ArrayIndexOutOfBoundsException when changing the input channel in Preferences (reversing Input and Loopback)

Running in Windows 10 Pro. I've had it running for quite some time (many days). I'm doing a lot of testing/learning. For these I'm using the Delta 1010LT unbalanced out to balanced in. This was a loopback connection, calibrating the balanced inputs. Ch1 has been calibrated for several days. I was switching to Ch2. The moment I clicked the Ch2 input in the pulldown list it crashed.

I've attached a zip with the roomeqJ_wizard1.log and an extract of the rew_output.txt files.

Those errors are related to deleting a measurement. Had something just been deleted?

John Mulcahy said:

Those errors are related to deleting a measurement. Had something just been deleted?

Click to expand...

I think that I had. I try to keep the running measurement total down as I'm experimenting with the RTA for amplifier distortion for optimal settings and discard those I no longer want to maintain. Part of that is due to finding your post in an older thread about added distortion. The results are nothing short of amazing. The "Select from RTA" is a fantastic time saver, especially when making many RTA runs for different generator and/or RTA settings.

As an aside, I've settled on always running the RTA using Coherent averaging since the distortions tend to be near or in the noise floor. The issue is, of course, that the HD results (in my case) include DAC and ADC. At this point I haven't found a way to separate them. I have an ADCiso that may allow me to make the card DAC the primary contributor, but it would be helpful if there were some way other than having separate hardware with self-noise low enough to isolate DAC or ADC.

dlr said:

The issue is, of course, that the HD results (in my case) include DAC and ADC. At this point I haven't found a way to separate them. I have an ADCiso that may allow me to make the card DAC the primary contributor, but it would be helpful if there were some way other than having separate hardware with self-noise low enough to isolate DAC or ADC.

Click to expand...

Separating DAC and ADC distortions from the combined loopback measurement is by principle not trivial. The standard way is using a steep notch filter on the ADC side which suppresses the DAC distortions before the AD conversion. REW supports measuring the notch filter params and compensating its FR during measurements.

There is a way to separate the DAC and DAC distortions in digital domain but it requires an additional calibration circuit https://www.diyaudio.com/community/...on-compensation-for-measurement-setup.328871/ https://patents.google.com/patent/US11283458B1/

phofman said:

Separating DAC and ADC distortions from the combined loopback measurement is by principle not trivial. The standard way is using a steep notch filter on the ADC side which suppresses the DAC distortions before the AD conversion. REW supports measuring the notch filter params and compensating its FR during measurements.

There is a way to separate the DAC and DAC distortions in digital domain but it requires an additional calibration circuit https://www.diyaudio.com/community/...on-compensation-for-measurement-setup.328871/ https://patents.google.com/patent/US11283458B1/

Click to expand...

I'll read through the patent link. I had found and read through some of the first link thread that you started, was hoping that in the intervening years some other method may have been found. Evidently not. Other threads have had useful info that I've spent a lot of time scanning (500+ pages for one thread is too time consuming). Fortunately google searches have brought up links directly to some of the more helpful posts in several threads (many are yours) such as a response by John that pointed out the added distortion option in REW.

The post by REW added distortion by John may have been in one of your discussions. I'll be posting some of my results in my thread on modest audio interface testing, but I'll add a brief note here. I had found issues earlier (no doubt others will have as well), one being that the "sweet spot" in a loopback of an audio interface is may not the best spot for the ADC or DAC side. Another was that the best DAC point was about -15dBFS for both 1010LT and 2i2, but using the added distortion option coupled with the Delta control panel output level control I was able to set the 1010LT output to -0.01dBFS and maintain excellent HD that meant being at the absolute best point for self-noise.

Edit: Added link to the post about added distortion. Response to Jan Didden, but you were involved in that thread.

Yes, if the ADC distortion component to the combined distortion is negligible, compensating (i.e. adding distortions of opposite phase) on the DAC side only works OK.

The problem is when the ADC distortion is not negligible, while being also compensated on the DAC/generator side . Then the output signal from the DAC is not distortion-free, but contains the ADC antidistortions. Then if the DUT (because measuring DUT distortions below distortion level of the uncompensated loopback is the goal) rotates these distortions (due to DUT FR of non-zero phase at the distortion frequency), then the ADC anti-distortions do not cancel out the vector-added ADC distortions and distort the measurement. That's the reason for splitting the DAC and ADC components and compensating DAC contribution on the DAC side (yielding clean sine signal) and ADC contribution on the ADC side - removing the ADC distortions from the combined DUT + ADC distortions.

But it's true that today's ADCs add so little distortion that often it can be ignored.