REW + DATSv3 = Impedance and TSP measurements made easy (Mac and Windows)

[Olen Rasp]

I may be late to the game, but couldn't find anything posted about DATSv3 when searching the REW forum.

I've been using my DATSv3 with REW to measure impedance and Thiele-Small parameters on my Mac and posted a how-to guide on diyaudio a couple of years ago (see attachment). It's become even easier in recent versions of REW to setup and run calibration on the DATS. I much prefer REW to the DATS software (which was Windows only last I checked) and haven't looked back ever since. DATS+REW works just as well on Windows and Mac. BTW I really like the dual added mass feature.

Have a look and give it a try! If there's interest I can update the instructions for a more current REW version.

Thanks,
Olen

 

[dcibel]

I'm surprised DATS uses such a large vale sense resistor, 10kOhm? It would seem such a large value would create a high noise sensitivity. Reference resistor calibration effectively nullifies the Rsense manually entered value, so easy confirmation would be to simply complete an impedance measurement without reference calibration, ie complete step 1 and 2 only for calibration process.

I think I mentioned in your old thread at DIY Audio, but it's great to see that DATS users have an option to use REW as an alternative to the proprietary Dayton software.

 

[Olen Rasp]

It's a bit noisy but averaging multiple sweeps helps a lot (averaging is common in acoustical measurements), as does running the sweeps as close to full-scale (0dBFS) as possible. One thing I love about REW is the real-time readout showing the noise floor before the measurement starts.

The 10k is in series with the load, meaning for any load/DUT impedance under 100-ohms or the total impedance varies by less than 1% - it goes from 10.0k (with a dead short) to 10.1k (with 100-ohm load). Say the output of the opamp driving the resistor+load is 1V. DATS measures the voltage across the load and can assume with 1% error that the current is 1V/10kohms = 0.1mA. If the load impedance increases to 100-ohms then then current is 0.099mA, close enough for most cases.

The DATS ADC measures the voltage across the load, (with a stated max impedance 10k) I'm guessing they set 10k as the full-scale input of the ADC (anything above that probably gets clipped), but I think this depends on which SW you're using. With a 10k load and the 10k series resistor there should be 0.5V across each. The DATS 16-bit ADC only has 65k steps, so you must scale accordingly - if you set the max impedance as 10k the smallest impedance you can measure is 10k/65k = 0.153=ohms (full-scale digital). If you set full-scale at 1k then the smallest impedance is 1k/65k = 0.015-ohms.

I'm curious how the DATS measures impedances above 1k ohm accurately (as it states on the produce page), I haven't tried. I usually calibrate my DATS in REW usinga known resistor value closer to the impedance range of the speakers I'm measuring (like 10-ohms or thereabout), and get accurate/consistent results with 10-20milliohms of resolution.

 

[dcibel]

Yes, of course the high value of Rsense provides very much a "constant voltage" source, as it would be under normal use. I should really try a comparison of 10kohm sense vs the 10 ohm that I use just to see how bad the SNR difference is. What I've found in the past, is that with a small Rsense value, highest voltage will be applied to the driver at Fs peak, which is what you want anyway, and smaller voltage applies through the low impediance frequencies has little effect, so the result of TSP is very much the same even though the voltage applies varies somewhat over frequency.

For context, I am using a Motu M4 with my own jig (detailed here) and any power amp on hand, so it's a bit more flexible as I have full control over the applied voltage, and of much higher SNR than the DATS. FWIW, you have assumed above that SNR is limited by the 16 bit ADC resolution, which is not often the case. You may find actual SNR may limit effective bit depth to say 14 bits.