Help wanted recording the outputs of the DAC and Headphone Amplifier

[Roland68]

I would like to digitally record the analog output of DACs and headphone amplifiers as high-resolution as possible.
Unfortunately, I have little to no knowledge of it.

What do I need as hardware?
I have a Macbook and a ZOOM TAC-2R Thunderbolt Audio Converter. Is that enough or do I need a special A / D converter for it?
What software do I use to record the audio signal?
What do I have to watch out for when recording?
Questions upon questions and I am grateful for any help.

 

[Helicopter]

Yes you can use an audio interface and a laptop to measure a headphone amp output. You may need to attenuate the amp so you don't fry your inputs. You may need to make some cables. Of course you will be limited by the ADC performance so if you're measuring SINAD, the interface will work better for a tube amp with 77dB SINAD than a SOTA amp with 120dB. You can start with a loopback to measure your interface, cables, and any external attenuator to get a baseline for the performance capability of the setup.

 

[AnalogSteph]

Apparently the TAC-2R is much the same as the UAC-2 (just TB instead of USB), and the measurements I found of the latter were quite uninspiring... sure the dynamic range is mostly there but distortion-wise it's not good, I guess there are some crummy ceramic capacitors or really bad low-power opamps in the signal path. Enough to get your feet wet, I suppose, but hardly measurement grade. It should be good enough to determine the noise floor of just about any DAC out there when using the mic input at high gain though, you just need to calibrate your levels e.g. by generating a low-level sine (e.g. 1 kHz @ -60 dBFS).

I must say I am not familiar with measurement software options on the Mac. REW would work, I mean it's written in Java and all, and you can always use your favorite audio editor or DAW to record (of which there should be no shortage, given how popular Apple gear is among artists).

 

[Helicopter]

Quant Assylum QA402 or Focusrite Scarlett would have better ADC. With QA402 you would just want a dummy load for headphone amp. With Scarlett you may want an attenuator too.

 

[Roland68]

I don't want to get started with measurements of devices, headphones and speakers until later.

My current goal is to record pieces of music at the exits.
I would like to archive the audible current state of devices in order to understand changes to my DIY devices, but also to compare them with devices that I have already handed in.

For this it would be necessary to create a workaround that can no longer be changed, which digitizes these recordings in the highest possible resolution.
It is clear that a current recording must always be made in order to compare, as only the recordings allow a comparison with each other.

Yes you can use an audio interface and a laptop to measure a headphone amp output. You may need to attenuate the amp so you don't fry your inputs. You may need to make some cables.
Quant Assylum QA402 or Focusrite Scarlett would have better ADC. With QA402 you would just want a dummy load for headphone amp. With Scarlett you may want an attenuator too.

Making the appropriate cables is no problem.
What do I have to do in order not to burn the devices down and weaken the signal? Just 32 ohm resistors in the signal path?

Apparently the TAC-2R is much the same as the UAC-2 (just TB instead of USB), and the measurements I found of the latter were quite uninspiring... sure the dynamic range is mostly there but distortion-wise it's not good.
You just need to calibrate your levels e.g. by generating a low-level sine (e.g. 1 kHz @ -60 dBFS).

I must say I am not familiar with measurement software options on the Mac. REW would work, I mean it's written in Java and all, and you can always use your favorite audio editor or DAW to record .

I was afraid that the TAC-2R would not be enough for that.
Is there something affordable to digitize the output in a sufficiently high resolution? Or do I have to look around in the pro audio area?
The Focusrite Scarlett is already very cheap.

What software can I use for recording? Audacity? Pro Tools First? or is there something better?

 

[Helicopter]

What do I have to do in order not to burn the devices down and weaken the signal? Just 32 ohm resistors in the signal path?

A voltage divider with 2 or more resistors per channel is better.

Take a look at My thread where I got advice on making a speaker attenuator. You could do something similar for headphone amps. The circuit and resistor values will depend on the attenuation you want. Actually you could probably just get a 2i2 and copy my box and be fine with the 2i3 volume pots into mic inputs. With the right cables my attenuator would work fine for measuring a speaker amp or a headphone amp that had outputs with speaker connectors, TRS, TRRS, XLR4, etc. If you are measuring lone level gear, you can set the 2i2 to line level and go straight into the inputs.

