Thetis QSK Delayed Sidetone

User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Thetis QSK Delayed Sidetone

Postby W4ATL » Thu Nov 11, 2021 6:23 pm

When using QSK with Thetis you have to turn off the Sidetone so that only the routed TX signal tone comes through. But it is delayed making it difficult to use paddles at high speed. Is there a way to turn off the routed TX signal and use only the sidetone that is not delayed?
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Sat Dec 04, 2021 4:24 pm

The delay can be reduced by going into Setup/DSP and setting Filter Size/CW = 1024. This is at the expense of filter sharpness but to me it is worth it. The delay is still there but much improved. Perhaps a lower number can eventually be implemented.
W4WMT
Posts: 209
Joined: Sun Apr 09, 2017 10:12 pm

Re: Thetis QSK Delayed Sidetone

Postby W4WMT » Sat Dec 04, 2021 10:54 pm

W4ATL wrote:Is there a way to turn off the routed TX signal and use only the sidetone that is not delayed?


The only way to turn off the routed TX signal would be to mute the receiver during each cw element. Since all cw generation is now being done in the fpga, there would not be a software solution to do what you ask. I believe there's a way to do this in gateware, but that's waaay above my pay grade. Perhaps there are some bright & talented GA Tech students, that have some recently acquired OpenHPSDR derived hardware in their club station, that could engineer a solution?

73!
User avatar
w-u-2-o
Posts: 4152
Joined: Fri Mar 10, 2017 1:47 pm

Re: Thetis QSK Delayed Sidetone

Postby w-u-2-o » Sun Dec 05, 2021 4:08 pm

Firmware generated sidetone should still be working perfectly. If you attach a speaker or headphone to the ANAN hardware, and a key also, then things should be just peachy.

The problem is with sidetone via VAC. The QSK full duplex fix is a Band-Aid, albeit a very clever and welcome one. Fixing this properly is conceptually trivial, with low technical risk, and requires no firmware development. If we require the key to be attached to the PC, something the software already supports very well, sidetone can be injected into the VAC audio output with extremely low latency, almost certainly less than 10ms.

If one wants to attach a key to the ANAN hardware and listen via VAC (not sure why you would want to, but still...), even that does not pose a huge technical challenge. Flagging key down/key up events in the high priority UDP data stream that already comes from the hardware would allow very low latency sidetone in the VAC audio stream using the VAC sidetone injection proposed above. However this would require firmware changes.

All that is needed is someone who wants to tackle low latency sidetone injection into VAC, and operators willing to attach keys to their PCs via the serial port mechanism already provided.

Philosophically, it was never clear why the original design went in the direction that it did. This being software defined radio and a thick-client architecture, it never made sense to add audio and key/paddle interfaces to the RF hardware unit. Perhaps there was a belief that these things could not be implemented with reasonably low latency with normal PC interfaces, but now we know that is not the case.

BREAK

Why do you refer to firmware as "gateware"?
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Thu Feb 24, 2022 3:21 am

w-u-2-o wrote:If one wants to attach a key to the ANAN hardware and listen via VAC (not sure why you would want to, but still...),

In CW QSK Mode, I'm listening to VAC with audio being ported from a Presonus Studio 192 Mobile to my Presonus StudioLive 16R mixer. Latency is less than at the ANAN 7000'S headphone jack.

Presently, when listening to CW in QSK mode at the headphone jack, latency builds over time. Other CW users (incl. W2PA and N4LQ) have experienced the same with no resolution yet found. The only practical solution is to stop Thetis, then re-start it. After a few minutes, hand-ear coordination is slowly lost at fast operating speed. Stopping and re-starting Thetis is a no-go in the middle of a QSO.

With Worldclock active and buffers run down to zero, latency is very low with ASIO VAC engaged -- so low that I can have a QSO at 45 WPM and still maintain hand-ear coordination.

When in QSK mode, we're hearing our own transmitted CW, just like any other signal on the band. For die-hard CW ops, it's the Holy Grail. Sending at QRQ speed with paddles requires really low latency and this setup delivers. I wouldn't want it any other way.

Paul, W9AC
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Fri Feb 25, 2022 1:19 pm

What ASIO capable box would you recommend I try?
User avatar
w-u-2-o
Posts: 4152
Joined: Fri Mar 10, 2017 1:47 pm

Re: Thetis QSK Delayed Sidetone

Postby w-u-2-o » Fri Feb 25, 2022 2:12 pm

They are all very similar. It's easy to suspect they mostly share the same chipsets. The biggest discriminator, and contributor to latency, is the quality and adjustability of the driver software. IMHO the best driver software comes from Behringer and Presonus. As luck would have it, the Behringer UMC202HD is pretty much the lowest cost solution. The Presonus Studio 24c 2x2 is also a very good choice. Although more expensive it comes with some very nice DAW software. You can find used ones on eBay all the time for $120'ish.

