Measuring Frequency Response and Distortion of amplifiers （4x150w RMS and 4x80W RMS）

[icbcodc]

For the calculation of the divider I considered the measurement of the distortion at the level of maximum output power. In our case the maximum power was 150W which on 8 ohm corresponds to a voltage of about 35Veff while the maximum level accepted by a sound card has been set at 1Veff therefore the 2.2Kohm and 82Kohm partition guarantees that at the maximum power the input voltage the sound card is less than the expected maximum level. with 35 Veff the input level to the sound card is 0.91Veff .

Hi Antonio

I still can not understand how to calculate the final resistor value of the divider, cause my basic knowledge is poor.

Cause the structure of divider in the circuit looks different with that in the picture of online divider calculator.

This below is the divider in the circuit:

This below is the divider in the online calculator:

For me,they are different. Should I just leave the online calculator away?

 

[dotnet]

In our case the maximum power was 150W which on 8 ohm corresponds to a voltage of about 35Veff while the maximum level accepted by a sound card has been set at 1Veff

If you look up the specs of the Behringer device, the maximum input level is actually +20dBu, i.e. ~7.75V. The ~3V over the 2.2k resistor at 150W will be fine.

 

[trobbins]

R2=22k, R1=2k2, RL=1Megohm (instrument input). Option A: forget about RL equations and just use 'no load' equations for instrument input. Vout = Vin x 2k2/(22k+2k2) = Vin x 0.091. But max input is -3dBu (0.55Vrms).

R2=22k, R1=2k2, RL=?? (line input). Option B: use RL equations but assume RL =10k. Vout = Vin x 1k8/(22k+1k8) = Vin x 0.076. Max input is +20dBu (7.7Vrms).

So 35Vrms is either divided down to 3.2Vrms (which is too high for instrument input) or 2.7Vrms which is not too high for line input.

Normally any amplifier testing with a soundcard uses an rms voltmeter connected to the amplifier output, as that confirms the output power of the amp, and can be used to confirm what amp output voltage can be applied to the soundcard input before it distorts too much.

 

[icbcodc]

R2=22k, R1=2k2, RL=1Megohm (instrument input). Option A: forget about RL equations and just use 'no load' equations for instrument input. Vout = Vin x 2k2/(22k+2k2) = Vin x 0.091. But max input is -3dBu (0.55Vrms).

R2=22k, R1=2k2, RL=?? (line input). Option B: use RL equations but assume RL =10k. Vout = Vin x 1k8/(22k+1k8) = Vin x 0.076. Max input is +20dBu (7.7Vrms).

So 35Vrms is either divided down to 3.2Vrms (which is too high for instrument input) or 2.7Vrms which is not too high for line input.

Normally any amplifier testing with a soundcard uses an rms voltmeter connected to the amplifier output, as that confirms the output power of the amp, and can be used to confirm what amp output voltage can be applied to the soundcard input before it distorts too much.

Thanks a lot, trobbins.

I'm sorry, cause I forgot to tell you the resistor value of speakers is 4 Ohm (car speakers) , and the amplifiers are also car audio products.

So, the max. output of the 150W amplifier under 4 Ohm is 24.49V rms . (square root (150W * 4Ohm))

Vin is 24.49V, R1=22000 , R2=2200, RL=10000 and Vout calculated is 1.85V , is it right?

 

[trobbins]

Yes. But note that RL=10k is not confirmed in the device specs afaik, so you still need to be cautious when increasing amp output signal level and observing any rapid increase in harmonics, and visible clipping of sinewave, to assess when either the amp or the soundcard is starting to affect measurement integrity.

 

[icbcodc]

Yes. But note that RL=10k is not confirmed in the device specs afaik, so you still need to be cautious when increasing amp output signal level and observing any rapid increase in harmonics, and visible clipping of sinewave, to assess when either the amp or the soundcard is starting to affect measurement integrity.

Hi trobbins

Thank you, I have a RMS VOLTMETER, I will monitor the output voltage of the amplifier.

Should I buy a 4Ohm 200W resistor to replace the 8Ohm 50W in the diagram?

I will also buy a 22K Ohm 1/4W resistor and a 2.2K Ohm 1/4W resistor, and two TS like this.

 

[trobbins]

You need to read your amp specs as to what range of load impedance is acceptable, and also note the speakers you aim to use the amps with. If your speakers are 4 ohm and your amp can work in to 4 ohm, then yes a 4 ohm load is worthwhile for testing. You need to read your amp specs/operating instructions for the connectors they use.

 

[icbcodc]

You need to read your amp specs as to what range of load impedance is acceptable, and also note the speakers you aim to use the amps with. If your speakers are 4 ohm and your amp can work in to 4 ohm, then yes a 4 ohm load is worthwhile for testing. You need to read your amp specs/operating instructions for the connectors they use.

Thank you trobbins.

Yes, my amplifiers support 4Ohm, and impedance value of most of my speakers are 4 Ohm. Cause 4 Ohm is typical value supported by car audio products. One of my amp supports 2 Ohm load also.

