Thetis v2.8.11 - UPDATES and current issues

[ON7WP]

_HPSDR hi-res IQ recording.pdf

(650.47 KiB) Downloaded 171 times

In Protocol1 / PowerSDR an American HANM programmed the so called hidden "Scientific mode" that was used to do some research on NVIS propagation on multiple antennas streaming the RX data to harddisk

It would be great of this could be implemented in Thetis too, enabling to continue scientific on propagation and antenna research.

Some info enclosed

[ramdor]

Richie--probably too late for the impending release, but a simple option to simply turn off SEQ error reporting might be worth it. If someone was having audible artifacts they could turn it on to see how bad bad is.

yes, I will add an option to ignore them on the infoBar in the next version.

R.

[w9mdb]

I've started seeing ADC1 Overload! messages on transmit with my 7000DLE MKII on the last couple of versions....not sure exactly when it started.
This is with JTDX and VBCable. I had this same problem with the ANAN 100 at one point.

Even reducing JTDX to 5W output power with the audio slider would not get rid of the error. I do have RX2 muted on TX (at least the box is checked).

I put a 50Ohm plug on the Ant2 connector and that seems to have cleared it up.

[w-u-2-o]

ADC overload messages are not an indicator of a bug, but rather system configuration or station configuration issues.

[ramdor]

...and ADC1 in the warning is actually ADC0 when referencing the setup ADC tab.

Previously, 31dB of 'TX' attenuation was always applied when PS-A was off OR a CW mode was used in TX. In other words, whatever a user had set for the TX atten would always be over-written by 31dB.

It has been changed, to only apply 31dB when PS-A is off AND a CW mode is used.

TL;DR; increase your TX attenuation assuming you are not using PS-A. You can test this by dumping a tune/carrier into a DL at say 100w, with max TX atten, then gradually dropping it. When I reach around 6dB attenuation, the adc1 warning is shown.

R.

[sv1jso]

Richie could you please let us know why the change in AND behavior ?
For me it is a problem when I use ssb with linear and then I switch to digital with higher anan power (without linear).

Regards

Mike SV1JSO

[ramdor]

I thought I did.

In other words, whatever a user had set for the TX atten would always be over-written by 31dB.

It perhaps needs a checkbox to force it always for such cases.

Richie.

[ULTIMAX]

Richie--probably too late for the impending release, but a simple option to simply turn off SEQ error reporting might be worth it. If someone was having audible artifacts they could turn it on to see how bad bad is.

yes, I will add an option to ignore them on the infoBar in the next version.

R.

Thank you Richie you're the best buddy 73's

[DL2XY]

Regarding VAC issues (Buffer misses) and maybe sequence errors in latest 21k9.

After running 21k9 for a longer time I experienced a distinct higher number of VAC warnings and seldom sequence errors.
The probability is about one to two events an hour on my system, but i heard from other OM's with much shorter intervals.

On such an event Thetis display is freezing and sound is muted for a very short time (20ms to 200ms) without sequence error but rarely up to 1s with sequence errors.

Maybe a hidden race condition or thread deadlock?

This behavior was introduced with rc5(b?), rc4b and before where not affected.

sorry for inconvenience,
Walter

[Joe]

I also noticed the more frequent vac instability as I had reported earlier but did not pay too much attention as I raised the timing on the ring buffer setting, I actually thought that the problem was a windows update in my case but not sure? Thanks for all of the hard work.

73’s
Joe
WD5Y

[oe3ide]

Very interesting. I run Thetis for over 48h during the CQ WPX contest last weekend. Not one VAC over/underrun, no SEQ errors...

73 Ernst

[ramdor]

Very interesting. I run Thetis for over 48h during the CQ WPX contest last weekend. Not one VAC over/underrun, no SEQ errors...

73 Ernst

Same here. I am currently 4hrs into a non stop session to close out a special event callsign. I have not one single SEQ error, or VAC under/over logged. I have however noticed a very slow memory leak, which I will be attempting to track tomorrow.

Richie.

[w-u-2-o]

Not sure exactly how many hours, but station has been running non-stop for somewhere between 24 and 36 hours. No SEQ errors. An inconsequential number of VAC under/overruns (less than 10).

21K9, running real-time priority on Thetis, dedicated NIC, NIC metric favors dedicated NIC, network throttle index tweak ON.

[K1LSB]

Richie,

Not sure if this is a software issue or not, but my mouse wheel doesn't have the desired effect on any of the slide controls.

If I click on any slide control (Master AF for example) and drag it wherever I want with the cursor then try to use the mouse wheel for further adjustments, the slide only moves a single notch in the desired direction then stops, regardless of how much further I spin the wheel.

However, if I position my cursor in the spectrum scope the mouse wheel will happily move the receiver whichever direction and however far I care to spin the wheel, so I know Thetis can detect how much I'm spinning the wheel. So I'm clueless as to why the sliders only move one notch then stop?

The behavior described has been observed in every version of Thetis for the last two years at least.

Thank you for everything,
Mark

[W9BHI]

All of my sliders work with the mouse wheel.
Running 7000DLE blue face, 21k9b, P2, Win 10.
No seq. errors.
So far, best Thetis version yet.
Don

[ramdor]

Richie,

Not sure if this is a software issue or not, but my mouse wheel doesn't have the desired effect on any of the slide controls.

If I click on any slide control (Master AF for example) and drag it wherever I want with the cursor then try to use the mouse wheel for further adjustments, the slide only moves a single notch in the desired direction then stops, regardless of how much further I spin the wheel.

However, if I position my cursor in the spectrum scope the mouse wheel will happily move the receiver whichever direction and however far I care to spin the wheel, so I know Thetis can detect how much I'm spinning the wheel. So I'm clueless as to why the sliders only move one notch then stop?

