apache-labs.com

DL2XY wrote:Just some Bugs:

Thetis v2.8.11 21d6b,ANAN100, WIN7 X64

• PSform Button Advance: Black Text on black background.
  
• Setupform/Display/RX2: Band indication is fixed on random Band.
  Settings are applied global to all bands.
  
• Console:
  Command line option "-datapath:" is broken. If used target path is ignored but "C:\" is used instead.
  This leads to database corruption if multiple instances are installed.

• the PSForm button. Advanced? not seeing black background here, checked the code, no colour change being made to btnPSAdvanced
• the disp->rx2 has been changed/reworked
• surely if you run mutliple instances and point it all to the same datapath, be it C:\ , D:\foo\ or as default C:\Users\<user>\AppData\Roaming\OpenHPSDR\Thetis then issues could occur at DB write time (last one shut down will be the one you keep). If you want multiple instances like that you'll need to use/also use -dbfilename. I could create a lockfile and check it etc at start up etc I guess, with option to clear it etc if there

DO2ZA Erwin wrote:ATT /S-Att on RX1 works perfekt, on RX 2 still higher Spectrum and S-Meter when I use the Att on RX2
73 Erwin

• If you use ADC0 for both RX1 and RX2 (ddc2 + ddc3 with ps/div off) then RX2 will be off as it will always attenuate ADC1. The result of this is that if you use RX2 attenuate, everything will move up the display in the RX2 area, because the display shift is being made but not the attenuate to ADC0 which is what you are looking at.
  
• If you use ADC1 for both RX1 and RX2 (ddc2 + ddc3 with ps/div off) then RX1 will be off as it will always attenuate ADC0. The result of this is that if you use RX1 attenuate, everything will move up the display in the RX1 area, because the display shift is being made but not the attenuate to ADC1 which is what you are looking at.
  
• If you use ADC1 for RX1 and ADC0 for RX2 (ddc2 + ddc3 with ps/div off) both RX1 and RX2 will be out, as they will each attenuate the incorrect ADC. Nothing will be right for either RX.

ramdor wrote:

DO2ZA Erwin wrote:ATT /S-Att on RX1 works perfekt, on RX 2 still higher Spectrum and S-Meter when I use the Att on RX2
73 Erwin

Yes, there is a big problem in there, and always has been. I tried to resolve it last week, but gave up on the idea after the number of issues it caused. I may look at it again at some point.

w-u-2-o wrote:Thanks for trying, Ritchie. You put in a ton of effort on it. It's most unfortunate that the code spaghetti factor was just to great for this one

Cheers!

Scott

ramdor wrote:

DL2XY wrote:Just some Bugs:

Thetis v2.8.11 21d6b,ANAN100, WIN7 X64

• PSform Button Advance: Black Text on black background.
  
• Setupform/Display/RX2: Band indication is fixed on random Band.
  Settings are applied global to all bands.
  
• Console:
  Command line option "-datapath:" is broken. If used target path is ignored but "C:\" is used instead.
  This leads to database corruption if multiple instances are installed.

Thanks for those Walter.

• the PSForm button. Advanced? not seeing black background here, checked the code, no colour change being made to btnPSAdvanced
• the disp->rx2 has been changed/reworked
• surely if you run mutliple instances and point it all to the same datapath, be it C:\ , D:\foo\ or as default C:\Users\<user>\AppData\Roaming\OpenHPSDR\Thetis then issues could occur at DB write time (last one shut down will be the one you keep). If you want multiple instances like that you'll need to use/also use -dbfilename. I could create a lockfile and check it etc at start up etc I guess, with option to clear it etc if there

Re. the VAC stuff above. Can you put a solution in place into the last tapr 2_8_11 source to calculate those values at runtime?

Cheers, R.

DL2XY wrote:
Thanks for response.

I found out it depends on Desktop Theme, I am using "Windows Classic" .
Please check PSForm "Advanced" Button in Form Designer. Property BackColor is set to "Web /Black", it should be " System/Control".
This bug exist since 2.8.11 release.

I do have distinct directories the datapath options are pointing to. But all instances are using "C:\".
This is not only for database.xml but for all files usually in datapath (ImportLog.txt, ke9ns8.dat, wdspWisdom00, SWL.csv, andromedadata.xml, CCALLib8WS.log).


I have got another one for you: Open WB and expand it's width to more than about 4300 pixel (you may need multiple monitors ) - Thetis crashes completely.

greets Walter

ramdor wrote:

DL2XY wrote:
Thanks for response.

I found out it depends on Desktop Theme, I am using "Windows Classic" .
Please check PSForm "Advanced" Button in Form Designer. Property BackColor is set to "Web /Black", it should be " System/Control".
This bug exist since 2.8.11 release.

I do have distinct directories the datapath options are pointing to. But all instances are using "C:\".
This is not only for database.xml but for all files usually in datapath (ImportLog.txt, ke9ns8.dat, wdspWisdom00, SWL.csv, andromedadata.xml, CCALLib8WS.log).