I couldn't find a better ADC implementation for this than the 2i2 at the price.

 

[AnalogSteph]

I don't want to get started with measurements of devices, headphones and speakers until later.

My current goal is to record pieces of music at the exits.
I would like to archive the audible current state of devices in order to understand changes to my DIY devices, but also to compare them with devices that I have already handed in.

What you want may not be what you need. You'll be making it very hard on yourself (to say the least) if you try to evaluate the performance of electronics by ear, actually it's a good way of driving yourself crazy. Even advanced null testing software like pkane's DeltaWave still has lots of quirks. Dedicated test signals exist precisely because of a need to tease out the weaknesses easily, reliably and with good accuracy even beyond the limits of human hearing. Engineers typically don't enjoy spending time in mental institutions, in fact they - or their employers - are even willing to pony up for rather expensive tools to take automated measurements for faster progress (case in point, the Audio Precision analyzers). At the same time, devices "designed by ear" regularly tend to contain glaring faults that their designers failed to pick up on yet are glaringly obvious in a basic measurement suite.

You'd be doing yourself a massive favor if you were to record not only music samples but also e.g. a complete RMAA 6.4.5 test signal matching the recording sample rate. RMAA is Windows-only but can do offline analysis as well (saving and loading WAV files), so you just need access to a Windows (virtual) machine. If you want, I could save a range of test tones (up to 32 bit, 384 kHz) and resample them for use at various lower playback sample rates - it's not really difficult to do, more tedious. Do you have a player that'll accept 32-bit int samples and FLAC? I'd have to install an encoder for ALAC first but this would also be feasible.

As far as the hardware side goes, do you have any specific budget in mind? You may want to keep an eye on the neighboring "best ADC" thread and fellow member @IVX's Cosmos ADC release.

 

[Roland68]

What you want may not be what you need. You'll be making it very hard on yourself (to say the least) if you try to evaluate the performance of electronics by ear, actually it's a good way of driving yourself crazy.

You'd be doing yourself a massive favor if you were to record not only music samples but also e.g. a complete RMAA 6.4.5 test signal matching the recording sample rate. RMAA is Windows-only but can do offline analysis as well (saving and loading WAV files), so you just need access to a Windows (virtual) machine. If you want, I could save a range of test tones (up to 32 bit, 384 kHz) and resample them for use at various lower playback sample rates - it's not really difficult to do, more tedious. Do you have a player that'll accept 32-bit int samples and FLAC? I'd have to install an encoder for ALAC first but this would also be feasible.

As far as the hardware side goes, do you have any specific budget in mind? You may want to keep an eye on the neighboring "best ADC" thread and fellow member @IVX's Cosmos ADC release.

Yes, unfortunately, that's exactly what I need. However, I've been doing this for 30 years and I'm far from going insane because of it (there are much worse things; o)
So far, I have left the measurement to others who understand more about it.
I use pieces of music that I've already heard live and that were recorded with the same musicians and instruments in the studio. Or unplugged recordings that I was there myself.
I've done this before, but I only really realized it when I came into contact with musicians with some very well-known instruments 20 years ago. These musicians complained that they did not recognize their own (acoustic) instrument even on expensive hi-fi systems. That was very interesting and insightful back then, especially when a musician played his recording on CD and then played the same piece again live on his instrument.
This is also what concerns me when I listen to devices to see whether the instruments and voices are reproduced as I know them.

Thanks for the offer to create test files for me, if I start with measurements, I'll be happy to come back to them. Player for 32-bit int samples and FLAC is no problem.
I don't have a Windows PC, but I can borrow a high-performance notebook from our company, if necessary a large HP workstation.

Budget, preferably as little as possible and as many as necessary.
But I've already seen that if I want to achieve more than with a normal audio interface, then I'm in the four-digit range.
I will first test how far I can get and whether the whole thing can be implemented in a promising manner.

 

[Roland68]

I looked at the circuit in the other thread (Measuring Speaker and Headphone Outputs With Focusrite Scarlet 2i2 v3) and have a few questions about it.

