I have read thru the thread, too quickly, I'm sure. Can someone provide a link to download Speaky's software for the QA403?
Hey @tomchr
You might take a look at this program I'm working on. Compared to the stock app, well...
See: https://github.com/MZachmann/QA40xPlot
You might take a look at this program I'm working on. Compared to the stock app, well...
See: https://github.com/MZachmann/QA40xPlot
thanks for sharing.
wow, that's good enough to make me dig up my SB1240 and do some surgery ...
what OS do you use, windows 11? do you do anything special for drivers?
and how did you get around the mysterious CS5381 replacement issue that googlyone experienced?
Last edited:
Imho no. A soundcard is effectively just part of a measurement system, where the noise and distortion can depend on many aspects, including additional system parts (like attenuators and notch filters and probes). If 100dB span is your aim, then perhaps determine what a base level system can achieve in a shielded environment with heavy REW filtering to suppress noise floor, then review what notch filters can do for you.
Thanks, trobbins. Maybe I wasn't clear enough: I have been chasing sub- minus 100 dB differences in noise floor between two channels of a system. While REW does show reproducible differences, I went back to Tom's loopback noise plot for the 3rd gen Solo in his first post, and that seems to tell me that I'm simply at or below the limit of the range the Solo can cover in a single measurement. I'm not asking about optimizing the measurement strategy (although that is a worthwhile topic), I just want to avoid wasting time by getting sidetracked into trying to understand features that are literally 'in the noise'. I'm simply asking whether I interpret Tom's plot correctly.
Post #6 indicates noise level is -140dB of test signal in loopback for 2nd gen. Have you searched/checked what the difference is wrt noise floor for the 3rd gen Solo?
If you want to look at THD much below -100 dBc the lower end Focusrite products won't be of much help. I don't know if the more expensive models are better in this regard or if they just have more inputs.
There's an RME ADC/DAC that offers THD around -130 dBc if I recall correctly. It's around $2500 AFAIR.
There's the Cosmos stack, but that's three different gizmos that you have to tie together with a medusa head of cables and adapters.
The QA403 offers the best performance per dollar. It's just unfortunate that the software isn't at the same level as the hardware.
Tom
There's an RME ADC/DAC that offers THD around -130 dBc if I recall correctly. It's around $2500 AFAIR.
There's the Cosmos stack, but that's three different gizmos that you have to tie together with a medusa head of cables and adapters.
The QA403 offers the best performance per dollar. It's just unfortunate that the software isn't at the same level as the hardware.
Tom
trobbins, plots of that type are exactly what confused me, but I think I'm slowly getting my head wrapped around it. Thanks for helping me with this very basic problem of understanding. The 3rd gen Solo does indeed show a similar level near -140 'dB' in loopback mode. The 'spectrum' plot units really are dB/sqrt(Hz), not dB, so a flat -140 'dB' line integrates out to -140+43=-97 dB over a 20kHz window (20*log(sqrt(20,000))=43). So, I guess the answer to my original question is really 'no', because the Solo does meaningfully measure noise density down to -140 dB/sqrt(Hz), for a full scale value of say 1 V that's 100 nanovolt (per sqrt(Hz)), or 14 microvolt over 20 kHz bandwidth. Pretty amazing. I hope I'm finally getting the math right.
These numbers also tell me that it is pretty hopeless to chase noise sources at this level with a scope, given that I can't really see/measure a 100th or even a tenth of a mV.
Tom, no, I'm doing this just for fun, so I don't need better equipment. But if I did or wanted, then the QA403 seems like a nice compact option. The Cosmos family of devices I must say I still find a bit confusing.
These numbers also tell me that it is pretty hopeless to chase noise sources at this level with a scope, given that I can't really see/measure a 100th or even a tenth of a mV.
Tom, no, I'm doing this just for fun, so I don't need better equipment. But if I did or wanted, then the QA403 seems like a nice compact option. The Cosmos family of devices I must say I still find a bit confusing.
My 12 cents - I am using Topping D10s + E1DA Cosmos ADC as a relatively cheap measuring system, with total cost of about $300. THD vs. frequency measurement limit of the direct loop is shown below:
In case of single-ended output measurements, I use USB-ISO for the D10s, if necessary. I also use a set of balanced dividers to cover output voltage levels of power amplifiers, up to 70V.
