Codec ADC/DAC latency

[parkersound]

What's a typical value for the latency associated with a codec latency? I am looking at the Wolfson WM8960. Could it be responsible for ~1ms of loopback latency when using a very small buffer size (2 samples)?

 

[DVDdoug]

It's buffers. You may not need buffers inside a DAC/ADC unless it's doing sample-rate conversion or some other processing. (I guess most DACs upsample and digitally-filter these days.)

2 samples is less than 1ms (depending on the sample rate).

With a computer and a multitasking operating system, which is always multitasking and interrupting even when you're only running one application, you'll probably never get-away with such a small buffer. Those buffers are in the computer.

 

[voodooless]

You cannot control buffers on this thing, so it's pointless to talk about that.

Most of the propagation delay comes from the digital filters in the DAC and ADC, in both cases, the group delay of these is 18/fs. At 48 kHz, that is about 0.75 ms in total.

 

[parkersound]

Digital filters can be controlled though - I am playing around with the configuration of the codec via I2C. I have switched off all the digital filters I could find, but I still see ~1.2ms round trip latency (copying the ADC out to the DAC in with a microcontroller which I believe should have ~250us of round trip latency).

 

[parkersound]

Although from the datasheet I see that there is a digital DAC filter and a digital ADC filter that have a group delay of 18/fs as you say and they cannot turned off as far as I understand. That would imply 750us of latency only for the filters + 250us due to the microcontroller which wouldn't be too far off from the latency I measure of ~1.2ms.

 

[Philbo King]

I'm not sure what the end game is here. Latency is not an issue in playback-only systems; it only makes sense for studio work. Even then 1.1 mSec is the time it takes for sound to travel 12 inches. If that amount of delay throws off a musician ("I can't synchonize my playing to somebody 1 foot away, it takes too long for the sound to reach me!") then I humbly suggest latency is not the problem.

 

[parkersound]

Sure, the problem is that if you start adding up the latency coming from 10 different components in a signal chain and each of them has 1ms of latency you end up with 10ms of latency. And that's definitely noticeable (I'd say things start getting messy around 5ms, if not less). If latency was not an issue, I don't think guitar effect companies would go so far to obtain latencies in the sub-millisecond range.

.

 

[weesch]

What's a typical value for the latency associated with a codec latency? I am looking at the Wolfson WM8960. Could it be responsible for ~1ms of loopback latency when using a very small buffer size (2 samples)?

hi
i'am not shure but when i want to know the latency i look at the group delay in FS or (samples)

WM8960 datasheet(82/92 Pages) WOLFSON

WM8960Production DatawPD, August 2013, Rev 4.282REGISTERADDRESSBITLABELDEFAULT Datasheet search, datasheets, Datasheet search site for Electronic Components and Semiconductors, integrated circuits, diodes, triacs and other semiconductors.

www.alldatasheet.es

here its DAC 18 fs and ADC 18 fs so it's make 36 fs or samples.
so at 44khz it makes 36/44 = 0.8 milliseconds only for AD/DA convertion
i remenber that the fastiest converter is made by Asahi Kasei Microdevices