Software (REW) can correct for soundcard induced distortion in a loopback configuration whereby the harmonics are suppressed down to the noise floor achieved by the test configuration. The harmonic suppression can be done manually across the typical range of harmonics, or can be an automated process (a lot easier!). REW can assess the noise floor behaviour, as noise floor can be interpreted and described in different ways, and 'high' doesn't really mean anything.
thanks trobbins. I must look into this further. I did some kind of sound card calibration through REW. However, I assumed this was just to set the level. Does that calibration also correct the noise floor/distortion error as well? I've been reading the lengthy documents. I've probably missed the error correction calibration file.
I must have done something wrong. After calibration of the sound card, I open RTA and loopback the sound card to take a 1khz reading. I still see high 2nd and 3rd order harmonics. I'd assume the error correction file would zero those out.
I'm looking at page 271 of the REW document.
I'm looking at page 271 of the REW document.
The calibration process is a loopback test that derives a cal file to normalise the frequency response for constant gain and phase across the bandwidth selected - ie. to give a 'flat dc to daylight' frequency response. That is not related to the harmonic suppression process, or related to how the noise floor can be assessed.
The harmonic suppression process is just for a single test sinewave frequency and signal level, and the supression outcome will vary as the signal frequency or magnitude is varied. So this is not related to a step related sweep or other process unless each step is done separately and re-calibrated for harmonic null.
Noise floor is very dependent on the processing set up, such as the fft process window and config, and the averaging mechanism used. For sinewave plotting, coherent averaging is a further processing technique to suppress noise between harmonics. This is all quite a detailed and technically imposing topic, and can take quite some learning curve to better appreciate.
The harmonic suppression process is just for a single test sinewave frequency and signal level, and the supression outcome will vary as the signal frequency or magnitude is varied. So this is not related to a step related sweep or other process unless each step is done separately and re-calibrated for harmonic null.
Noise floor is very dependent on the processing set up, such as the fft process window and config, and the averaging mechanism used. For sinewave plotting, coherent averaging is a further processing technique to suppress noise between harmonics. This is all quite a detailed and technically imposing topic, and can take quite some learning curve to better appreciate.
Perhaps look at response when signal level is varied. Keep reading the help topics and gaining experience as settings are changed.
Greetings, friends. I've been setting up my testing rig and had a couple Q's. I'll be testing tube power amps, mainly <5w SE amps and a 10w PP amp. Dummy load is Thick Film resistors wired to give 4- and 8-ohm loads. I'll be using a Pyle ASIO interface.
Should I build a XLR probe? Way back in Post 1 we were presented with single-ended and differential input rigs. Forgive if this has already been discussed, but since I'm not testing a BTL amp, would I see any benefit from the differential probe? As I didn't have any suitable plugs on hand for that, I have built a SE probe with a 22:1 voltage divider and gotten usable data from REW.
Are there standards for testing tube amps? Measuring at 1w output may give a certain characteristic, but since distortion is proportional to input and tube amp distortion is sought after, should I be testing with a higher input voltage to capture the exaggerated distortion profile?
Should I build a XLR probe? Way back in Post 1 we were presented with single-ended and differential input rigs. Forgive if this has already been discussed, but since I'm not testing a BTL amp, would I see any benefit from the differential probe? As I didn't have any suitable plugs on hand for that, I have built a SE probe with a 22:1 voltage divider and gotten usable data from REW.
Are there standards for testing tube amps? Measuring at 1w output may give a certain characteristic, but since distortion is proportional to input and tube amp distortion is sought after, should I be testing with a higher input voltage to capture the exaggerated distortion profile?
Here's a shot of my EL84 Triode amp. It makes about 1.2w with a 1vrms input. I set the output to 2.83v/1w. 36% 2nd Harmonic.
Old tubes? Is the amp working perfect?
37% distortion is maybe on the high side.. 2-3% 2nd harmonic is not hurting at all. But this was actually a little surprise.
37% distortion is maybe on the high side.. 2-3% 2nd harmonic is not hurting at all. But this was actually a little surprise.
I guess they have about a year of casual listening on em. Come to think of it, these are 6P15P tubes I got from Ukraine, with a 12AU7EH driver tube. The amp has a switch for Triode or U/L and the 2nd harmonic distortion doubles when the amp is measured in U/L mode. Amp makes about 1w in Triode mode and about 1.2w in U/L, with a 1vrms input. I guess I outta run some tests on 6P14P and EL84 tubes as well.
w
w
You sure it isnt a case of yeah, 36% - but at 20 Hz? Much less at "normal" listening frequencies...
I would think distortion is a level dependent thing. Traditionally done at full rated output because it's assumed to be less at "normal" listening levels. What would be great is a 3D plot with level, frequency and distortion. Then you can see what the distortion does with level for the whole audio frequency band - highly non linear, where 99.9% of the time you're nowhere near full rated output, while listening "casually".
I would think distortion is a level dependent thing. Traditionally done at full rated output because it's assumed to be less at "normal" listening levels. What would be great is a 3D plot with level, frequency and distortion. Then you can see what the distortion does with level for the whole audio frequency band - highly non linear, where 99.9% of the time you're nowhere near full rated output, while listening "casually".
Ok, how do I do that? I'm still figuring this stuff out. I ran tests of each channel on the amp, and the numbers all matched up. Tubes are good, or at least measure the same.
Here's the right channel run at -3dbFS
and here's the left channel run at -3dbFS
2nd Harmonic is closer to 11% on each channel with the output set to 2.83v, but the calibration seems to be drifting. Here's a link to all the measurements.
w
Here's the right channel run at -3dbFS
and here's the left channel run at -3dbFS
2nd Harmonic is closer to 11% on each channel with the output set to 2.83v, but the calibration seems to be drifting. Here's a link to all the measurements.
w
- Home
- Design & Build
- Software Tools
- How to - Distortion Measurements with REW