Strange THD using Load Resistor with Attenuating Taps

I’ve had a QA402 for a few years now and I’m starting to know it quite well. As an avid audio DIY hobbyist I recently set an objective of bench testing the mighty Carver TFM-75 amplifier. It’s a dual mono, 750 watt per channel monster. Thus, I set about building a load resistor that could handle its full output and attenuate the signal to levels acceptable to the QA402. I read all I could on this and other forums on what others have done. I elected to use 5 200 watt resistors per channel. 3 of them 2 ohm, the other two 1 ohm. I also elected to provide 4 taps providing a selection of attenuation levels. The schematic is shown below.

Once finished, I set about testing. On of my tests was to set the inputs levels to generate 200 watts into 8 ohms (40 volts). I then measured THD and output voltage for each of the four taps and the unattenuated output of the load resistor. As you can see from the test data below. the THD increases as the attention rises to -18dB. Changing attenuation levels in the QA402 has little effect on the THD measurements. The measurements were taken at the full 42 dB attenuation of the QA402. FFT size was 64K. Sampling rate was 96K.

I don’t understand why the THD would increase so dramatically as I connected to different attenuation taps on the load resistor. Is something screwed up here?

Thanks in advance for any ideas.

Here are the charts for 0dB and -18dB attenuation. Right channel only. Ignore the numbers for the left channel, it wasn’t connected to anything.

Note in the -18dB chart that the fundamental at 1Khz attenuates as expected, but the even harmonics actually increase. Note further that the 60Hz spike and all of its harmonics attenuate as expected as well. What’s going on with the even harmonics?

I think you may need to go into the dBv tab and set the input attenuation to -18dB


Thank You.

I did that but the only thing that changes is the power, gain and output numbers. THD and SNR remain the same.

Hi. When during a measurement I find something strange and absolutely unwarranted, I the first thing I do is check the various connections, moving them around and reconnecting them. Especially BNC connectors are often subject to introducing unexpected results, but other types of connections can also give “strange” results (semi-rectifying connections). It happened to me quite often that after these checks the “strange” results disappeared. Of course, I don’t know if this is your case, but doing a check of the kind I have described costs nothing.

1 Like

It’s the odd harmonics that increase actually, (3/5/7/9/11 etc).

How are the resistors physically arranged on your load/heatsink? Is there some magnetic coupling going on at high powers?

Matt also discussed loads and placing the sense resistor in a series load array, in the centre of the string, (not at the end) to reduce the peak voltages the QA40x sees. At 200WRMS, you are at the limit of the 56Vp of the QA402/3.

"Finally, why is the load sense resistor R2 in the middle of the string and not in the top and bottom location? Theoretically it could be in any location and still perform the same task. But practically, if we locate the sense resistor in the middle of the string, we reduce the peak voltages seen by the QA401. "

I based my suggestion as there was no attenuation displayed on your plot showing -18dB. Here is my load/tap set up, and I can (in theory) do 800w into 4 or 8 ohms. I have done 700w into 8ohms w/out any problems…


Scott - Thank you again for your thoughts. I elected not to enter any external gains as I wanted to verify the various taps were doing what I expected. I watched the right channel RMS dBV output to make sure I was on the right tap. In the above two plots, the first one on the unattenuated tap shows 32 dBV. The second one on the -18dB tap shows 14 dbV. 18dB down from the first one.

I watched you video carefully. Thank you for all the effort you put into making it. I relied heavily on your design when crafting my own. The two designs are very similar. I have a couple questions.

  1. Are you using inductive or non inductive load resistors?
  2. Have you had a chance to compare THD measurement between the 0dB 8 ohm tap and the -18dB 8 ohm tap?
  3. Have you seen any odd behavior stemming from comments by @matt recommending the use of a middle resistor rather than one at the bottom (or top) of the string?



Thanks for your comments. I too have spent many hours chasing bad connectors, cables and adaptors. I initially suspected exactly that problem here. But after much wiggling of connectors and a bit of loop back testing, I gained confidence that my cabling was OK. Further, I used a completely different set of cables and tested through the left channel of the QA402 and got essentially the same results.

Thanks again.

A few more details on my setup.