- I understood correctly that the XLR / microphone input is the right one for recording / measuring?
- When connected to amplifier outputs, SE headphone amplifier output and RCA outputs of DACs, the two OUT- would be brought together in front of the resistors. Is this a problem?
- Do I have to significantly reduce the resistance values for recording / measuring DACs and preamplifiers? Output voltage via XLR is max. 6 VRMS.
- Do I need an additional load on the outputs when recording with headphone amplifiers?
- The XLR ground (Pin1) of the outputs is never connected to the XLR ground of the microphone inputs in this setup?

I will make the first recordings only once with the Zoom TAC-2R. If that is successful I will get myself a Focusrite Scarlet 2i2 v3 and an RME ADI-2 Pro FS R to test.
Is there an important difference between the Focusrite Scarlet 2i2 v3 and 4i4 V3 for the recordings / measurements?

 

[Helicopter]

My device takes speaker level signal into a SpeakOn connector and does 3 things with it: sends it (almost) full strength to an amp or dummy load, send it (almost) full strength to a multimeter, and sends it to a resistor circuit that drops the voltage / attenuates it, and sends a line level signal out via XLR. I further attenuate the signal at the 2i2 input with the potentiometer. I start at min volume, turn it up until clipping lights come on, then back it off a bit. I can measure the level with the multimeter and a 60Hz tone or whatever.

The 2i2 has good specs, especially for the price. Just depends what you want to measure. SINAD is ample for anything phono, virtually anything tube, and a lot of SS stuff like old amps or AVRs. It is nowhere near good enough to test SOTA DACs, amps, headphone amps, etc... but you basically need an AP if you want to measure a Topping DAC or something like that.

 

[AnalogSteph]

I looked at the circuit in the other thread (Measuring Speaker and Headphone Outputs With Focusrite Scarlet 2i2 v3) and have a few questions about it.

- I understood correctly that the XLR / microphone input is the right one for recording / measuring?

That'll depend on the interface in question. With some the mic inputs also have the best all-round performance, while for some others it's the line inputs that perform better (the Tascam US-2x2HR is one example I can think of... its mic inputs have super low noise but distortion rises quite early - see Julian Krause's review).

- When connected to amplifier outputs, SE headphone amplifier output and RCA outputs of DACs, the two OUT- would be brought together in front of the resistors. Is this a problem?

Nope.

- Do I have to significantly reduce the resistance values for recording / measuring DACs and preamplifiers? Output voltage via XLR is max. 6 VRMS.

You would definitely have to pick resistor values matching your input levels, yes. The example values given in the other thread were for something like 200 W / 8 ohm max, that's 40 Vrms. (For 6 Vrms, you might try e.g. R1 = 3.32k, R2 = 1.47k-1.5k. 2.61k / 1k should work alright, too. About 15 dB with a 2i2, input impedance >6 kOhms, output impedance <1.5 kOhms.)
Note that for attenuation less than 10-12 dB, reconciling a decently high input impedance with a decently low source impedance tends to be difficult.
Minimum distortion in mic input stages tends to be reached at a medium gain setting... near minimum, common-mode distortion tends to degrade performance somewhat, while higher up there is less excess gain-bandwidth available to keep distortion in check. Not sure where this point is for e.g. the 2i2 gen3. For a Behringer mixer it turned out to be in the 20-30 dB range, while for a Mackie the sweet spot basically extended all the way up to maximum gain as circuitry following was the limiting factor, not the preamp itself (though for obvious reasons, you don't want max gain either as it brings up the input noise floor way too much, limiting dynamic range at the bottom end).

- Do I need an additional load on the outputs when recording with headphone amplifiers?

In general, yes - otherwise all you're testing is their ability to drive your measurement attenuator. It doesn't have to be quite as fancy as the one Amir is using (Neurochrome HP-LOAD, reviewed recently).
.

- The XLR ground (Pin1) of the outputs is never connected to the XLR ground of the microphone inputs in this setup?

Both are already connected through the audio interface. You could no doubt include a jumper or something to make an optional connection given a DUT with balanced output, but be warned that this may expose Pin 1 Problem type issues in both the DUT and the measurement interface's inputs equally (and I wouldn't rule out that these might occur in a budget interface).

 

[AnalogSteph]

For RMAA test signals 24-44 to 24/192, see here.

 

[Roland68]

@Helicopter
@AnalogSteph
Thank you for your help.
I have now got myself a 3rd generation Focusrite audio interface and I am quite excited about it.
In the next few weeks I will start to create a standard for recording and also to take measurements with REW.