PC audio test set

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I have tried it long time ago (2020-2021 or so) and could not get it to work back then. Don't remember why anymore. I also tried FlexASIO, but did not work either. But your suggestion triggered me to try once again, because in the meantime, both REW and ASIO4ALL saw new releases. And guess what: it works flawlessly now 😃. Thanks for triggering me!

Jan
Perfect! (y)
 
I just got the QA403 and it's an impressive device...still learning it but I really needed to be able to provide clients with accurate numbers as well as establishing benchmarks on modules when servicing a console or tape machine to match channel to channel performance.
I used a motu, Visual Analyser and REW for years with good success but when the motu(firewire rig) finally gave up I went with the QA403 and it's great although I miss the scope functions of Visual Analyser.
 
For an interface I found many relay input selector board with relay step attenuator. Use a stereo card for balanced line. There are many models for around 50$ or so... This could be a good start for building an input interface for the sound card.
The input attenuator is the easy part. However, it may not be that simple depending on t ahectual input impedance of the souncard. If you need to measure low level signals, you may have to use mic inputs, which impedance is rather low.
Then you have to address the issue of generator output. Most souncards struggle to deliver signal high enough to challenge the DUT.
 
Most souncards struggle to deliver signal high enough to challenge the DUT.

This is a common issue indeed. AFAIK even the QA40x can't deliver enough juice for measuring the headroom. Long time ago, when I suggested higher output I almost got attacked by many members accusing me that I sacrifice THD and noisefloor for the sake of amplitude :)

I would love to build asome small amp device, which would balance and amplify the output of a signal generator but I don't know where to start... For the time being I've been using a step-up TX but the limitations are the fixed gain (+6db), the bandwidth and the linearity of course.
 
This is a common issue indeed. AFAIK even the QA40x can't deliver enough juice for measuring the headroom. Long time ago, when I suggested higher output I almost got attacked by many members accusing me that I sacrifice THD and noisefloor for the sake of amplitude :)

I would love to build asome small amp device, which would balance and amplify the output of a signal generator but I don't know where to start... For the time being I've been using a step-up TX but the limitations are the fixed gain (+6db), the bandwidth and the linearity of course.
Yes, max output level is always limited. I have two interfaces that I chose specifically for their max output: Tascam US-2X2HR (+20dBu/7.746Vrms), and a Motu 4Pre, which tops at about 6.5Vrms. Neither, of course, is hot enough when testing pro audio devices, but should be fine for any consumer level thingy. Consumer interfaces usually don't perform will at 0dBFS, just FYI, you'll want to be a couple dB below that.

Of course, the QA401, as great as it is, sticks at 3.76Vrms (differential), +13.7dBu.

So...two things on that:

1. What do you really need to drive that hard? You need to know and define that first, because in practice, there isn't a lot, and as they stand, you can do 80% of everything you need with a US-2X2HR barefoot.
2. What would you be willing to give up for more output level? You will give up something, but it may not matter.

My answer to 1: Transformers. If I want to profile a line-level transformer, I'll need to hit it pretty hard. A 111C, for example...yeah, you really have to punch those.

My answer to 2: I'll give up a bit of noise floor and a tiny bit of distortion for hotter output. But that's also based on what you're planning to measure. Again, transformers, noise isn't often an issue, so getting more level into them with a tad more noise, isn't a problem. Distortion...well, everything I'd use to amp up the analyzer output has far lower distortion than any transformer, so that's not a problem either.

Now, my solution, and this is in progress... I have several cards I designed for broadcast audio use in the early 1980s. They're well made, and with a little modification, will work fine to carry the active parts. And the active parts they were made for were 990 opamps, which in 1980 were the bees kneez. The cards are already configured for them, and I have two new 990s set up for +/- 24V power, which lets them provide a maximum of 27.6Vrms/ +31dBu, differential. The card is set up for enough gain so the interface will push them to max output, or nearly, at 0dBFS. They'll drive 75 ohms, and will drive 150 ohms to 145KHz, with an EIN of -133dBu. And folks, that's good enough for anything I need. Pefect? Nope. AP? No. But until somebody agrees to pay me to make measurements with an AP setup, new, used, or ancient, this rig will be fine.

