DIY amplifier performance checking - step by step ??
Hi
I am looking for instruction on simple audio amplifier performance testing procedures that a hobbiest could use to objectively compare some of these cheap amp boards available online. Perhaps this has already been documented somewhere - I couldn’t find much.
Basically I want to see which amps perform the best and learn from the experience.
These are the available tools:
Regulated power supplies
Power resistors and electronic dummy loads.
DMMs
Signal generator (eBay cheapie)
Digital storage oscilloscope.
With just these tools, what tests, if any, can be done and how?
I am sure someone has probably asked this question before. Maybe there is a useful text somewhere or a forum post.
I don’t expect all key performance specs can be determined with these basic tools but can things like clean power into loads, clipping behaviour, THD, etc be measured with what I have?
Thanks
Hi
I am looking for instruction on simple audio amplifier performance testing procedures that a hobbiest could use to objectively compare some of these cheap amp boards available online. Perhaps this has already been documented somewhere - I couldn’t find much.
Basically I want to see which amps perform the best and learn from the experience.
These are the available tools:
Regulated power supplies
Power resistors and electronic dummy loads.
DMMs
Signal generator (eBay cheapie)
Digital storage oscilloscope.
With just these tools, what tests, if any, can be done and how?
I am sure someone has probably asked this question before. Maybe there is a useful text somewhere or a forum post.
I don’t expect all key performance specs can be determined with these basic tools but can things like clean power into loads, clipping behaviour, THD, etc be measured with what I have?
Thanks
Last edited:
UCA0202 is pretty good to start with ARTA measurements.
Together with a laptop you get a high quality signal generator and you can measure distortion, noise and frequency response for instance.
Last edited:
Here’s what I do to start with - it generally leads down a rabbit hole measuring the poor amp to death.
1. Power amp up on lab supply with ~100mA current setting, input shorted, and no load. Check it’s able to power up without hitting the current limit.
2. Check output DC voltage with multimeter. I use an old Fluke 87 but really any meter will do. It should be a few millivolts. If it’s excessive, either adjust until it’s close to zero if there’s an adjustment or else work out what’s going on.
3. Check bias conditions with multimeter. I generally check the voltage across the output transistor emitter resistorsto ensure they’re balanced and reasonable. Adjust if there’s an adjustment. If there isn’t, ensure they’re what you designed them to be and if not work out what you stuffed up.
4. Connect amp output to spectrum analyser high impedance input. I use a venerable HP3585B, that does DC-40MHz extremely well. I check the whole 40MHz band for any oscillations, then zero in on DC-20KHz.
5. Measure output noise at 1KHz. My HP3585B literally has a button that does this, and gives you a figure in nV/sqrtHz.
6. Measure passband gain. Connect the input to the spec-an tracking generator and take a plot of the overall passband. I generally wind the span out to find where the band edges are and look for any out-of-band peaking that might point to instability.
7. Connect the amp input to a function generator (I use a HP3325B). Generate some tones and look at the output spectra. The 3325 is really only good for 90dB, so unless the amp is awful it’s not useful for THD measurements, but it’s terribly useful for exciting oscillations and looking at clipping behaviour etc.
8. Connect the amp output to an oscilloscope (I use a Tek TDS340A, which does 100MHz, 500Msps). Generate square waves with the function generator at various amplitudes and frequencies and record the output. This is a terribly good test to see if the amp is misbehaving - as you’ll see ringing and oscillations straight away.
9. Repeat the clipping tests with the CRO.
10. Use a pure sine source (I have a home-brew 1KHz oscillator that’s better than 1ppm THD) to measure THD performance. I use my spec-an with a 1KHz notch filter, then record the harmonic levels. It’s a tedious process, but one that gives really good insight into what’s happening with the amp and is good to around 1ppm.
11. Go back and repeat everything with appropriate load for the amp (generally 8 ohms).
It’s never this straightforward for me. I generally get sidetracked fixing an oscillation or playing with compensation, or some other thing like subtle distortion mechanisms.
I’d like to do other tests, like for example intermods with two sinewaves, but only have one good source. I generally look st slew rate with the square wave testing and use that to reassure myself that the amp will be fast enough to not have bad intermods.
1. Power amp up on lab supply with ~100mA current setting, input shorted, and no load. Check it’s able to power up without hitting the current limit.