I personally have a Presonus Studio 192 Mobile. This is no longer made, unfortunately. When it was available it was the lowest cost solution for an interface with an external Word Clock (audio engineer-speak for external clock reference) input. They are highly prized for this and still command a price of nearly $300 on eBay.
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Thu Apr 21, 2022 9:05 pm

I have successfully implemented QSK CW, using a paddle with minimum delay in the sidetone. The secret is to use an external audio box, such as a Behringer UMC202HD, for low latency audio from the PC running Thetis to your headphones or speakers. One advantage is that I now have a separate volume control for my headphones while I have Fostex 6301NB monitor speakers connected to the Behringer.

The resulting audio is outstanding and the delay in the TX tone when keying is minimal even at 30 WPM. Sending with paddles is no longer a problem and the resulting QSK is perfect.

This post is a good start when using an external ASIO sound board.

https://apache-labs.com/community/viewtopic.php?f=14&t=3920

And of course read this.

https://apache-labs.com/community/viewtopic.php?f=14&t=4149

But you need the secret ASIO box to make it work well with paddles.

The default AGC settings when going to QSK mode did not work for me as I detected some AGC pumping on strong signals. I increased the Decay to 5 ms and that got rid of the pumping without affecting QSK.

It is important to make sure all the Buffer Latency settings are set to 0 ms in Settings/Audio/VAC1.

Be aware that you need to engage VAC1 and you will not have RX2 audio with the existing stock Channelmaster.DLL. If you try this and like it, you can download a custom DLL that will route audio to your ASIO device and you will have RX1 and RX2. You can also engage the VAC for data modes such as FT8 and have that VAC audio routed to the sound device connected to WSJT. The downside to the DLL is you lose the ability to get audio from the Apache Anan but you won't need it. Another downside is that I've lost my sidetone in semi break-in. I think that is generated in the hardware. That is not a problem since I exclusively use QSK.

Get the new DLL here.

https://www.dropbox.com/sh/rz26sn8oigg4wnz/AAAnvcWukSG0pIF1ZCwHXF0Fa?dl=0

Go to C:\Program Files\OpenHPSDR\Thetis and rename the existing Channelmaster.DLL to something like Channelmaster_OLD.DLL and put your new Channelmaster.DLL in that directory. Restart Thetis.

You may need to adjust buffer settings and the AGC for your particular computer. Initially I had distorted TX audio and I cleared it by increasing the RX buffer size. I was not able to reproduce the distortion and went back to the W9AC buffer settings.

Thanks to W4WMT and W9AC for their posts and help.
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Thu Apr 21, 2022 10:55 pm

>"The resulting audio is outstanding and the delay in the TX tone when keying is minimal even at 30 WPM. Sending with paddles is no longer a problem and the resulting QSK is perfect."

Amazing, isn't it?

>"Another downside is that I've lost my sidetone in semi break-in. I think that is generated in the hardware."

Right. Two changes need to occur when switching from QSK to SEMI: (1) sidetone needs to be checked; and (2) CW delays need adjusting.

I've requested that the delay settings independently follow the CW mode but in addition, the sidetone checkbox should be automatically set as an option when switching from QSK to SEMI mode.

Like you, I don't use CW semi-break-in, but many CW ops can only operate that way.

Paul, W9AC
W4WMT
Posts: 209
Joined: Sun Apr 09, 2017 10:12 pm

Re: Thetis QSK Delayed Sidetone

Postby W4WMT » Fri Apr 22, 2022 11:54 am

w9ac wrote:Like you, I don't use CW semi-break-in, but many CW ops can only operate that way.

True!
But that notwithstanding, wouldn't those ops be better off just increasing the "CW Break-In" delay to effect the same timing as semi? That way you'd have the best of both worlds, with that silky full-duplex monitoring! Sidetone is so 1980's :-)
73 de Bryan W4WMT
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Fri Apr 22, 2022 12:09 pm

W4WMT wrote:
w9ac wrote:Like you, I don't use CW semi-break-in, but many CW ops can only operate that way.

