Page 2 of 2

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 6:13 pm
by w-u-2-o
Paul,

There is no sidetone via VAC. I am curious what you are hearing. Are you using PowerSDR or Thetis? If Thetis, do you have QSK turned on?

73,

Scott

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 6:22 pm
by w9ac
w-u-2-o wrote:Paul,

There is no sidetone via VAC. I am curious what you are hearing. Are you using PowerSDR or Thetis? If Thetis, do you have QSK turned on?

73,

Scott

Yes, QSK is engaged in Thetis and VAC is active. So, the CW side-tone comes from the receiver and not the side-tone generator.

Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 6:48 pm
by w-u-2-o
That's fabulous, Paul! And it really demonstrates the big difference that ASIO drivers and hardware can make.

Chris, W2PA, and I did an experiment using a non-ASIO drivers and could not get the latency down low enough to do what you are doing. Unfortunately Chris does not have any ASIO hardware.

Were you also using the low latency filter settings?

Thanks,

Scott

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 8:19 pm
by w9ac
w-u-2-o wrote:That's fabulous, Paul! And it really demonstrates the big difference that ASIO drivers and hardware can make.
Were you also using the low latency filter settings?

Yes, although the linear phase setting with a 64/2048 filter size produces similar results with extremely low latency.

I do see some random VAC overflows (like 10 an hour) but I don't hear any audio glitches and Var Ratio hovers near 1.0 with only the last two digits (five and six places to right of decimal) slowly bouncing around. Mostly, the 5th digit is solid. So to recap:

Bodnar GPSDO Output 1 set to 10 MHz and routed to ANAN 7000DLE Ext. Ref. In
Bodnar GPSDO Output 2 set to 48 kHz and routed to Presonus 192 Mobile word clock input (block size auto set by Thetis);

Thetis settings:

Driver: ASIO
Input/output Device: Studio 192 Mobile
VAC To/From "force" boxes checked
Buffer size = 128 locked to Bodnar GPSDO at 48 kHz
RingBuffer In/Out both fixed at 1
PortAudio both fixed at 0

Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 9:13 pm
by w-u-2-o
You save at least 10ms in both TX and RX directions when using a low latency filter of any size compared to a linear phase filter of length 2048. IMHO it's worth using the low latency filter. Because the low latency filters are minimum phase FIR filters their latency does not change with size and remains fixed at around 10ms.

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 9:25 pm
by W4WMT
Paul,

Interesting experiment: Without changing any of your settings from above, remove the 48kHz word clock going to the Presonus and see what that does to your hourly overflow underflow count.

73, Bryan W4WMT

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 11:35 pm
by w9ac
w-u-2-o wrote:You save at least 10ms in both TX and RX directions when using a low latency filter of any size compared to a linear phase filter of length 2048.


With some time to kill this afternoon, I measured CW latency of the CW side-tone in Thetis' new QSK mode. In the screen captures below, the first set is linear phase, 64/2048. The second capture set is low-latency, same buffer and filter size.

The yellow trace represents side-tone audio from the ANAN 7000DLE's headphone jack. The lavender trace is sampled directly from the Studio 192's headphone jack. Horizontal division = 20 ms.

The green trace is a keyed pulse from a modified WinKeyerUSB. This output is used to trigger the scope.

Recall that when the new QSK mode is used, the internal side-tone is bypassed and the side-tone is derived from the receiver. I wasn't expecting this outcome at all. The ASIO driver's are operating faster than the Orion's DAC audio stage. Obviously, there's a lot common in Orion's signal processing path, but the on-board DAC path is slower than the VAC path with ASIO drivers.

Look at the side-tone delivered from the Studio 192's headphone jack when Thetis' filter type is set to low latency. That's 30 ms after key closure. The ANAN headphone jack delivers it in about 70 ms. Now, 30 ms may seem unimpressive, but remember this is the total delay in the transmitter and receiver.

By contrast, the linear phase filter type delays audio by 50 ms and 90 ms respectively. To Scott's point, the low latency filter delivers audio about 20 ms faster than the linear phase type.

Paul, W9AC

SDS00002.jpg
SDS00002.jpg (107.28 KiB) Viewed 15439 times
SDS00003.jpg
SDS00003.jpg (107.58 KiB) Viewed 15439 times

Re: Presonus Studio 192 Mobile

Posted: Sun May 19, 2019 11:51 pm
by w9ac
Bryan W4WMT wrote:Paul,

Interesting experiment: Without changing any of your settings from above, remove the 48kHz word clock going to the Presonus and see what that does to your hourly overflow underflow count.

73, Bryan W4WMT

Bryan,

If I pull the 48 kHz clock cable between the GPSDO and Presonus 192, then the Presonus ASIO 192 Input and output drivers disappear in Thetis' Virtual Audio Cable Setup menu. As soon as I reconnect the cable, the driver selection is restored.

