Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[thomaspf]

I came across a strange problem with JRiver. I just finished building a new DAC/Amplifier combo and as a final step wanted to fine adjust the bias voltages and zero voltage differential for the balanced symmetrical power amp.

The problem seems to be that when I hit Pause or Stop on JRiver then the digital output does not seem to send zero samples but some other constant value resulting in a DC offset. That DC offset at the DAC does get amplified into about 0.5V at the speaker outputs so there is a constant output current stressing the drivers in the speakers.

I first suspected the USB->S/PDIF converter I am using and contacted the vendor who pointed out to me (I verified this) that with Foobar the output goes to zero on Stop or Pause. In fact this is always 0V when no music is playing except when I use JRiver.
 
What is the expected behavior here?

Cheers

   Thomas

[JimH]

What version of MC are you using?

Does the device use the Thesycon driver?

[thomaspf]

I discovered this on MC16 and then tried the latest stable version of MC17. The device is a Lindemann DCC-24/192 using the USB Audio class 2 driver that comes with it. I am not sure whether that os a derivative of the Thesycon driver.

I can replicate the behavior with an older AudioTrak Optoplay that is using the built-in Microsoft USB audio driver.

Cheers

   Thomas

[JimH]

Try the build at the top of the MC17 board.  It's 17.0.178.

[thomaspf]

Interesting. I just upgraded to MC v178 and the behavior is somewhat different.

While before the offset was always around 0.4V left and 0.2V right on my speaker posts. It now varies between stop and pause.
For Pause it is now very close to 0V but for stop it still sits around 0.04V DC offset. I also enabled play leading and trailing silence on the options.

You can hear a faint plop when the speaker snaps back to normal position when the playback starts again.

Cheers

   Thomas  

[Matt]

What audio output mode are you using?

Please try 'Wasapi - Event Style' because it sends silence on pause so the hardware never sees a pause / flush.  This makes it the most resilient to hardware differences.

[thomaspf]

Hi Matt!

I have been using kernel streaming because in my experience and with this specific device that was the most stable output method with switching between different resolution material.

I just tried WASAPI event style and pause now always goes to zero but on stop i am seeing 0.096V left and right.

I don't even quite understand what is happening here. Doesn't JRiver close and release the device on stop? Why is there a constant non zero signal coming out. Maybe sending a few zeros before closing the device tricks it into staying a zero?

Cheers

   Thomas

[Matt]

I just tried WASAPI event style and pause now always goes to zero but on stop i am seeing 0.096V left and right.

I don't even quite understand what is happening here. Doesn't JRiver close and release the device on stop?

We close and release everything on stop.  If it's still outputting a voltage when we're stopped, I think it's a driver / hardware problem.

To test your silence idea, pause and hit stop while paused.  Does that zero the output?

Feel free to give the DAC maker my contact information.  We'd be happy to work with them to solve the problem.  I'm matt at jriver dot com.

Thanks.

[thomaspf]

Hi Matt,

Pause followed by Stop seems to do the trick!

I send an introductory mail.

Cheers

   Thomas

[Skogkatt]

As Matt says it might be a driver / hardware problem but then it is a quite common one as I verified it with both Realtek HD and ASUS Xonar Essence ST (both using SPDIF output) under all WASAPi flavors.

Silence after stop was added for ASIO with 17.0.30:

17.0.30 (11/7/2011)
5. NEW: ASIO outputs a little silence when stopping so that hardware that plays its last buffer on the next start won't click.  About 1/4 of a second.

This worked reliably for ASIO but limited to that interface type. A workaround I found was using "stop: fadeout" so output goes to zero before it gets released.

Still I wish to have an option to choose "output always engaged" after stop as probably happens with pause.

A little note about "stop: fadeout" : it works much better (using it together with convolver and its latency) since it was made independent from internal volume but sometimes (randomly) output is muted before it gets smoothly faded to silence. When this happens a little click can be heard on the following play (ASIO): is it possible that output is released too early?

[Matt]

it is a quite common one as I verified it with both Realtek HD and ASUS Xonar Essence ST (both using SPDIF output) under all WASAPi flavors.

I have a Realtek and Xonar at work and test with them a lot.  I've never seen this.

It makes me wonder if something later in your audio stack is looping, not the device itself.

[Skogkatt]

Xonar analog outputs seem immune (tested the headphone output) but SPDIF no, even with ASIO.

[Matt]

but SPDIF no, even with ASIO.

But what's receiving the S/PDIF?  How do you know it's not the thing that is looping a buffer?

[Skogkatt]

But what's receiving the S/PDIF?  How do you know it's not the thing that is looping a buffer?

There is a S/PDIF receiver chip decoding the signal to I2S followed by an ASRC and the DAC chip: till there is a valid S/PDIF signal at the
input, the thing keeps moving data refreshing constantly the analog output. So, no looping buffers there: first in first out.

The only way to understand something is by measuring what data the S/PDIF carries out the XONAR after a stop while playing music.

No problem whatsoever if MC is stopped while the track is silent.

The XONAR uses muting relays on each analog output so the problem can't be seen there.