I have got another one for you: Open WB and expand it's width to more than about 4300 pixel (you may need multiple monitors ) - Thetis crashes completely.

greets Walter

button reset to defaults. Strange it was the only one there like that.

The datapath issue. Other than adding command line options for each of the different datafiles we dump into the datapath folder not much we can do. If you use the same datapath for more than one instance, there will be issues. Unless we add a lock file to that path and check its exitance there is no clean way to do that really, unless you want something that adds 1 or 2 or N depending how many instances are running, or some sort of instance ID on the command line, which is no different to providing a different datapath for each instance. Not sure I am understanding the issue tbh,

Yep, 4096 or around there is buffer size on that WB display probably.

R.

ramdor wrote:Please send me your command line argument(s), better still, a screenshot of your shortcut.
R.

ea3aqr wrote:New bug detected:

When clicking on FM, AM, SAM, DIGL or DIGU, VAC1 is automatically turned off. All other modes working fine.

DL2XY wrote:

2) VAC resampler initialisation issues (Scott)

The phase margin of the FLL control loop is supercritical, depending on buffersize and samplerate setting.
I had to adjust the proportional gain and slew rate in source.

The patch below give me stable conditions for my setting (48ksmp,10ms,512bytes).
This involves a slightly higher settling time (about 1 to 2 Min) but is perfect stable after one 10% overshoot.

For a perfect solution these values would have to be calculated at runtime according to setting.

73 Walter


file: wdsp/rmatch.c Line 501

void* create_rmatchV(int in_size, int out_size, int nom_inrate, int nom_outrate, int ringsize)
{
return (void*)create_rmatch (
1, // run
0, // input buffer, stuffed in other calls
0, // output buffer, stuffed in other calls
in_size, // input buffer size (complex samples)
out_size, // output buffer size (complex samples)
nom_inrate, // nominal input sample-rate
nom_outrate, // nominal output sample-rate
0.0, // fc_high (0.0 -> automatic)
-1.0, // fc_low (-1.0 -> no low cutoff)
1.0, // gain
3.0, // startup delay (seconds)
1, // automatic ring-size [not implemented yet]
ringsize, // ringsize
1024, // R, coefficient density
1.0, // initial variable ratio
4096, // feed-forward moving average min size
262144, // feed-forward moving average max size - POWER OF TWO!
0.01, // feed-forward exponential smoothing
4096, // proportional feedback min moving av ringsize
16384, // proportional feedback max moving av ringsize - POWER OF TWO!
1.0e-06, // (4e-6) // proportional feedback gain
1, // linearly interpolate cvar by sample
0.001 // (0.003) // slew time (seconds)
);
}

DL2XY wrote:

ea3aqr wrote:New bug detected:

When clicking on FM, AM, SAM, DIGL or DIGU, VAC1 is automatically turned off. All other modes working fine.

Switching modes implies automatic switching to TX-profile last used in new mode. Is VAC enabled in these profiles?

73 Walter

ramdor wrote:First stage I think, will be to add a 'visualiser' to the DDC tab, that will show RX1 RX2 PS Diversity and Sync DDC's in use, for both P1/P2 and the various models. Warren has given me some fantastic ddc/adc info from his spreadsheets that he keeps all the wdsp/cmaster/hardware routing. Adding that visualiser will help, and can be easily checked.

ramdor wrote:Added an option to highlight everything everywhere that is included in a TX profile. Figured it might be of use, at least so we can work out if too much/too little is being stored against them.

There may be a 21e version out at the end of the weekend which has a couple of features and fixes.

Richie.

w-u-2-o wrote:Richie: this is a brilliant idea. it will arguably provide 80% of what the Path Illustrator is supposed to do with a lot less code.

ramdor wrote:21e update released

Just a couple of fixes. Also, MNF support for transverters. Some changes to the hotkeys for MNF. See notes.

Cheers, Richie.

21e
fix: status bar item widths now forced in code as designer was not setting widths correctly check
fix: background colour fix for Advanced button in PS form check
add: band edges added for 1/2/3 region if you have those selected and extended turned off
add: ignore dupes option for bandstack filter added. If the frequency is already in the list, the slot you are leaving will not be updated at all check
add: RX assignments shown on ADC setup page. This is changed by being in different states, such as MOX/Diversity/PS check
add: Manual notch support for transverters. Notches limited by xvter coverage check
add: option to highlight anything that is stored as part of a TX profile check
change: active band display for RX2 grid min/max in setup changed to handle all bands check
change: change to the way bands are obtained from frequency for vfob/rx2 check
change: Use CTRL right click to add notch 200Hz wide. CTRL+SHIFT right click 100Hz wide. Changed from middle mouse when in crosshair mode. Still use middle to enabled/disable or shift middle to quick delete. Right mouse over a highlighted notch will show menu check