True!
But that notwithstanding, wouldn't those ops be better off just increasing the "CW Break-In" delay to effect the same timing as semi? That way you'd have the best of both worlds, with that silky full-duplex monitoring! Sidetone is so 1980's :-)
73 de Bryan W4WMT

Yes! That's yet another possibility. Selecting Semi would then be accomplished with a high delay value in <Setup><DSP><CW>.

However, there's still a sidetone latency creep issue in QSK unless plunging into ASIO audio. Many ops may not want to go to that extent. Hopefully, that bug can be killed in the future.

The screen capture below was taken a couple years ago but the effect remains the same today. What's shown is receiver audio latency creep that builds up over time when using my ANAN 7000's internal CODEC.

Each vertical division is 20 ms. The green trace shows CW keyer closure (pulsed logic-high on closure). The yellow scope trace shows approximately 70 ms. of sidetone delay -- an eternity with QRQ ops. The magenta trace shows latency from my Presonus 192 with ASIO audio. ASIO latency remains constant over time and I've no hand-ear timing issues when sending at 40 WPM.

To reset the internal CODEC requires stopping, then re-starting Thetis. All is well until perhaps 15-20 minutes later when latency build again.

Paul, W9AC
Attachments
SDS00003.jpg
SDS00003.jpg (107.58 KiB) Viewed 2889 times
VE6IV
Posts: 5
Joined: Thu Mar 12, 2020 10:07 pm

Re: Thetis QSK Delayed Sidetone

Postby VE6IV » Tue Apr 26, 2022 4:39 am

Just to clarify, can you run a "fast break-in" config in QSK mode using a sufficient delay and use the ASIO driver with the Behringer outboard interface (UMC202HD)?
I am not a fan of full QSK but will need to avoid this latency creep during contests. Sure hope this is being looked at by the powers that be.

Kirb VE6IV
(My 7000dle mkII arrived today!!)
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Tue Apr 26, 2022 9:45 am

VE6IV wrote:Just to clarify, can you run a "fast break-in" config in QSK mode using a sufficient delay and use the ASIO driver with the Behringer outboard interface (UMC202HD)?
I am not a fan of full QSK but will need to avoid this latency creep during contests. Sure hope this is being looked at by the powers that be.

Kirb VE6IV
(My 7000dle mkII arrived today!!)


If you want to run Semi Break-In you don't need to do all the stuff I've outlined here. You can just attach your headphones or speakers directly to the 7000. The Sidetone in Semi Break-in is not delayed at all in this configuration.

If you are concerned about getting the lowest receive latency possible, then the Behringer is the way to go.
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Tue Apr 26, 2022 10:55 am

>"Just to clarify, can you run a "fast break-in" config in QSK mode using a sufficient delay and use the ASIO driver with the Behringer outboard interface (UMC202HD)?"

Yes. In Thetis, you would set the desired CW break-in time under <Setup> <DSP><CW>. This will create the semi-break-in effect while retaining the attributes of ASIO through your UMC202HD. The sidetone will be your own transmitted CW signal.

Or, as W4ATL suggests, simply use semi-break-in mode. In this case, the sidetone is internally generated just as it is with any other transceiver.

Paul, W9AC
VE6IV
Posts: 5
Joined: Thu Mar 12, 2020 10:07 pm

Re: Thetis QSK Delayed Sidetone

Postby VE6IV » Sun May 01, 2022 3:34 am

Thanks guys. Appreciate the clarification.

Kirb, VE6IV
VE6IV
Posts: 5
Joined: Thu Mar 12, 2020 10:07 pm

Re: Thetis QSK Delayed Sidetone

Postby VE6IV » Fri May 06, 2022 4:45 am

Other than the want/wish postings on the Thetis sub-forum, is there a bug report repository somewhere? I'm just wondering where these issues are being captured and prioritized, i.e latency creep, sidetone record/playback in semi break-in bug etc.
It would be nice to see that list with a bunch of check marks against said items, indicating someone is looking at addressing the problem(s).

Kirb - VE6IV
User avatar
w-u-2-o
Posts: 4152
Joined: Fri Mar 10, 2017 1:47 pm

Re: Thetis QSK Delayed Sidetone

Postby w-u-2-o » Fri May 06, 2022 11:28 am

VE6IV wrote:Other than the want/wish postings on the Thetis sub-forum, is there a bug report repository somewhere? I'm just wondering where these issues are being captured and prioritized, i.e latency creep, sidetone record/playback in semi break-in bug etc.
It would be nice to see that list with a bunch of check marks against said items, indicating someone is looking at addressing the problem(s)

Kirb,

This isn't your normal open source development environment where there are a lot of busy contributors. The total number of developers who have worked on Thetis and its predecessor can be counted on the fingers of one, maybe two hands. The number of developers actively working at any given time is approximately equal to 1, and many times this number goes to 0 for long periods.