My solution is basically a cheat, because I had what I already had, I'm lazy, and I don't have as much free time in my old age as I thought I might. I got the new (and actually, improved!) 990s from John Hardy, along with a pair of his "output isolators", and a set of fresh sockets. Though honestly, my old 990s work too, though they are +/-15V designs.
 
I would love to build asome small amp device, which would balance and amplify the output of a signal generator but I don't know where to start... For the time being I've been using a step-up TX but the limitations are the fixed gain (+6db), the bandwidth and the linearity of course.
In my front-end thingy, I use a THAT 1646. It gives +27dBu balanced into 600 ohms with minimal THD. +2.1dBu unbalanced
Not an AP, but enough for most instances.
 
Shouldn’t that be +21dBu unbalanced?
Yes, it's a typo. I'm not abbey though :)

In my front-end thingy, I use a THAT 1646. It gives +27dBu balanced into 600 ohms with minimal THD. +2.1dBu unbalanced
Not an AP, but enough for most instances.

I remember seeing your schematic some time ago and it looked very interesting. Did you ever make any PCBs for this?
 
This is a common issue indeed. AFAIK even the QA40x can't deliver enough juice for measuring the headroom. Long time ago, when I suggested higher output I almost got attacked by many members accusing me that I sacrifice THD and noisefloor for the sake of amplitude :)

I would love to build asome small amp device, which would balance and amplify the output of a signal generator but I don't know where to start... For the time being I've been using a step-up TX but the limitations are the fixed gain (+6db), the bandwidth and the linearity of course.

Attached schematic is what I have in my Audio Interface Frontend / aka DIY Audio Analyzer. It takes the single-ended output signal from the DAC output filter op-amp of the modified UMC202HD that I use as Audio Interface. Output level can be adjusted between 0-1V RMS and 0-10V RMS, single-ended and unloaded. That's +28dBU differential, unloaded. Selectable Single-Ended output impedance of 50 or 150 Ohm. It is short-circuit protected and protected against voltage spikes on the output, which could occur when disconnecting reactive loads under test. Although the op-amps are supposed to be short-circuit proof, I placed polyfuses in the power lines. Probably not needed, but better safe than sorry.

I've used a multiturn potmeter to adjust the output level, but it's inductive and affects the frequency response. Tried to flatten the curve as good as possible with C31, but it's still not perfect. If I were to build one again, I would use an ALPS RK27 potmeter with Vernier reduction.

Jan
 

Attachments

  • AA_Line_Out_Schematic Rev B.PDF
    386.1 KB
A midi controlled relay attenuator for REW would be nice ,
the program has midi functionality built in ,
John Paul Mulcahy might be willing to add something to the official hardware list ,

E1DA Cosmos now produces an autoscaler ,
Im not all together sure why the gain increments are all over the place ,
https://e1dashz.wixsite.com/index/cosmos-scaler
 
I always use the Scarlette line in and turn the gain to minimum and cal just once.

Cheers

Ian
Ian I'd be curious to see if you've found a quieter gain sweet spot with the Scarlett than I have.
I use a slightly different calibration strategy with the Scarlett 2i2 Gen 3.

The normalized 0 dBFS level for almost all 5V analog stage differential input converters is 2V RMS.
In AudioTester I set the FS calibration to "2000 mV," apply 1.00V RMS into the input and adjust the front panel level control for -6dBFS.
This makes the overall gain unity from input connector to A/D modulator input.
It may not be the ultimate sweet spot for noise but I'm able to measure down to about -96 dBu unweighted over a 20 kHZ BW with FS at 96 kHZ.
That's a DR of only 104 dB so for lower level noise measurements I use the PCM4222 which sits around -109 dBu.
The only real issue I have with this calibration strategy is not bumping the knobs.

The Gen 4 version of the Scarlett 2i2 claims to improve input DR.

Line Input DR now spec'd at 115 dB A-weighted (2i2 G3 Measured 104 dB unweighted.)
Line Output impedance 100Ω. (Was 300Ω/leg.)
Has external USB-C power input in addition to bus power. 900 mA requirement.

I do use a USB iFi Defender ground isolator on all the bus-powered converters and inject 5V from an external supply.
 
