QA451 causing incorrect THD measurement

Dear @matt and anyone else who is able to assist.

I have previously posted in this forum asking questions about the QA403. I am currently using the QA403 and QA451B to measure an amplifier I am building.

My setup is shown below:

I recently found that when I run a test, the left channel is coming up with a THD of around 0.02%. In the same test, the right channel measured a THD of around 0.00052%.
This is shown in in the screenshot below where the QA451 has been turned on to 8 ohms.

Initially I thought my left channel was faulty so I swapped the measurement cables around to see if the cables are swapped, if the left channel would continue to have such a poor THD. When I did this, I found the right channel was having the poor THD. So this means that both the left and right channel are working well and there was something else happening.

I then accidently had the 8 ohm turned off in the QA451 before I tested again. I ran the test without realising and then saw the results. Here the left and right channel both have similar THD. This is shown as below. I had not touched the cables or anything in the test setup. The only difference is the 8 ohm was not selected on the QA451.

I wanted to ask, what could be the cause of this in the QA451? Has anyone had this issue before?

Thanks for your help!


When you do a loop thru measurement, say with a 0dBV signal, are the two channels matched THD wise? It looks like your amp channels are not identical as far as gain and then power, goes. The THDs are pretty small and sometimes I think we worry a lot more about these things than we should, but understand your concern…unfortunately I do not have a QA451 so can’t comment directly on that aspect.

Hi @Bartosz, I think VAR’s on the right track here. First verify the QA403 is acting as expected in loopback with a 0 dB signal, 10 dB, etc. Second, the left channel THD is shown as -75 dB on the troublesome plot. Roughly, you should be able to see that visually on the display. What this means is the highest harmonic should be 75 dB below your fundamental. Or, if you have two harmonics of equal heigh, then they would be 81 dB below your fundamental. Unfortunately, the QA451 app is covering the most interesting part of your plot. Do you see a harmonic that is -75 dB below the fundamental???

@VAR and @matt

Thanks for getting back to me. So I did the loop test with the following setup:

For -10 dB I had the following result:

For 10dB I had the following result:

I had compiled the results into a table and plotted the graphs:

I was shocked at how much difference there was between the two channels. I then decided to swap the coax cables just as a troubleshooting step. You will be shocked at the results once I retested:

This is after the cables had been swapped for -10dB:

And 10dB after swapping cables:

Again, I compiled the results and plotted graphs:

Have you seen this before where cables can make such a drastic difference?

Hi Bartosz. When I find an excessively unexpected result in a test, the first thing I do is disconnect and reconnect the BNCs on both the DUT and QA403. Very often (obviously not always!) I find that this operation makes the unexpected results disappear. In my opinion, I believe this is due to the imperfect connection between male and female BNCs, and the simple operation of removing and reconnecting the BNCs often solves the problem. This is what I have found in my experience.

Hi @Bartosz, yes, I have seen cables that are on their way to failing exhibit this. Like @Claudio notes, unexpected results can take some time to diagnose. You can connect the cables in loopback and twist them, bend them, etc to see if you can force them to fail. And before you start a measurement, it’s not a bad idea to first look at the cables you will use in loopback. And sometimes it’s possible to not completely lock a BNC if you are moving quickly.

I’d bet you could take the bad cable and twist and bend it a bit and “fix” it in a certain position. Note the bad cable is letting in a lot of powerline. The shield is probably compromised someplace.

Since my professional career was as a microwave/rf engineer who did a lot of measurements above 8GHZ, cleaning connectors, namely SMA/3.5mm before each use and moving expensive coax cables around to see if they were going bad (mainly phase unstability) was part of the process. I still do it with the QA40x, but have not had as many problems now that all my load cables (banana connectors) are soldered. Where I still see a problem sometimes is when doing a measurement of say a preamp, where the “-” terminals are connected to a termination (or short), and I get more noise of hum on one channel. Swapping the termination with another one almost always solves the problem.

1 Like