First measurements of my headphone setup using REW- Are they accurate?

[Gitfiddle]

I finally got around to taking some measurements in REW and decided to try and measure my headphone setup. I am curious if my methods are accurate at all. If they are not, some tips would be nice on how to improve them. I assume an actual audio interface and multimeter would be recommended?

My loop:

Pc -> powered external USB hub -> Moondrop dawn pro in NOS mode 3.5mm out max volume (should be 2vrm according to specs) -> liquid spark amp -> pc Realtek input

I used REW's soundcard calibration at this point. Attached below are the measurements generated from that named "dawn pro.zip"


Here is a 1k sine that I generated. Unfortunately I could not find a corresponding REW file. I must not have saved it.

I'm mainly curious what is going on at 2k and beyond. I assume a mix of things. NOS, user error, poor measurement input, and other errors that are probably my fault. I'm also fine if this is accurate too, since I am very much enjoying the music.

 

[Keith_W]

Your topic title is very confusing. You said you are measuring your "headphone setup" but your REW measurements look like a DAC of some kind, and the measurement you posted looks like an extremely poor performing DAC with a heck of a lot of harmonic distortion. All those spikes you see labelled 2, 3, 4, etc. are harmonic distortion.

If this is a DAC measurement, then it is important that the performance of your ADC exceeds that of your DAC, otherwise you will be measuring the performance of your ADC instead of your DAC!

If you want to measure headphones, you will need binaural microphones, or a headphone test fixture like MiniDSP EARS. But if you are measuring a DAC, you will need a very low noise ADC like this.

 

[DVDdoug]

Noise and distortion at -60dB isn't great but it's not terrible. It looks like harmonic distortion which isn't as bad as noise. Distortion tends to be masked (drowned out) by the signal and it's (somewhat) proportional to the signal. Noise tends to be constant and it becomes audible when the signal goes quiet or silent.

pc Realtek input

Is that a line input (usually blue on a soundcard). Most laptops don't have line-in.

A microphone input (millivolts) on a laptop or soundcard is not compatible with line level signals (about 1V) and the mic preamp is usually low-quality and noisy.

If you aren't hearing noise (hum, hiss, or whine in the background) and if your headphones go loud enough, whatever you have is probably OK.

 

[Gitfiddle]

A microphone input (millivolts) on a laptop or soundcard is not compatible with line level signals (about 1V) and the mic preamp is usually low-quality and noisy.

If you aren't hearing noise (hum, hiss, or whine in the background) and if your headphones go loud enough, whatever you have is probably OK.

This is precisely the information I needed. That means the measurements are flawed from using the mic input. Thanks.

Your topic title is very confusing.

Apologies for presenting it in a confusing way. The information you and doug provided is still enough to give me all I needed to know. Thank you.

 

[Sokel]

Depends at what you're measuring.
Use exclusive I/O when you can, check levels so you don't clip either inputs or outputs, keep your rig clean.

You don't need to pre-generate the test signals, REW has a generator which you can configure any way you like and watch results in real time.
Use at least 4 averages, turn off the lock as REW tells you.

What we see there seems to be near clipping, or a lot of things have gone wrong.

 

[Sokel]

Example:

 

[Gitfiddle]

What we see there seems to be near clipping, or a lot of things have gone wrong.

Considering what information doug shared, it seems to simply be a case of me making a novice mistake and using the wrong input on the soundcard.

You don't need to pre-generate the test signals, REW has a generator which you can configure any way you like and watch results in real time.
Use at least 4 averages, turn off the lock as REW tells you.

Yes I have tinkered around with the generator and RTA window, thanks for the tips. This will help me more properly use it in the future.

Example:

Thanks for this as well. Seeing the settings you use is very helpful.

 

[Gitfiddle]

Is that a line input (usually blue on a soundcard). Most laptops don't have line-in.

After properly switching to line input the distortion after 1khz is much lower. User error, case closed.

If this is a DAC measurement, then it is important that the performance of your ADC exceeds that of your DAC, otherwise you will be measuring the performance of your ADC instead of your DAC!

I tend to think this is the case here along with the input mistake. I'll take a look at ADC's such as the one you linked in the future If curiosity gets the better of me.

Thanks for the help, everyone that replied.