it in the Citation 12 mod. article for Audio Amateur. Go to PassDIY and find the Citation article.

Craig

[unclejed613]

actually this is a good way to measure output impedance (or inversely damping factor). if the output of amp A is 10Vpeak (20Vp-p), and the resistor is 10 ohms, then you are feeding 1Apeak into amp B's output. so a measurement with an oscope on amp B's output will show a relationship of 1mV/milliohm. this will also reveal if the amp is underbiased, because you will see spikes at the zero crossing due to the changes in output impedance as the transistors move through the crossover region. this may be an effective way to correctly set the bias on an amp without a distortion analyzer (i really need to experiment with this a bit more before i recommend it). i'm not yet sure what the effects of amp A being underbiased or overbiased would be, which is another reason i need to explore this a bit more

[ontoaba]

Good idea, change the resistor with real loudspeaker too they are reactive.
My test is little different, using sweep AC current source connected to amp output.

[ontoaba]

Originally Posted by unclejed613

this will also reveal if the amp is underbiased, because you will see spikes at the zero crossing due to the changes in output impedance as the transistors move through the crossover region. this may be an effective way to correctly set the bias on an amp without a distortion analyzer (i really need to experiment with this a bit more before i recommend it). i'm not yet sure what the effects of amp A being underbiased or overbiased would be, which is another reason i need to explore this a bit more

Not only underbiased, but bad combination of output filter and feedback loop in classB or small bias classAB creates some noise when crossing, while most classA is free.

[unclejed613]

i have done some investigation of this. the best method is to use a "power" version of a Howland current source instead of an amp and resistor. if the amp output impedance is more than 100 milliohms, the measurement begins to show a bit of error. about 1% per 100 milliohm. a howland current source would also be useful for charting the impedance vs frequency curves of speakers, crossover networks, etc... and the current is set with a simple scaling resistor. the one i built to test the idea has switchable scaling for 1A, 100mA, 10mA, 1mA.... i just have to experiment with it more to find out if there are other sources of error i'm overlooking. i also want to try using it as an ESR meter (or even better a curve tracer for in-circuit testing of components).

[jez]

It's an old idea. IIRC Tim De Paravicini described this method for testing damping factor many years ago.

[dirkwright]

Anyone build a DIY version of the Power Cube by Audiograph?