Do you think it's worth measuring the amp under 2 Ohm load?

BTW, does the quality of TS plug matter?

 

[icbcodc]

Attention
the scheme proposed by our friend icbcodc works only with amplifiers that have the output referred to ground. use this circuit under penalty of breaking the power amp.

Greetings Antonio

Hi, Antonio

Though the amplifiers are car audio products, I'm using them in my listening room.

I think the power transformer (220V to 12V) is grounded like a home audio product.


220V Power Transformer (power supply) plug:

POWER TRANSFORMER(Power supply ) OUTPUT: (220V to 12V output)

Amplifier Power

 

[icbcodc]

Hi trobbins
It's the first time that I have made this kind of circuit for amplifier measurement. Could you help me to check it. Thank you very much.

I made the circuit like this

The resistor connection detail:

The circuit diagram:

I measured the resistor value of the circuit( it's for the UMC202HD sound card input), it's 1.998KOhms,

I also measured the resistor for the amplifier high level ouput, it's about 4Ohms

 

[jschwender]

Looks good, should work. Just don't forget an approprate heat sink, otherwise you burn the resistor quickly.

 

[trobbins]

The different ground points may introduce noise at higher power levels, and assumes the amp's output neg terminal is internally connected to the signal input gnd terminal (is it?).

 

[icbcodc]

Looks good, should work. Just don't forget an approprate heat sink, otherwise you burn the resistor quickly.

Thank your, I don't have an heat sink, so I will monitor the tempreture of the resistor, if it's too high, I will put a wet towel on it.

 

[icbcodc]

The different ground points may introduce noise at higher power levels, and assumes the amp's output neg terminal is internally connected to the signal input gnd terminal (is it?).

Yes, I've tested it, the negative terminal of speaker output is internally connected to the negative terminal of RCA on the amp.

Everything looks fine.

 

[trobbins]

Once you have made some measurements that seem reasonable, then you need to appreciate that there is still another big step needed to make accurate measurements on aspects such as distortion, as there are many many test setup influences that modify the distortion readings you may make. There are a number of threads over on diyaudio that may help you if you keep going.

For example even the load resistor is an issue as it appears to have a temp coeff of 100ppm/degC, and unless you heatsink it really well and also derate its power level a lot then it may introduce noticeable distortion - also the 200W rating is only if you can keep the case at 25C, and its power rating falls to 100W if you can keep the heatsink and ambient to 150C.

 

[icbcodc]

there are many many test setup influences that modify the distortion readings you may make. There are a number of threads over on diyaudio that may help you if you keep going.

the 200W rating is only if you can keep the case at 25C, and its power rating falls to 100W if you can keep the heatsink and ambient to 150C.

Thank you so much, trobbins

I will read the articles with long long threads on DIYAUDIO, and it will take sometime. Actually, I hope there is a tutorial, cause there are many details (procedure) I need to know.

Today, I will try to measure the amp for frequency response and distortion.

And in REW Should I caliabrate the output dbu level of soundcard and the final input voltage level of the soundcard first?

 

[trobbins]

A thread started by xrk971 is worth scanning through (https://www.diyaudio.com/community/threads/howto-distortion-measurements-with-rew.338511/)

You should run and use a loopback calibration file for the soundcard and probes (if any) you are using, and over the frequency range you are measuring across. But also measure and note the raw response without calibration, as that indicates where your test setup is starting to roll-off and make adjustments (from a flat-line response). Your DUT then is the cause of frequency response measurements when the DUT is included (assuming you are not overloading or doing anything else with input and output connections).

Fwiw I just modify the signal level to give a certain mid-band output voltage (DMM measurement) that aligns with say 4V across 4 ohm to give 4W, and then save the measurement as a 4W measurement.

 

[icbcodc]

A thread started by xrk971 is worth scanning through (https://www.diyaudio.com/community/threads/howto-distortion-measurements-with-rew.338511/)

Fwiw I just modify the signal level to give a certain mid-band output voltage (DMM measurement) that aligns with say 4V across 4 ohm to give 4W, and then save the measurement as a 4W measurement.

Yes，I know the thread by xrk971, it's very long, and need sometime to read carefully.

I will do the loopback calibration in preference menu of REW.

I also read a simple one recommended by John , https://enjon.uk/2019/09/01/a-diy-approach-to-distortion-measurements-for-audio-amplifiers/

I think it's good for starters to make a simple but qualified circuit.

I will read both of them carefully.

 

[icbcodc]

Looks good, should work. Just don't forget an approprate heat sink, otherwise you burn the resistor quickly.

Hi jschwender

I disconnected the charger of my Thinkpad T410 and use battery to run REW.

The loopback distortion of UMC202HD is not very good. If I turn DIRECT MONITOR on, the THD is about 0.4%, if I turn it off, the THD is about 0.14%.

