DT9857E shows asynchronous behaviour between analog input and output channels

[Domme]

Hello,

I have been using the DT9857E for a while now, mainly for multi-channel microphone input reading, but I just recognized that the input/output operations are not being performed synchronously while trying to measure impulse response. Instead, there is a little lag which differs from measurement to measurement. This makes correct measurements of phase impossible.

I attached a script to show the problem. The figure below shows how the input is captured at different samples for 5 separate readwrite executions in an otherwise identical setup.

 

Is there a possibility to synchronize input and output channels using software-sided triggers? Or do I have to sacrifice an input and an output channel each for a loopback?

Kind regards,
Domme

sync-problems.m

Edited March 25 by Domme

[Fausto]

Hello @Domme.

Please confirm that your target system is using a previous version (v7.8.2) of DT Open Layers library for the DT9857E.

 

Disconnect your microphones and loop the analog output directly to the five analog input channels.  Please test and reply with screen capture(s) and comments.

Regards,

Fausto

 

[Domme]

18 hours ago, Fausto said:

Hello @Domme.

Please confirm that your target system is using a previous version (v7.8.2) of DT Open Layers library for the DT9857E.

 

Disconnect your microphones and loop the analog output directly to the five analog input channels.  Please test and reply with screen capture(s) and comments.

Regards,

Fausto

 

Hello Fausto,

thanks for your answer. I am using DT Open Layers v7.8.2. Just to clarify: I have not used 5 microphones for that example diagram, but did 5 separate (identical) measurements on the same input channel to compare sample deviation from measurement to measurement. Looping back the output to input channel 1 for 5 executions of readwrite gives the following signals on the input channel:

As can be seen, the output signal starts before the input channel is ready, resulting in additional phase between input and output. Also, the output ends a few samples before it should:

In summary, a part of the signal is missing. This could be "fixed" by adding silence to the beginning, looping back, subtracting the latency and so on, but it would be a kind of ugly workaround and I would need to sacrifice an input and an output channel. I hope that there is a better solution.

Kind regards,
Domme

Edited March 26 by Domme