2. Check output DC voltage with multimeter. I use an old Fluke 87 but really any meter will do. It should be a few millivolts. If it’s excessive, either adjust until it’s close to zero if there’s an adjustment or else work out what’s going on.
3. Check bias conditions with multimeter. I generally check the voltage across the output transistor emitter resistorsto ensure they’re balanced and reasonable. Adjust if there’s an adjustment. If there isn’t, ensure they’re what you designed them to be and if not work out what you stuffed up.
4. Connect amp output to spectrum analyser high impedance input. I use a venerable HP3585B, that does DC-40MHz extremely well. I check the whole 40MHz band for any oscillations, then zero in on DC-20KHz.
5. Measure output noise at 1KHz. My HP3585B literally has a button that does this, and gives you a figure in nV/sqrtHz.
6. Measure passband gain. Connect the input to the spec-an tracking generator and take a plot of the overall passband. I generally wind the span out to find where the band edges are and look for any out-of-band peaking that might point to instability.
7. Connect the amp input to a function generator (I use a HP3325B). Generate some tones and look at the output spectra. The 3325 is really only good for 90dB, so unless the amp is awful it’s not useful for THD measurements, but it’s terribly useful for exciting oscillations and looking at clipping behaviour etc.
8. Connect the amp output to an oscilloscope (I use a Tek TDS340A, which does 100MHz, 500Msps). Generate square waves with the function generator at various amplitudes and frequencies and record the output. This is a terribly good test to see if the amp is misbehaving - as you’ll see ringing and oscillations straight away.
9. Repeat the clipping tests with the CRO.
10. Use a pure sine source (I have a home-brew 1KHz oscillator that’s better than 1ppm THD) to measure THD performance. I use my spec-an with a 1KHz notch filter, then record the harmonic levels. It’s a tedious process, but one that gives really good insight into what’s happening with the amp and is good to around 1ppm.
11. Go back and repeat everything with appropriate load for the amp (generally 8 ohms).
It’s never this straightforward for me. I generally get sidetracked fixing an oscillation or playing with compensation, or some other thing like subtle distortion mechanisms.
I’d like to do other tests, like for example intermods with two sinewaves, but only have one good source. I generally look st slew rate with the square wave testing and use that to reassure myself that the amp will be fast enough to not have bad intermods.
suzyj, sounds like you are in a much better starting position than Dunk02.
I can suggest that, like Dunk02, you are well advised to seek out a soundcard audio analyser path - as time and software has really brought that tool to the front for so many audio related tests.
I just use free and well supported REW5, but many here happily use other software. That gets you in to all the THD, THD+N, IM type tests and spotting noise floor and oscillatory issues, and with a dynamic range and distortion performance that is only limited by your soundcard and probe interface.
I happily use a slightly modified $1 USB soundcard, but also have a 2nd hand emu 0404 as a benchmark. There is certainly a growing number of equipment paths that can be taken in this area - and perhaps that is a hassle in itself for new starters. My $1 soundcard is for valve amp testing in case the 10:1 or 100:1 probe lets through something bad - so not a risk as compared to over-voltage glitching a venerable vintage spectrum analyser, and pretty much equals/exceeds that devices dynamic range.
I can suggest that, like Dunk02, you are well advised to seek out a soundcard audio analyser path - as time and software has really brought that tool to the front for so many audio related tests.
I just use free and well supported REW5, but many here happily use other software. That gets you in to all the THD, THD+N, IM type tests and spotting noise floor and oscillatory issues, and with a dynamic range and distortion performance that is only limited by your soundcard and probe interface.
I happily use a slightly modified $1 USB soundcard, but also have a 2nd hand emu 0404 as a benchmark. There is certainly a growing number of equipment paths that can be taken in this area - and perhaps that is a hassle in itself for new starters. My $1 soundcard is for valve amp testing in case the 10:1 or 100:1 probe lets through something bad - so not a risk as compared to over-voltage glitching a venerable vintage spectrum analyser, and pretty much equals/exceeds that devices dynamic range.
Thanks for the advice - much appreciated . I am going to get an emc8000 mic (or parts express equivalent) and usb audio interface for kicks as I also want to test the Helium micro monitors I recently built. I have fli ked through the Arta handbook and, while I have a technical baclground, this may be one complex hobby too many for me.