E1DA Cosmos now produces an autoscaler ,
Im not all together sure why the gain increments are all over the place ,
https://e1dashz.wixsite.com/index/cosmos-scaler
Ivan explained it in one of his posts. He uses one single point of gain control and resistors that are wired in series, so it's not lke a binary combination. The lower value resistors give better resolution than the higher ones. The actual number of bits is 5. He could have had a better resolution but it would have needed a more complex logic and more relays.
 
I use AudioTester, REW, ARTA, YMEC, Visual Analyzer and have RMAA but don't use it any more.

My "go to" is usually AudioTester with a Focusrite 2i2 Gen 3. Will be getting a 2i2 Gen 4 soon as it has higher A/D DR and lower output impedance.

AudioTester was last upgraded in February 2023.

For noise measurements that exceed the DR of the 2i2 I use a TI PCM4222 EVM and for super-low (-140 dBc residual) 1 kHz distortion measurements I use ULDO-Nacho with the 2i2. The one thing I don't like about the PCM4222 is that it "leaks" Vcm out of the input XLRs.
Thank for all these tips. I am actually considering buying some of your PCB's to build my mastering console come next summer/autumn.
I have REW, it works quite nicely with the UMIK-1 (MiniDSP) but I fail to see how to measure circuits such as preamps, EQs and the like. Mind you I had it only for a few days now and did not do a complete RTFM as of yet. :)
I also downloaded the free version of AudioTester and just had a 10 minutes look into it. With this one a "front end" is definitely required.
 
Hmmmm!...
The gentleman who developed ARTA is retiring and no longer sells licences so I can use only the demo mode :(
I am looking into it right now and find it very interesting.
It does all what I need it seems, but I will dig further into REW today and definitely I need to build a front-end.
The one suggested by RuffRecords above is a one day project and I think I have everything needed on hand.
 
Bonjour everyone!
I tried a few things yesterday with REW, Visual Analyser and Audio Tester and boy oh boy thsse softwares are deep!
So far, I am inclined toward VA for nothing else maybe that I like the user interface more in general.
But definitely I have in hand what I was asking for in the OP.
I found that yes indeed, we need a form of input/output hardware front end for the sound card so I will start a new thread about this very subject.

Thanks to all, this discussion was very interesting, Luc
 
A software I re-discovered recently is Open Sound Meter. I had it installed for a while, but forgot about it. It seems popular with FOH engineers and it's available for Mac, Windows, Linux, and iOS. See:

https://opensoundmeter.com/en/download

Software that is available for most platforms always looks interesting to me. Desktop versions are free and Open Source. iOS version is 69$.
 
Ian I'd be curious to see if you've found a quieter gain sweet spot with the Scarlett than I have.
I use a slightly different calibration strategy with the Scarlett 2i2 Gen 3.

The normalized 0 dBFS level for almost all 5V analog stage differential input converters is 2V RMS.
In AudioTester I set the FS calibration to "2000 mV," apply 1.00V RMS into the input and adjust the front panel level control for -6dBFS.
This makes the overall gain unity from input connector to A/D modulator input.
It may not be the ultimate sweet spot for noise but I'm able to measure down to about -96 dBu unweighted over a 20 kHZ BW with FS at 96 kHZ.
That's a DR of only 104 dB so for lower level noise measurements I use the PCM4222 which sits around -109 dBu.
The only real issue I have with this calibration strategy is not bumping the knobs.
I have not attempted to find a noise sweet spot yet. My main noise measurements are of mic pres with 60dB gain which with a 150 ohm source will have an output noise level of at least -70dBu. I seem to rememeber trying the Scarlette in loopback (to create a cal file for REW) and its noise was well below -90dBu. Most of my tube circuits achieve output noise levels in the high -80s.

Cheers

Ian
 
Thanks Ian. Like yourself I find the 2i2 quiet enough for measuring preamp noise. It's only when I need to check a line-level circuit's noise floor that I break out the PCM4222 board.

Good luck with your projects blue-luke.
 
Thanks Ian. Like yourself I find the 2i2 quiet enough for measuring preamp noise. It's only when I need to check a line-level circuit's noise floor that I break out the PCM4222 board.

Good luck with your projects blue-luke.
As an aside, at Neve back in the 70s we had a little 40dB preamp in a die cast box that was used to raise console output noise up to a level where the Radford test gear we used could actually make a measurement.

Cheers

ian
 
Back
Top