Amplifier Gain Varies with Frequency

[gni]

I measured the gain of several amplifiers and came up with some
numbers that just don't seem to add up. Shouldn't the gain be
close to the same at all frequencies? Is it possible that my
DVM doesn't read the correct voltages at higher frequencies?
Or is it due to the fact that some of the amps have zobels and
others have inductor/resistor circuit in the output?

Rotel RB-850
26.40dB @ 60Hz
26.81dB @ 1,001Hz
31.25dB @ 9,999Hz
36.34dB @ 9,999Hz
44.91dB @ 15,9999Hz

Yamaha M-45
27.99dB @ 60Hz
28.31dB @ 1,001Hz
40.30dB @ 9,999Hz
52.40dB @ 15,999Hz

Accurian Amp w/o preamp
14.30dB @ 60Hz
14.58dB @ 1,001Hz
22.78dB @ 9,999Hz
17.25dB @ 15,999Hz

Radio Shack wireless amp
27.1dB @ 60Hz
23.9dB @ 1,001Hz
39.1dB @ 9,999Hz
50.8dB @ 15,999Hz

[Donk]

Hi,

DVM gives correct RMS values only at about 10...200Hz.
Use an o'scope or an audio-band rated voltmeter instead.

Regards,

Donk

[AndrewT]

Hi,
you cannot use an ordinary DMM to measure frequency reponse.

I use a switched attenuator between the source signal generator and the amplifier. I set the attenuator loss to exactly equal the amplifier gain.
This makes the input signal to the attenuator exactly equal to the output of the amplifier.
Now set the attenuator to X+1db. i.e. from -30db to -29db.
adjust the generator frequency until the amp output = attenuator input.
at this frquency the amp response is 1db down from it's nominal gain.
This can be done at many different attenuations eg -0.1db, -0.5db, -1db, -3db etc).
Due to comparing voltages that are identical at each end of the testing set-up the accuracy of the DMM is eliminated from the measurement.

My switched attenuator does -0.0db to -60.05db in steps of 0.05db using 12 DPDT switches and allows frequency response measurements from 0.6Hz to 500kHz. The lower limit is set by my frequency counter and the upper limit is set by my DMM, but I generally never go above 200kHz.

[gni]

In a couple of minutes I would like to follow up on AndrewT's talk about
frequency response. I think my question has been answered to the
point that the gain measurements are inaccurate when I used the
DVM above its frequency limit. Could I say then that the gain results
at 60Hz are fairly accurate?

I would like to get a better voltmeter. . . .maybe Santa will drop one
down the chimney this year.

I really don't understand AndrewT's circuit for measurement of
frequency response.

[Conrad Hoffman]

Most DVMs are very limited in frequency response. Since you seem to have a signal source to feed the amp, why not feed the meter directly and see how good (or bad) it is?

Andrews method is almost foolproof as long as the meter has linear amplitude response at the frequency extremes, which is very likely. Using null meters and comparison techniques, combined with accurate voltage dividers and ratio transformers, was the mainstay of calibration labs and people who needed super accurate measurements for decades. These days test equipment is direct reading and, if you buy the right stuff, will cover any reasonable frequency range, but the divider technique is still the gold standard.

[AndrewT]

Let's assume your test amp has a gain of exactly 31.3db.
attach an attenuator to the front end of the amp which reduces the test voltage by exactly 31.3db.
The signal at the input to the attenuator is exactly the same as the signal at the output of the amplifier.
Your inaccurate DMM will read the same voltage at both locations.
Now reset the attenuator to -28.3db. At the same frequency the amplifier will now read too high by about 40%. As you increase the test frequency the test voltage will hold steady and then start to fall.
When the input voltage =output voltage the gain of the amplifier has fallen to match the attenuation of the attenuator. ie. the gain =28.3db.
This is -3db compared to the start voltage. Read off the frequency and you have the F-3db frequency. The inaccuracy of the DMM does not affect this reading.
The accuracy of the switched attenuator is critical to this method.

Conrad,
the generator frequency reponse errors are also cancelled by this method.

[jcx]

I wouldn't bet on cheap a DVM even being linear at higher AC frequencies

If you've got a pc you've likely got a built-in audio codec with way better frequency flatness than a cheap meter

make a few atenuators to reduce the power amp output V to safe levels and use the free download of RMAA for a quick look

http://audio.rightmark.org/products/rmaa.shtml

[MJL21193]

Originally posted by jcx

make a few atenuators to reduce the power amp output V to safe levels and use the free download of RMAA for a quick look

http://audio.rightmark.org/products/rmaa.shtml

Hi jcx,
I'm building an amp here:

http://www.diyaudio.com/forums/showt...05#post1342905

I lack a distortion analyzer. I see the RMAA performs a distortion test (among others). Would this be accurate enough for an indication of my amps performance?
I have a fairly decent sound card in my lab computer.

Thanks.

[jcx]

loop-back testing gives an indication of the soundcard's limits

you should be able to make basic measurements with the free RMAA, down to the noise/distortion limits of the soundcard

but you would be limited to the functionality of RMAA - only a few test options, also sampling frequency/Nyquist prevents soundcards from being useful for amp stability testing

[MJL21193]

Originally posted by jcx

but you would be limited to the functionality of RMAA - only a few test options, also sampling frequency/Nyquist prevents soundcards from being useful for amp stability testing

Thanks,
I have a 400MHz scope for looking at stability. The distortion testing seems adequate for some preliminary results. All I'm looking for is if I'm on the right track, and to prove out the sim.