Below is a photo of the resistor array. Two arrays actually, one for left and one for right. Signals from the amp enter from the top using the binding posts. The QA402 is connected at the bottom using signals from the terminal block. The big black terminal is of course negative and the big red one positive. The four red ones in between are the four taps from the array. The top two resistors are the 1 ohm resistors shown in my schematic. The bottom three are the 2 ohm resistors. The switches are used to engage or disengage the load.

I have not tried the middle resistor tap point since I did not want to redo my loads. I bought wirewound resistors for the loads- no specification on inductance. For lower power levels, say <5w/8ohms, I use the 0dB tap, and say I then amp going to be measuring 150w/8ohms and go to the 2.5dB tap. I then go into the dBV menu and set the input gain to -2.5dB and move the attenuator to 30dB instead of 18dB (where it was for the 5w measurement). I find very little difference in the THD measurements when I’ve looked at the two. I think I talk about this in the MC2300 video when I will do a THD vs Freq & output power levels as I will use the 18dB tap and set the input gain to -18dB and start off at say 3w and go to 200w all without changing the QA40x’s attenuator. I have gotten pretty good results that way for the lower power levels.


Thanks for the comments. I’ve tried to respond below.

Well crap. Proves once again that some times I can’t count. Thanks.

Photo above of the physical layout. Initially I thought you had nailed it with your comment about magnetic coupling. The resistors are just about as close and they can get. So I unwrapped things a bit to separate the resistors and re ran the tests. Photo below. Unfortunately, I got nearly identical results. Same increasing THD as you work your way up in attenuation levels.

I saw that. But I was puzzled as to why there would be any difference in peak voltages between middle or bottom sensing resistors. In any case, I wanted to feed the QA402 singled ended, so I was doomed to using the bottom resistors. I remain puzzled as why they are less desirable.

Good point. I’ve dialed down the power to 150 watts, re ran the tests and got significantly improved THD for the un attenuated tap. Unfortunately the attenuated taps still show significantly higher THD as the attenuation level goes up.

Are you connectiong to the QA differentially or single ended? The single ended connection can get ground currents mixed up with the signal you are trying to measure. If the amp is BTL then you want the connection to a sense resistor in the middle of the string. The goal isto minimize the common mode the QA sees and at the same time keep ground currents away from the test inputs. Twist the wires together to minimize the loop area on the test/probe wires.
Your current is 10-A RMS or 14A peak. That can generate a pretty strong magnetic field so you would want to interleave the resistors to again minimize the loop area as the current goes around the circuit.
The power resistors I have measured had inconsequenctial inductance. 4 feed of zip cord has more so not an issue. However every connection can cause distortion when the current gets high enough. Make sure the connections to the resistors are really good. At high currents (100A) solder is not recommended, it can melt from the heat. Crimps and screws and nuts can be better but need regular maintanance. (True for your hose wiring as well.)

I don’t think I understand your setup.

To me this looks like a flexible load with taps at different impedances/resistances. It would be useful for testing how the amplifier performs with different loads.

I think what you want to do would be have a high impedance voltage divider in parallel with the load, the QA403 gets the signal from the voltage divider and because it is parallel with the load it makes negligible difference to the impedance loading the amp.

This has some drawbacks. Shot noise in resistors etc. Matt’s approach of measuring across a small sense resistor in series with the load is superior in that respect.

Here is an attenuator board I made, I used it to knock off 6dB and this enabled me to measure a 300W beast.

It looks like you are loading your amp with a 1 Ohm load on the maximum attenuation setting, could this be why?

Sorry for the slow response.

My intent was to use single ended. The measurements I provided earlier were all single ended. I have done the same tests differential and gotten similar results.

Nope. The amplifier is always connected to the fuel 8 ohm load as shown in my diagram. The multiple tap points are used to select from multiple attenuation levels. The analyzer is connected to the tap points.

Good suggestion. This is Plan B if I can’t confirm what is going on with the original design.

Think you should make the resistor divider differential such that you have same impedance to both sides of the load resistor. Meaning r1, r2, r3 with r2 is the measurement resistor (differential voltage measurement) and secure that value r1 = value r3…

The goal is to avoid large asymmetrical common mode modulation which can introduce THD.