Thanks Kevin. $2495 is too expensive for me (I'm a modest diyer
). I was thinking of a similar price to what was offered in the group buy (USD 999/ea). I'm now using an E-MU 0404 USB sound card, which doesn't do a bad job for the price, but I'd like something more "professional". For a significantly lower cost there is the QA402, maybe a bit lower in performance but it seems to be a good buy. Can anyone give an opinion on this?
Good to know... I don't like to depend on a specific SW, that's a severe drawback...
The software in this case is quite good. I actually like software tailored to the hardware and application which is why I use Virtins with my RTX. QA was working on an option to allow their hardware to work with standard audio applications so I would investigate to see how far they got.
An open source ASIO driver was written for the QA's Cypress USB interface https://github.com/dechamps/ASIO401 . Etienne Dechamps is an ASIO guru, look at his github repos.
For linux QA have a headless tool communicating with the HW via userspace libusb. Not a soundcard interface though.
For linux QA have a headless tool communicating with the HW via userspace libusb. Not a soundcard interface though.
Maybe "professional" is not the right word... I'm basically looking for a set of features that you don't get by using a sound card:
- An effective protection against overvoltage, as well as the possibility to apply at least up to 30V RMS to the input (despite the care I take when I do my measurements, I have already damaged two cards in several careless, relatively easy to make mistakes...).
- That the system is self-calibrating and automatically selects the range. (Although it's not complicated to calibrate a sound card, it's a real pain in the ***...)
- Distortion and noise levels several orders of magnitude lower than those of the devices I want to measure (for my particular applications, phono equalizers and preamplifiers are usually the most demanding). For inputs, at least:
· Frequency Response: +/- 0.02 (20Hz-20kHz)
· Input Noise Floor (input shorted, 20 kHz BW, unweighted): -115dBV
· THD (1kHz @ 0dBV): -114dB
· Input Impedance: at least 100K (unbalanced) , preferably bigger
- Analogue outputs with similar characteristics and output impedance not higher than 150 Ohms.
Last edited:
One alternative is to use a measurement interface for soundcard. Here in diyaudio you can find 2 very good alternatives:
- Pete Millett's: https://www.diyaudio.com/community/threads/test-measurement-interface-for-soundcard.155405/
- Jan Didden's Autoranger
Several orders of magnitude better than modern preamps is a strong requirement, considering the performance of the current audio technology.
For the last year I have been working on an analyzer which should satisfy the above requirements (with distortion-compensated THD, THD+N less likely), including fully automated level calibration of all I/O ranges and compensation of DAC/ADC harmonic distortions, input autoranging up to some 350Vpp balanced, 200k input impedance balanced (noise?), 100R output impedance balanced , Vout 4Vrms to 100ohm balanced load (i.e. 22Vpp balanced no load). DAC/ADC sampling rate 384kHz flat BW to fs/2, maybe 768kHz. Connected to PC as a regular duplex UAC2 audio device (no drivers required for Win/OSX/linux) with 32bit full-resolution samples adjusted for the current range, composed with a USB ethernet device (like e.g. some RME/Motu devices do) for web browser control, integration with John Mulcahy's commercial REW Pro (autosetup for the calibrated level, basic duplex communication REW <-> analyzer). I will decide about going to production after learning prototype parameters and production costs, maybe with a crowdfunding campaign. The automated calibrations make it a rather complex device (lots of relays), it would be well above the "DIY incidental shopping" price range. The current component shortages and the resulting crazy price hikes are not friendly for projects like this though
Last edited:
Cool! How does the autoranging work? Autoranging can be a nice feature, but depending on the application it is sometimes better to leave the autoranging to the software, or to turn it off completely.The automated calibrations make it a rather complex device (lots of relays), it would be well above the "DIY incidental shopping" price range. The current component shortages and the resulting crazy price hikes are not friendly for projects like this though
Yes, autoranging requires co-operation with the measurement software. That's why the mentioned integration with REW (which is not done yet, but agreed on). For sweep the analyzer must know the max. out level first, to allow using the largest range without switching. For input - you can pre-measure DUT's min/max first (while disabling/zeroing stream to the measuring PC) and based on the result choose the largest required input range for the sweep. However, stepped measurements are better in this regard - FFTs more precise, time available for averaging, the distortion compensation can be active at every step, range can be switched at every step appropriately. Because the input and output are always a single range of 32bit (24bit measurement + 8bit ranging), no level calibration switching is required in the measurement software. The ranges will be mutually calibrated, but still switching a range would be visible on the swept output data, it cannot be continuous down to LSB. Of course it requires a complex controlling software running on the analyzer, which is actually the point of the device
Last edited:
Well, I know this will make some people shake their heads, but, for diy, I prefer the tube preamps, then the figures are no so demanding... but I still enjoy solid state as well and tend to use it when it has a clear advantage in certain applications.
I am delighted to hear that you have a project underway. When could we see the fruit of it?
Would you be interested to make this available to other software developers, so that they can add support to their software packages? Would you consider borrowing boards to test with the software?
Honestly, I have not considered integrating with other SW but REW Pro for now. It will take some substantial work on the REW side and something needs to be offered in exchange. But it would behave as a regular UAC2 device, standard UAC2 samplerate selection, with web interface and documented network API for high-level control (manual ranging, calibration/compensation control, etc.). It would be shipped as a finished product, case with LCD, certified third-party power adapter, warranty, not just a board. The assembled board would be too expensive for being just a board
As of the prototype time frame - this year would be nice. May not seem very ambitious but things just take time. Hopefully the semiconductors market will recover by then.
If anyone is interested in more details, please PM for now, I do not want to hijack this thread.
As of the prototype time frame - this year would be nice. May not seem very ambitious but things just take time. Hopefully the semiconductors market will recover by then.
If anyone is interested in more details, please PM for now, I do not want to hijack this thread.
FJC, one aspect of aiming to do measurements with very low distortion levels is that the oscillators available for diy tend to be single frequency (eg. Victors). In that situation, the use of REW harmonic cancellation of a soundcard like EMU0404 can be in effect make an oscillator with much better HD levels than default - so the soundcard is then not a significant hurdle to undertaking that type of measurement. Also, at that single frequency, the use of a frequency compensated divider (like a scope 10:1 or 100:1 probe) is not significant hurdle to allowing a few hundred volt level sensing.
Autoranger has various models.....limit can be changed.....please read site.