All I really want to do is estimate how far the amp volume can be turned up before detectable clipping happens or signifucant distortion is introduced by the amp being over driven, or speaker damage is incurred.
If have seen vidoes where an input sine wave is monitored using a scope and the waveform is observed to distort when the volume is excessive for a 4/6/8 ohm resistive load, for example.
I don’t have a good ear for sound quality so just want some simple electronic tests (if there are any) so I may compare some chip amp boards myself. I have read many reviews of these boards but am inquisitve and want to do some testing myself.
All I really want to do is estimate how far the amp volume can be turned up before detectable clipping happens or signifucant distortion is introduced by the amp being over driven, or speaker damage is incurred.
If have seen vidoes where an input sine wave is monitored using a scope and the waveform is observed to distort when the volume is excessive for a 4/6/8 ohm resistive load, for example.
I don’t have a good ear for sound quality so just want some simple electronic tests (if there are any) so I may compare some chip amp boards myself. I have read many reviews of these boards but am inquisitve and want to do some testing myself.
Last edited:
Here is the cookbook...
Dunk02,
Read through this and it will give you some guidance about
how to actually perform the tests; (i.e., it will give you answers
to your questions).
I came across this by accident when looking for some of the
old US standards that were implemented in the 50s and 60s
(I think).
No, you don't even need the exact gear that is in the cookbook,
which explains how to set them up, take the measurements,
along with how they should look and finally how you should
document them.
You mentioned that you have a technical background so
you should have some fun while doing the tests it describes.
Tektronix Cookbook of standard Audio Tests: LINK
Dunk02,
Read through this and it will give you some guidance about
how to actually perform the tests; (i.e., it will give you answers
to your questions).
I came across this by accident when looking for some of the
old US standards that were implemented in the 50s and 60s
(I think).
No, you don't even need the exact gear that is in the cookbook,
which explains how to set them up, take the measurements,
along with how they should look and finally how you should
document them.
You mentioned that you have a technical background so
you should have some fun while doing the tests it describes.
Tektronix Cookbook of standard Audio Tests: LINK
Yep, there is some good stuff here in Ye Olde Cookbook.
I have been trying to establish a test method using either a test tone from a cheap signal generator, or I have used downloaded test WAV files from JohnAudioTech on Youtube. Having some trouble setting it up at the moment (not much time). The Siglent SGS1102CNL FTT function is tricky to setup.
To be honest I don’t fully understand dB yet.
I think power is calculated from the DSO sine wave Vrms^2 divided by the load impedence, just before the input signal clips.
The unmodified TDA7297 amp seems to start clipping at about 25-30% of volume rotation when loading only one channel with a resistive load. This seems low but I think I need to measure the volume where 1% distortion occurs when both outputs are loaded, for clean power measurement. Waiting on another 4 ohm resistor (my diy bookshelf speakers have ND91 4” drivers) to arrive so I can connect both channels up together.
Will post some pictures if I get anything sensible out of this.
I have been trying to establish a test method using either a test tone from a cheap signal generator, or I have used downloaded test WAV files from JohnAudioTech on Youtube. Having some trouble setting it up at the moment (not much time). The Siglent SGS1102CNL FTT function is tricky to setup.
To be honest I don’t fully understand dB yet.
I think power is calculated from the DSO sine wave Vrms^2 divided by the load impedence, just before the input signal clips.
The unmodified TDA7297 amp seems to start clipping at about 25-30% of volume rotation when loading only one channel with a resistive load. This seems low but I think I need to measure the volume where 1% distortion occurs when both outputs are loaded, for clean power measurement. Waiting on another 4 ohm resistor (my diy bookshelf speakers have ND91 4” drivers) to arrive so I can connect both channels up together.
Will post some pictures if I get anything sensible out of this.
*Driven with what??* Put a ribbon mike in, it won't overload with the knob at 100%. Put in a 25V H-P 200AB sig-gen, it will overload at 2%. Yes, DAC outputs are semi-standardized and most commercial trax hit the max. But gain is often scaled for less-intense analog sources, and 25% turn is probably right.
Should make very little difference unless your power supply is really saggy. Perhaps 20% on power, which is not the difference between
and .
So you aren't measuring performance then.
To just check performance, I'd just play it, listen, and feel/smell for trouble.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.