I'm sure those drivers would stay unaffected if the Presonus 192 was set to the internal clock.

Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Mon May 20, 2019 12:02 am
by w9ac
Well, I rebooted Thetis and now the ANAN 7000DLE and Studio 192 headphone jacks show nearly the same response time. Not sure what that was all about. Captures below. Top capture is linear phase; bottom is low latency.

Paul, W9AC


SDS00004.jpg
SDS00004.jpg (107.3 KiB) Viewed 15435 times

Re: Presonus Studio 192 Mobile

Posted: Mon May 20, 2019 1:03 am
by w-u-2-o
That's great data, Paul, thank you!

Delay is only a single "PARIS" dot at 40WPM, or half a dot at 20WPM.

The important take-away from this is that, with an properly tuned ASIO sound interface, virtualized CW operation is possible with no degradation of QSK performance or sidetone in QSK mode. In fact it's at least a few milliseconds better than using a connection to the hardware unit.

Edited to add: full virtualization would also need to test keying from the Thetis serial port keying interface.

73!

Scott

Re: Presonus Studio 192 Mobile

Posted: Mon May 20, 2019 8:19 pm
by w9ac
Interesting. The timing disparity between audio at the Studio 192 headphone jack and the 7000DLE's headphone jack "creeps in" with time but corrects itself by disabling, then re-enabling VAC in Thetis. Sometimes it occurs within a few minutes, other times it happens hours later. It also clears up when re-booting Thetis. After correcting, the delay is nearly the same between the two as shown in the most recent screen captures above. Anyone care to hypothesize?

Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Mon May 20, 2019 10:44 pm
by w-u-2-o
My guess would be buffers associated with the stream to the CODEC on the SDR board are slowly filling up, whereas the buffers associated with the ASIO stream are being kept reliably empty.

Re: Presonus Studio 192 Mobile

Posted: Tue May 21, 2019 12:56 am
by w9ac
w-u-2-o wrote:My guess would be buffers associated with the stream to the CODEC on the SDR board are slowly filling up, whereas the buffers associated with the ASIO stream are being kept reliably empty.

Thanks Scott. Perhaps a future release of Thetis could flush and refresh the CODEC buffer at periodic intervals if it can be accomplished unobtrusively.
Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Tue May 21, 2019 11:34 am
by W4WMT
Can you tell which one is creeping?

Re: Presonus Studio 192 Mobile

Posted: Tue May 21, 2019 12:00 pm
by w9ac
Bryan W4WMT wrote:Can you tell which one is creeping?

Latency from the Presonus output always remains constant (the lavender trace). But latency from the ANAN headphone jack (yellow trace) increases to the delay shown in the first set of screen captures.

Paul, W9AC

Re: Presonus Studio 192 Mobile

Posted: Tue May 21, 2019 2:57 pm
by w-u-2-o
Paul,

I got hit with a Windows update today and almost forgot to add this cautionary note: I've had the last few Windows updates stall at restart, with the a black screen and spinning dots forever. Two updates ago I figured out that there is something about the Presonus driver that causes this. If you ever find a Windows update induced restart stalled like that, simply disconnect the USB3 cable from the Presonus and that should do the trick, the dots should stop spinning and the restart will continue. Once the restart and update is complete you can plug the Presonus back in.

Normal system restarts appear unaffected, of course.

73!

Scott

Re: Presonus Studio 192 Mobile

Posted: Wed May 22, 2019 3:59 am
by w9ac
w-u-2-o wrote:Paul,

I got hit with a Windows update today and almost forgot to add this cautionary note: I've had the last few Windows updates stall at restart, with the a black screen and spinning dots forever. Two updates ago I figured out that there is something about the Presonus driver that causes this. If you ever find a Windows update induced restart stalled like that, simply disconnect the USB3 cable from the Presonus and that should do the trick, the dots should stop spinning and the restart will continue. Once the restart and update is complete you can plug the Presonus back in.

Normal system restarts appear unaffected, of course.

73!

Scott

Scott,

Thanks for the tip. I also had to clean up a mess after a recent Windows update. The first problem showed up in a latency monitor program. The fix was to re-install a NIC driver from the Intel website.

The second problem showed up when the mouse cursor started jumping around the screen. The root cause is a compatibility issue between Windows and a FTDI USB-to-Serial adapter driver. The "serial enumerator" box must be unchecked or the mouse pointer takes on a life of its own.

Anyway, I have to revisit this stuff after most Windows updates. At my remote site, all updates are completely shut off. That took some work because it requires several system registry changes to kill all updates and "phone home" background apps. The PC is used only as a server so there's no outbound traffic generated at the site. As such, there's no need for any Windows updates. In fact, there's zero virus and malware protection. Not needed in this application. The router firewall has been adequate. I also have an identical PC preconfigured that's sitting in a box at the site. Worst case scenario is I take out one PC and roll in the other with very little down time due to the back-up PC being preconfigured.

Paul, W9AC