 

[jschwender]

That confirms my experience with this device. The pre-amplifiers are advertized as "warm and organic MIDAS sound". That explicitly means it adds harmonics. If musicians like that, fine. But it makes it unsuitable for measurement purposes.

 

[trobbins]

Perhaps for starters you could use RTA to view the spectrum of the loopback, and the HD levels, and adjust the tone generator level through a wide range, and similarly adjust the output level knob, and the input level knob, and swap over to the phones output and do a similar level sweep, and use the instrument input select as well as the line select. You need to make yourself aware of what is able to be generated from your interface for all settings and levels, as there may well be a relative sweet spot. As far as the inputs, its likely that the knob level can remain at min for your application and to maintain the lowest noise floor. RTA mode allows you to visually appreciate when HD levels rise and fall, and the Distortion box confirms that as % level changes. You also need to change the sampling rate, and see if you can use ASIO driver (as the best driver for high rate performance).

 

[icbcodc]

RTA mode allows you to visually appreciate when HD levels rise and fall, and the Distortion box confirms that as % level changes. You also need to change the sampling rate, and see if you can use ASIO driver (as the best driver for high rate performance).

Hi trobbins,
I tried many times, and get the best loopback result.

Laptop is running with the battery.

Windows version

Behringer UMC202HD hardware OUTPUT and GAIN1 settings.

Control Panel settings:

4096 samples is the highest value accepted by current REW and current windows.

Input volume: 0

Output Volume: -22 higher , more distortion

Calibration soundcard settings in REW:

input : 18.96dBFS , if higher , more distortion.

THD: 0.072%, not too bad, maybe enough for normal car audio products measurement.

 

[trobbins]

It may be enlightening to use the 'stepped sine' feature and the Step level function in RTA to see how the 2nd Harmonic level changes with line out level when doing a loopback test at a mid-band test frequency like 1kHz. For that type of assessment it won't really matter if the higher order harmonics are lost in the noise floor given that 2HD appears to be quite a lot higher. The span of levels to step could be say -40dBFS to 0dBFS in 5dB steps, where 0dBFS doesn't cause gross clipping. You may also want to adjust Gain 1 to min level (or near min) to lower the noise floor, and play with a few adjustments that lower the noise floor, such as FFT length and averaging. I would expect to see a rising level of 2HD% as the level approaches 0dBFS. Each sweep of tests can take a few minutes, so watch the progress bar to confirm all is proceeding.

Another test of interest would be to generate a sinewave for loopback, and then use the 'add harmonic distortion' feature to just select the addition of a 2nd harmonic signal, and then go through a process of adjusting the added 2nd harmonic level and phase and watching what change that can do to the 2nd harmonic level in the RTA screen. That adjustment process can be quite confusing, but is often iterative and requires the adjustment level of the added signal to be tweaked to where the RTA level of 2HD just stops reducing below a certain level, and then adjusting the phase to see if the RTA 2HD level can be enticed to reduce further, and then returning to the level adjustment to iteratively try and coax the 2nd harmonic down further, then back to the phase... and try and get the 2HD level down in to the noise floor.

 

[icbcodc]

It may be enlightening to use the 'stepped sine' feature and the Step level function in RTA to see how the 2nd Harmonic level changes with line out level when doing a loopback test at a mid-band test frequency like 1kHz. For that type of assessment it won't really matter if the higher order harmonics are lost in the noise floor given that 2HD appears to be quite a lot higher. The span of levels to step could be say -40dBFS to 0dBFS in 5dB steps, where 0dBFS doesn't cause gross clipping. You may also want to adjust Gain 1 to min level (or near min) to lower the noise floor, and play with a few adjustments that lower the noise floor, such as FFT length and averaging. I would expect to see a rising level of 2HD% as the level approaches 0dBFS. Each sweep of tests can take a few minutes, so watch the progress bar to confirm all is proceeding.

Hi trobbins

It's really a tricky thing to find out the sweet point of UMC202HD distortion settings.

I followed your suggestions , using RTA and Signal Generator function in REW, I found that the GAIN1 of UMC202HD should be set to higher levels, and the OUTPUT of UMC202HD should be set to mid-low levels.

Like this:

THD down to 0.032%, THD+N down to 0.70%

Increase signal level to -5.50 dBFS, THD up to 0.036%, but THD+N down to 0.67%.

So which one should be more important for now, THD or THD+N ?

 

[trobbins]

My suggestion would be to re-read post #48 and see if you can do a suite of automated test measurements that show how the 2nd harmonic % varies with signal level and other important settings. Imho that should help characterise what the device can do (compared to just two spot measurements). An example plot of step level changes is in post #3 in https://www.audiosciencereview.com/...s/tascam-us-2x2hr-loopback-measurments.29212/ but note that imho only H2 is initially of interest to benchmark as that is relatively high and needs to be confirmed as to whether it is always that high (in %) or whether it has a region where it is relatively low (ie. a sweet spot). And even if it is relatively high, then the next query is whether it can be nulled to a much lower level using 'add harmonic distortion' feature, or it can't be as jschwender has noted.