Keyboard Shortcuts
Likes
- SoftwareControlledHamRadio
- Messages
Search
Re: MF changes
?
Hi Bill,
I'm just about to start populating a Main board that has the SGTL5000 and the SPI driver onboard. I've also done away with the 3.5mm Jacks and replaced them with JST connectors, along with moving the regulators and the shutdown circuitry to a separate PS board.
I've no idea if it'll all work out, but that's the beauty of being able to experiment and do your own thing!
?
Rob G7WKE
?
?
? |
Re: MF changes
I like the idea of breaking the LO onto a daughter board that mounts to the RF board through header pins. That'll allow for more experimentation. -------- Original Message -------- On 5/27/25 11:48 AM, Tim via groups.io wrote:
|
Re: T41 Hilbert Filter Design
Thanks Oliver for the response curves.? Interesting to see.? Also, I had missed the sampling rate? on the v50 curve being 24 kHz.? I added a note to the index page to cover that.
?
Unless I am missing something, I conclude that there is no reason to pursue the double filter +/-45 degree case.? For the same computational effort you get better performance using the single filter 90 degree along with a delay.? Anybody see anything different?
?
Thanks for the discussion! |
Re: T41 Hilbert Filter Design
I made a production line of sorts and tested about 20 different Hilbert transform filters only to confirm what I learned during my DSP experiment days: "at a certain point, filter design is as much an art as a science".? I also realized that what I've been calling spurs are just harmonics of my input signal.? I'm not experienced enough to know the significance of those.? As seen, the Hilbert filter can be narrowed to eliminate them, but I know that for some, that has a negative effect on audio quality.
?
I think Bob's Hilbert 90 deg plus delay question remains open.? There are 90/0 degree coefficients in some versions of the software but these don't relate to what Bob's referring to.? Rather, they go back to the 90/0 degree phase shift that what was described in my edition of the T41 book.? There, the left channel was passed through a 90-degree Hilbert transform and the right through a 0-degree Hilbert transform.? The method Bob mentions, does the first transform, but just delays the right channel to compensate, avoiding the associated multiplications/additions.? As Bob showed, you get a better filter with the same number of computations.
?
Anyway, I started this thread wondering if the Hilbert filters were causing the poor signal I was seeing during transmit IQ calibration.? It turned out that was due to the dec/int low pass filters.? Oliver is working on those, so I'll return to my project of trying to automate the transmit IQ calibration.? If possible, I'd like to extend this to automatic calibration of all bands with one button press like I did for the receive IQ. |
Re: MF changes
¿ªÔÆÌåÓýMy apologies.? I¡¯ve been on the road since before Hamvention and unavailable for the most part.? I¡¯ll be back home as of Thursday AM and have a lot of catching up to do with orders from Hamvention (and a few prior) and documentation with the latest board revs.? Some of the changes to boards are so minor that I did not update the documentation for them.? On the other hand¡ a few more extensive changes were made to the RF board in the MF section that makes assembly much easier¡ however not documented yet (IN THE ASSEMBLY MANUALS ONLY¡ the schematics and BOMs are ALWAYS up to date¡ the assembly manuals take a complete review to update). ? I¡¯ll be making a few additional minor updates to the MAIN board in the next round of boards I order¡ the new ISO SPI driver will be incorporated into the MAIN board so the plug-in Display Driver board will then be built-in.? It¡¯s been a big success with over 250 kitted and sold to date. ? Here is some food for thought for next Hardware directions. ?
