I ran the Power Output Impedance automated test on a power amp and am not sure of what to make of the result- I could not find any wiki on that test…
Hi @Var, what hardware and software are you running? Is it possible you flipped your specified load impedances?
QA402 and 1.161. THe test prompts you to connect to either the 8 or 4ohm load so I am pretty sure I did it correctly, but can certainly try it again…thanks for getting back with me
OK, thanks. When you watch the measurement happen, set the frequency range to something narrow and make sure you have the peak DBV display active. The first test will run (with 8 ohms, for example). And when it’s finished, note the peak amplitude at the last frequency point. And then the second test will run. When it’s done, note the peak amplitude at the last frequency point.
With 4 ohms, the amplitude has to be less. And that will yield a positive output impedance. But if the amplitude with 4 ohms is MORE than the amplitude with 8 ohms, then that will yield a negative output impedance.
I tried it again, this time making sure I was on Peak DBV (was on dBR) and got better results:
That looks more reasonable and now I can calculate the damping factor (for what that is worth) !!
1 Like
Just a Wish List thought, but it would nice if the damping factor wasplotted on the right side. I assume that it would be just for an 8ohm load…
Since this post I have pretty much not had any problems when measuring output impedance of an amplifier- I have measured several, both tube and solid state. A few days ago I was measuring an Onkyo SS amp that has a discrete transistor output stage (class ab, direct coupled). The output impedance looks normal above 200hz, but below that , it has a linear slope going down to -3ohms at 20hz:
I re-ran the measurement and got the same result. I am wondering if the methodolgy below is being used by the Q40x to calculate the output impedance (from the CEA standards):
Though it is not mentioned here, the earlier IHF spec had the output power at 1w/8ohms, which is what I run the test at. If the ratio of the Open ckt voltage and the loaded ckt (8ohms always in my case) is less than 1, the output impedance will be negative, but am curious if this actually happens or is something else going on…? Thanks…
Hi @Var, the software takes a look at two impedances, not necessarily an open circuit. So, you can run the test at 4 ohms and 8 ohms, or 4 ohms and 1000 ohms.
This is using stepped tones, is that right? If you are seeing a negative impedance, it probably means the amplitude at the lower impedance is greater than the amplitude at the higher impedance.
At 100 Hz, you are showing a negative impedance. Can you set the generator to 100 Hz and manually measure the peak amplitude with your two difference impedances and report those?
Thanks for getting back with me, Matt. Am using an open ckt for the 2nd impedance, though the actual impedance setting in the plug in is 1,000,000 which is the max allowed and an open ckt in my books. The plugin appears to be stepping tones. I will test the amp as you ask and get back later today. Probably will run it at 250, 150 and 100hz…
Updated- I re-ran the test and got mostly the same plot as before:
Here are the setting used:
Then I did the manual measurements with (~1w) 8ohms and then removed the load for the next measurements:
The calculations do not show anything that would cause the impedance to go negative…?
It would be interesting to play a single frequency to see what is going on with voltage measurements. That line looks very straight which makes me slightly suspicious! 
Hi. I wonder, but if instead of inserting 1 Mohm (so no load at the time of measurement) you insert a load, say a resistor with a value of 100 Ohms or less actually connected, does the problem still occur? It would be interesting to try.
The measurements I posted were made setting the QA403 to the noted frequencies
The reason I did not is that the spec definition is for the open ckt voltage, but I suppose it would not be too difficult to do that.
Yes, you are absolutely right in saying that the specs talk about open circuit, but it would be interesting to see if the plugin problem recurs even when the amplifier is connected to a resistor that is also quite large but does not approximate the open circuit
Hi @Var, is it possible that your FFT size is too small and the readings below 100 are erroneous due to noise?
I just did a measurement on a QA461 with 10 ohms and 1Mohm as the loads, stepping from 20 Hz to 400 Hz and I can see it negative in places when the FFT is very small (1K)
From your table, the unloaded amp has an open circuit voltage of 3.047Vrms. If the amp impedance is 0.0163 ohms, then the loaded (8 ohm) measurement should be 3.047 * 8/(8.0163) = 3.040 Vrms. But you measured 2.998Vrms.
Or, looking another way, 3.047 * 8 / (8+X) = 2.998, where X is output Z. Solving for X give X=0.131 ohms. What is your last column showing (Vopen/V8)-1?
The last column was just put there to see if that portion of the formula ever got negative, which it did not. The FFT size is 128k, and the sample rate is 96k. Looking at the 1st plot of the output impedance it looks like it is about 0.15ohms at 1khz, which is pretty much what your calculations showed. This is not the lowest output impedance I have measured, and have not had problems at the lower part of the band except one other time long ago. This particular amp is pretty quiet with a great frequency response. (at least with 8/4 ohm loads).
OK, thanks @Var! I wonder if the user specifies the max impedance if it should just default to the single load calculation at that point…Or maybe add a checkbox so the user can specify they only want an open-circuit and a single impedance measurement. What do you think?
The manually measured numbers should be delivering a negative impedance. That suggests the measured voltage is greater than the source voltage.
@Matt- the manually measured numbers I made always give a ratio >1, so the impedance should be positive. I am not sure what calculation the software is using, but if those numbers are ared, does it show a negative impedance? My main point in doing this measurement is to get the Damping factor as that is what is specified most often in the vintage gear I test, and that is obtained by dividing 8 by the output impedance that was measured. If there was an option (or automated test) that you clicked for Damping Factor, that asked you to connect an 8ohm load, ran through the measurements at a specified output level (1w/ohms suggested), and then prompted you to remove the load and run another stepped sweep and plotted out the damping factor, that would be great. I would also have the automated routine pull in the recommend FFT, Sample rate, etc. Plus while I am asking for a lot, have a built-in frequency steps from 20hz-20khz (10hz steps to 100hz, 100hz steps to 1khz, and then 1khz steps to 20khz). There is a low frequency damping factor that is specified at 50hz, but this would probably give a pretty accurate plot for most folks.
@VAR Are you setting the output level of the QA in the automated test to match the sensitivity of the amplifier to give 1W (2.83V@8R)?
Dampling factor was as you say, always quoted with repsect to 8R back in the day and measured at the nominal 1W either swept or plotted at specific frequencies.
With my old analyzer, I just ran a 492 point frequency response plot of 8R loaded and unloaded, 20-20k, exported the data calculated and plotted either DF or output impedance in Excel. I did have to take into account amplifier DC offset if it was more than a few millivolts as the analyzer had both AC and DC inputs (flat to DC, but offset and ADCs with quick sweeps/aquisitions can throw a big spanner in the numbers).
Like if I changed attenuator/gain ranges too fast, there was settling time which would throw off the lowest frequencies.
I’m just mentioning that as a potential for negative DF numbers at low frequencies. Not sure about the QA’s coupling/front end. My analyzer was essentially just a cap with software compensation for AC, like a scope front end.
There are some changes in the software Matt has made in relation to DC offset, not sure if that affects the test you are using either. I will have to experiment with mine and see how it behaves.
@Restorer-john- I set the 1w/8ohms at 1khz and enter that value into the automated plugin- pretty close to 1w anyway. Most amps seem to be within 0.6dB over the 20-20khz range. The amp used for this measurement was only down .1dB at 10hz. I don’t know anything about the DC offset voltage- I have never done anything with the QA40x in relation to that. Since this amp performed so well, I did not measure it- that is not something I typically do unless I think the amp is running hotter than normal and have bias/offset settings for the unit…