The behavior described has been observed in every version of Thetis for the last two years at least.

Thank you for everything,
Mark

Mouse settings in windows
Scrolling
Scroll inactive windows when hovering over them <--- make sure that is ON

Richie.

[K1LSB]

Thank you so much Richie, that fixed it!

Mark

[ramdor]

Memory Leak

Ok, there is a memory leak in Thetis when CPU% display is set to [Thetis Only]. It has existed in windows core/.net for a very long time. It seems to leak with specific counters, and a counter relating to a process name is one of them. In my testing it is around 300MB over 4-5hrs.

I do not have a fix at this time, so I would advise setting the CPU% display to SYSTEM. You can do this by clicking on the % on the status bar.

Thanks, Richie.

systemOnly.png (5.72 KiB) Viewed 9374 times

[DG9FDA]

Screenshot 2022-03-29 174152.png (24.48 KiB) Viewed 9364 times

Using Tetis for a long time and love it! Everything is working fine with 21k9b. I miss the CPU load information. Under General - H/W Select I see "PLL Not Locked. What is my problem?

Thanks, Georg DG9FDA

[ramdor]

Using Tetis for a long time and love it! Everything is working fine with 21k9b. I miss the CPU load information. Under General - H/W Select I see "PLL Not Locked. What is my problem?

cpu% will not be shown if your user lacks the permission to obtain the performance counter, and/or, your operating system does not provide it.

Not sure about pll not clocked. The info is returned as part of the high priority status packet. It should show the correct status in P2 when Thetis/radio is powered on/connected.

Richie.

[w-u-2-o]

Richie--will the CPU display be made to default to "system" in the upcoming production release?

[ramdor]

Richie--will the CPU display be made to default to "system" in the upcoming production release?

yes it does already. I am tempted to do a quick patch to send to Doug to remove the feature as the leak might actually be one of those things I can't fix. I will have to log a report with MS after building a small test bed.

R.

[w-u-2-o]

Removing it might be the best thing to do. There are plenty of other tools to measure CPU consumption: Task Manager and ResMon are built into Windows, and Process Explorer is an easy addition. All three will give different results, though

[DG9FDA]

Using Tetis for a long time and love it! Everything is working fine with 21k9b. I miss the CPU load information. Under General - H/W Select I see "PLL Not Locked. What is my problem?

cpu% will not be shown if your user lacks the permission to obtain the performance counter, and/or, your operating system does not provide it.

Not sure about pll not clocked. The info is returned as part of the high priority status packet. It should show the correct status in P2 when Thetis/radio is powered on/connected.

Richie.

No solution for cpu% found. I am not an Windows specialist...

PLL not locked was fixed with newest firmware 12.1.8. (replaced12.1 pre4). PLL not locked produced a very rough tx signal (FM on RF).

[DG9FDA]

Using Tetis for a long time and love it! Everything is working fine with 21k9b. I miss the CPU load information. Under General - H/W Select I see "PLL Not Locked. What is my problem?

cpu% will not be shown if your user lacks the permission to obtain the performance counter, and/or, your operating system does not provide it.

Not sure about pll not clocked. The info is returned as part of the high priority status packet. It should show the correct status in P2 when Thetis/radio is powered on/connected.

Richie.

No solution for cpu% found. I am not an Windows specialist...

PLL not locked was fixed with newest firmware 12.1.8. (replaced12.1 pre4). PLL not locked produced a very rough tx signal (FM on RF).

Thanks to Erwin, DO2ZA I could fix the cpu% / performance counter problem: http://www.w1aex.com/anan/anan.html#counters

[K9CDR]

Richie--will the CPU display be made to default to "system" in the upcoming production release?

yes it does already. I am tempted to do a quick patch to send to Doug to remove the feature as the leak might actually be one of those things I can't fix. I will have to log a report with MS after building a small test bed.

R.

@Ramdor and all contributors to this software-- a huge THANK YOU to all the hard work and hours you guys put into this that I for sure greatly appreciate as im sure most of us do! Is there a beer fund I can contribute to for you guys? I'd love to get the next round! I can't imagine the time and effort that's it taken to get this where it is and the direction it's going so thanks boys! Appreciate it all!

[ramdor]

UPDATE 21k9c

Post 1 has been updated.

Just a couple of changes, that could do with testing

BTW, since 21k9 you will be notified if you need to hold CTRL to update the database. I have therefore removed all instructions regarding this from the first post to minimise confusion.

Thanks, Richie.

21k9c
fix: cursor NorthSouth was being shown outside panafall mode (the splitter drag cursor)
fix: N1MM offset issue fixed (due to lack of accuracy in low/high frequency info sent with the n1mm xml data)
add: infoBar hide SEQ error option (option-2 tab)
change: some changes to the init & resize directX code
change: system only display of cpu% (Thetis only has memory leak and is under investigation)
change: Kenwood AI command broadcast functions moved into delegate system
change: PA volts + amps samples are made more 'uniform' over the sample period

[Joe]

Thanks Richie,

All loaded just now, looks great! I got it quick been waiting and checking alot!

Thanks,
Joe
WD5Y

[Tony EI7BMB]

Thanks for the update Richie. Skin W2NBC scope mesh (customised) got messed up with the panadapter background colour so I'll stay with K9b for now.

[ramdor]

Thanks for the update Richie. Skin W2NBC scope mesh (customised) got messed up with the panadapter background colour so I'll stay with K9b for now.

Do you have an alpha set on the background colour to make it 'see through' so you can see the skin as the background gets drawn over the top of the skin background.

Richie.

testW2NBC.png (529.62 KiB) Viewed 8902 times