How to - Distortion Measurements with REW

What result can be expected when measuring a low power SET amplifier in FFT mode?

I finally set up a rig for measuring amplifier distorsion i FFT-mode, using a Focusrite Scarlet Solo (3rd edition) and REW.
The loopback is shown below. From what I understand, it looks as it is suposed to do ? (pls. comment if not).
2A3 loop.jpg


The DUT in this measure is a regular SET amplifier with 6072A input stage i SRPP and a 2A3 output in single end. Filamets are AC powered with ballancing pots, tuned to minimal (not audiable) hum. Output signal looks good on a scope, and visibal clipping occurs earyest at apout 3,5-4W. The sound is good with no apparent distorsion, so I would expect it to measure quite well (AC heaters and B+ hum taken into account).

However. The distorsion spectra in FFT-mode shows quite some 50Hs distorsion peaks (and 2nd, 3rd, 4th, etc overtones of it..) contaminating the hole audioband. The picture below shows the amplifier commected to the test-rig. On but with no stimuli signal put to it.
Clearly those annying 50Hz mulitpples are shown, even when no stimuli signal to the amplifier.
2A3 0W.jpg


When stimuli-signal is put on (1kHz signal from Focusrite output) the 50Hz peaks modulates the 1KHz signal, making the whole measure quite hard to interprete. In this case the output is 1W over 8 Ohm load, so the total harmonic distorsion seems exxesivelly high (see below).
My question is. Is this what I can expect from a FFT-measure of at low power SET amplifier, or have I made someting wrong in this measure?
2A3 1W.jpg
 
My question is. Is this what I can expect from a FFT-measure of at low power SET amplifier, or have I made someting wrong in this measure?
I remember there was a tick box in the FFT mode that would take the measurements only after the fundamental, I don't know maybe it was another program!

BTW your measurements seperate THD from noise and it is detecting fundamental as number one. So it seems PSU noise is high and more filtering is needed. About the THD I think 1.18% is a not bad result for a 2A3 SET at 1W, and it's nearly all 2nd order. I think it's the main characteristic of SET sound.
 
Last edited:
What is your amp’s PSU setup like? What you see is typical but could be cleaner - look for radiated mains EMI contaminating your high impedance front end. Don’t place linear trafo so close to amp board. Use proper grounding strap from chassis heatsink to star earth ground connected to IEC ground. Use a ground loop breaker 10ohm NTC or resistor between chassis ground and amp analog 0v ground.

50Hz peak should be in -80dB range or lower else hum will be bad.

The SET FFT looks fine.
 
Component layout, power supply layout/ground scheme, wire layout/loop area affect level of noise.

Hard to say with this amplifier without seeing it and its construction details.

As others have commented the audio signal FFT looks in the ballpark for a 2A3 SET amplifier.
 
Thanks xrk
It’s a regular PSU (c-l-c filter). Nothing fancy, but surely possible to make better.
Your advice to make it below -80dB was a helpful limit to aim for.
What’s most important for me in this stage, is the conclusion that the measures fft values was to rely on.
The next step will be to improve the topology.

I’m understand that it’s no easy question to answer, but I keep wondering how much of those 50Hz peaks coming from ripple at B+. And how much coming from AC filament current.
But that an easy test for me to do attaching dc sources.
 
Yes, These measures were taken with my laptop and soundcard runned on batteries. Only the DUT were connected to the mains.

Thank you all for your replies 🙂
Perhaps one more ground loop check would be related to the amp itself and if the signal input gnd to the amp is not the same node as the output signal gnd that is being probed and measured.
 
Ripple is 100Hz.

50Hz from the AC filament is possible. 50Hz is also possible from EMI from AC line and power transformer. Layout and location of AC line and transformer may be factors. There are also 100Hz noise and harmonics that are high.

Views of the amplifier and including its under chassis may offer some ideas.
 
  • Like
Reactions: sajadho1988
The inside is not the best looking one..
I’ve inherited this amp as as half completed build. Just finishing what’s missing. So I have not every detail in it.
The ground rail is running from iec-input to all signal grounds. What’s a bit odd is the startup timer board, connected on the ground side.
The rca inputs are on the front. And PSU in the back.
 

Attachments

  • IMG_6847.jpeg
    IMG_6847.jpeg
    817.1 KB · Views: 91
  • IMG_7510.jpeg
    IMG_7510.jpeg
    702 KB · Views: 95
You want to separate earth ground (dirty) from analog ground (clean) using something like CL-60 or 8D-20 NTC. Connect earth ground from IEC shell to chassis metal star point. Connect NTC between chassis star to clean ground bus bar. Ensure that RCA ground is not connected to chassis directly (this often happens with cheap non-isolated connectors). Get isolated ones with bushings or polymer bodies to ensure signal input ground does not to touch chassis directly. This will be major source of hum and ground loops.

1738497105018.jpeg
 
  • Like
Reactions: DIYHarry