THD limit at middle frequencies is about -126dB, THD+N limit is about -112dB (flat 22kHz). Noise level is then estimated by the D10s. I use REW and Arta/STEPS software. In a direct comparison with QA403, E1DA ADC is able to measure lower distortion and lower noise, but it needs qualified operator.
The LF rise of distortion is completely due to distortion in input electrolytic coupling capacitors in the Cosmos. This was proven by bypassing them by a piece of wire, but the resulting input DC shift is then unacceptable.
Cosmos ADC has too low input impedance and for measurements of link level stages it is good to use an additional buffer with a good op-amp and input impedance >= 10k.
In case of single-ended output measurements, I use USB-ISO for the D10s, if necessary. I also use a set of balanced dividers to cover output voltage levels of power amplifiers, up to 70V.
THD limit at middle frequencies is about -126dB, THD+N limit is about -112dB (flat 22kHz). Noise level is then estimated by the D10s. I use REW and Arta/STEPS software. In a direct comparison with QA403, E1DA ADC is able to measure lower distortion and lower noise, but it needs qualified operator.
The LF rise of distortion is completely due to distortion in input electrolytic coupling capacitors in the Cosmos. This was proven by bypassing them by a piece of wire, but the resulting input DC shift is then unacceptable.
Cosmos ADC has too low input impedance and for measurements of link level stages it is good to use an additional buffer with a good op-amp and input impedance >= 10k.
Interesting. I can't find an exactly corresponding plot in Tom's comparison on Neurochrome.com, but this plot
indicates, if I interpret both plots correctly, that at 1kHz the QA403 is at least 6 dB worse than the 1 kHz point from your DAC and ADC combination. Both plots show the integrated distortion, but yours is integrated over 45 kHz bandwidth, and Tom's over 20 kHz, so not apples to apples (besides plotting integrated distortion as function of different variables).
indicates, if I interpret both plots correctly, that at 1kHz the QA403 is at least 6 dB worse than the 1 kHz point from your DAC and ADC combination. Both plots show the integrated distortion, but yours is integrated over 45 kHz bandwidth, and Tom's over 20 kHz, so not apples to apples (besides plotting integrated distortion as function of different variables).
Your observations are correct. Tom's plot shows THD vs. level at 1 kHz (BW20kHz), my plot shows THD vs. frequency (at about -6dBFS /ADCinput/ and with BW45kHz). There is a big advantage of QA403 auto ranging. Cosmos, on the other hand, can get deeper in the low level distortion and noise. If I make similar plot as Tom did, it looks like this (please note I have no auto-ranger):
X-axis in dBFS re DAC output.
P.S.: I also have a Focusrite 2i2 and it can only be used for acoustical measurements. It is in no way able to measure neither preamplifiers, nor low distortion power amplifiers. Noise level is too high and distortion at higher frequencies as well.
X-axis in dBFS re DAC output.
P.S.: I also have a Focusrite 2i2 and it can only be used for acoustical measurements. It is in no way able to measure neither preamplifiers, nor low distortion power amplifiers. Noise level is too high and distortion at higher frequencies as well.
Last edited:
That's pretty good.
Be careful with the units when comparing the graphs. The graphs I show of the QA403 shows THD versus input level (dBV) whereas some of the plots above show THD versus input level in dBFS. Without knowing what voltage 0 dBFS corresponds to we can't compare the graphs directly. That doesn't mean that comparisons are invalid; only that we should be a bit careful when applying them.
The ADC+DAC combo will have one sweet spot, i.e., lowest spot in the THD vs level graph. Dedicated analyzers like the QA403 and APx500 will have multiple sweet spots due to their input and output attenuators.
Tom
Be careful with the units when comparing the graphs. The graphs I show of the QA403 shows THD versus input level (dBV) whereas some of the plots above show THD versus input level in dBFS. Without knowing what voltage 0 dBFS corresponds to we can't compare the graphs directly. That doesn't mean that comparisons are invalid; only that we should be a bit careful when applying them.
The ADC+DAC combo will have one sweet spot, i.e., lowest spot in the THD vs level graph. Dedicated analyzers like the QA403 and APx500 will have multiple sweet spots due to their input and output attenuators.
Tom
My point is opposite. There is only one plot in dBFS and it in fact shows DAC performance, because ADC is much better. It is the limit. The rest is a mere scalling die to input dividers of the ADC. I am very well aware of units and scalling.