Also, historically there has never been an effort at making a true, collaborative, development environment. While source and installers are stored on GitHub they are not organized to support multiple branches of development, and previous developers have always developed essentially privately.

However, Richie has made some initial overtures in this area. As part of that he has stood up a GitHub issue tracker, but I think this is primarily for issues he is tracking and is interested in tackling, not necessarily all issues:

https://github.com/ramdor/Thetis-2.9.0/issues

73,

Scott
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Mon Jul 04, 2022 5:02 pm

I successfully implemented the W9AC modification for QSK on my 7000DLE MKII which involves replacing K5 on the PA Board with a reed relay and installing a circuit board that opens up the relay coil lines to RL17 and RL19. It appears the relay suggested by W9AC is out of production or not in stock so I used an alternate - AXICOM V23100V4312C000.

I decided to cut the trace from pin1 on both RL17 and RL19 to their bypass capacitors and solder lines from the W9AC circuit to the pads on each side of the cut trace.

QSK CW operation is now totally silent. The only way I can tell if I'm transmitting is to look at the power output on my wattmeter. I've made several QSOs and the rig is a pleasure to use on CW.
dl4zbg
Posts: 101
Joined: Sun Apr 09, 2017 5:33 pm

Re: Thetis QSK Delayed Sidetone

Postby dl4zbg » Mon Jul 04, 2022 7:47 pm

Can you share photos of the installation?

73

Volker
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Mon Jul 04, 2022 10:17 pm

I've got photos of the 7000 disassembly but then I forgot to take pics of the installed board and the new relay.
User avatar
W4ATL
Posts: 10
Joined: Tue Mar 09, 2021 9:14 pm

Re: Thetis QSK Delayed Sidetone

Postby W4ATL » Tue Jul 05, 2022 11:47 am

I went back and looked at my pics and I did take a picture of the new relay. It is mounted upside down with some silicone RTV. Wires run from the pads of the old relay to the pins of the new reed relay.

Old relay (Orange).
PXL_20220702_180853220.jpg
PXL_20220702_180853220.jpg (690.2 KiB) Viewed 903 times


New Reed Relay mounted upside down and before I soldered them to the pins.
PXL_20220702_190442500.jpg
PXL_20220702_190442500.jpg (519.78 KiB) Viewed 903 times
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Tue Jul 05, 2022 1:09 pm

W4ATL wrote:I went back and looked at my pics and I did take a picture of the new relay. It is mounted upside down with some silicone RTV. Wires run from the pads of the old relay to the pins of the new reed relay.

Well done. As W4ATL described, there's zero mechanical relay noise after the mod; it's as quiet as PIN diode switching. I have several QSK amplifiers including Alpha 87A and 89 models, but what gets used most is an SPE 1K-FA model with a PIN diode modification. All Thetis keying delays are set to zero with no hot-switching.

Future Apache Labs products should isolate the T/R relay coil from all others. By current design, the T/R coil is shared with the PureSignal function, thus unnecessarily activating PureSignal relays when transmitting in CW mode. Moreover, the T/R relay could easily be shock-mounted in production. Ten Tec did with the Omni 7, so could Apache Labs.

Paul, W9AC
dl4zbg
Posts: 101
Joined: Sun Apr 09, 2017 5:33 pm

Re: Thetis QSK Delayed Sidetone

Postby dl4zbg » Tue Jul 05, 2022 8:56 pm

Thanks for the fotos. With these kind of relays I managed to switch 5 watts on 144MHz. I would not dare to use them for 100 watts. What do you think how long they will survive?

cheers

Volker
User avatar
w9ac
Posts: 191
Joined: Sun Apr 09, 2017 4:01 pm

Re: Thetis QSK Delayed Sidetone

Postby w9ac » Tue Jul 05, 2022 9:16 pm

dl4zbg wrote:Thanks for the fotos. With these kind of relays I managed to switch 5 watts on 144MHz. I would not dare to use them for 100 watts. What do you think how long they will survive?

cheers

Volker

It's worth a try and easy enough to replace if it fails. And if it does, I'll send him an Aromat/Matsushita RSD-12V that's meant for the job and equally as quiet.

Paul, W9AC

Return to “Keys, Keyers, Paddles & Bugs”