? I¡¯ve been looking into getting BPF and RF boards made with the small semi-conductor parts prepopulated.? On the RF board, that means the PE4302¡¯s and the si5351 would come pre-populated.? The rest of the parts are easy to solder.? The switches on the BPF would come pre-placed.? All at a very nominal cost (less than $0.50 per part plus part cost in the quantity I buy). ? Always looking for your thoughts. ? ? ? ? ? Dr. William J. Schmidt - K9HZ J62K(J68HZ) 8P6HK ZF2HZ PJ4/K9HZ VP5/K9HZ PJ2/K9HZ VP2EHZ ? Owner - Operator Big Signal Ranch ¨C K9ZC Staunton, Illinois ? Owner ¨C Operator Villa Grand Piton ¨C J68HZ Soufriere, St. Lucia W.I. Rent it: ? Moderator: North American QRO Group at Groups.IO. Moderator: Amateur Radio Builders Group at Groups.IO. ? email:? bill@... ? ? From: [email protected] <[email protected]> On Behalf Of David Fine
Sent: Monday, May 26, 2025 10:45 AM To: [email protected] Subject: [SoftwareControlledHamRadio] MF changes ? Can anyone tell me if the changes made regarding the MF frequencies are documented anywhere?? Were changes made in the BOMs or assembly manuals for the affected boards?? I have been unable to find them.? Thanks. Dave. W0DF |
Re: T41 Hilbert Filter Design
For comparison, here are plots of the responses of the various versions of the Hilbert filters.? Here's the version that is currently in the code (FIR_Hilbert_coeffs_45,FIR_Hilbert_coeffs_neg_45): There were two sets of coefficients in the V050 code, one with 0/90 phase shifts (FIR_Hilbert_coeffs90,FIR_Hilbert_coeffs0; note the change in sample rate to 24 ksps): And one with +45/-45 phase shifts (FIR_Hilbert_coeffs_45,FIR_Hilbert_coeffs_neg45):
On Tuesday, May 27th, 2025 at 4:24 AM, Terrance Robertson, KN6ZDE via groups.io <tmrob4@...> wrote:
|
Re: T41 Hilbert Filter Design
Thanks.? Great program, though it doesn't totally work in my version of Octave.? I made some edits to get it mostly working.? That explains one thing.? It looks like the narrow response cuts off the spurs I'm seeing with the wider response.? I'll try that tomorrow.? Note, you used the 90/0 set from v50.? The +/-45 follow those but also give a similar response. |
Re: T41 Hilbert Filter Design
I posted the Octave .m files, the three sideband suppression curves, the coefficients for the 199 tap plus delay and the Hilbert design files.? They are at
?
Here is code for the OA_F32 code. It does not directly apply, but is an example of the 90 degree plus delay:
https://github.com/chipaudette/OpenAudio_ArduinoLibrary/blob/master/AudioFilter90Deg_F32.h
https://github.com/chipaudette/OpenAudio_ArduinoLibrary/blob/master/AudioFilter90Deg_F32.cpp
?
73, Bob? W7PUA
? |
Re: MF changes
If you have the latest set of boards from Bill, you'll see that the clock divider circuitry is no longer present on the board -- all that's left are two jumpers and four zero-ohm resistors: The prior generation of boards still had the clock divider components: If you have an older board, all you need to populate in the clock divider section is the zero ohm resistors (R35, R37, R40, R42) and the two jumpers (JP18 and JP19).
On Monday, May 26th, 2025 at 11:45 AM, David Fine <dfine100@...> wrote:
|
Re: T41 Hilbert Filter Design
I calculated the passbands for the Hilbert filter coefficients currently in the V066-9 code (FIR_Hilbert_coeffs_45 and FIR_Hilbert_coeffs_neg_45). Their passbands don't have the same cutoff frequency: The phase shift between them is 90 degrees up to about 5.5 kHz. Tomorrow (well, this week's) project is to figure out the process to generate a new set of filter coefficients using the scipy library in Python. I don't have a copy of the Iowa Hills software (though I did find the DSP code on github) and need a better long-term solution for this and future projects.
On Monday, May 26th, 2025 at 6:52 PM, Terrance Robertson, KN6ZDE via groups.io <tmrob4@...> wrote:
|
Re: T41 Hilbert Filter Design
Here's my new transmit chain with the v66-9 Hilbert coefficients.
?
?
Close in, I can see what Bob has found, but there are two spurs that aren't there with the older Hilbert coefficients.? Here they are side by side, v66-9 Hilbert coefficients left v50 right.
?
?
Bob, I'll try your 90/0 coefficients if you post them. |
Re: T41 Hilbert Filter Design
On Mon, May 26, 2025 at 03:01 PM, Bob Larkin wrote:
No, the v50 Hilbert pair from fir.cpp has barely 50 dB opposite sideband rejection, and that is only from 200 to 2300 Hz (at 12ksps).? Above that it is not good at all and so would need serious filtering to not transmit splatter or receive spurs.Hmm, that doesn't seem consistent with what I post about yesterday, but I was looking at the whole chain.? Perhaps I'll see something better circling back to the v66-9 Hilbert filter coefficients again with the new dec/int filters.? |
Re: T41 Hilbert Filter Design
Here is the result of my new DSP transmit chain, it's an improvement over the improved one I posted yesterday, perhaps a tad less than half the bandwidth used.
?
?
Because of the big improvement in the gain of the new dec/int filters, I was able to reduce various gains applied throughout the chain: microphone gain 20dB -> 10dB, scaling factors applied in?ExciterIQData 20 -> 2.6.? Note that this is before transmit IQ calibration.
?
|