Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
- QRPLabs
- Messages
Search
|
I completed a QMX+ last week and the output powers are very similar to yours as you can see below. Measured power is from an external power meter and reported power is from the diagnostic menu on the rig.
One thing I noted was that the 10m output using the diagnostic tab is lower (and the power meter confirms) than that measured just keying the rig, I'm not sure if that is a feature or a setting I've messed up somewhere.
?
--
G0TXL (Paul) |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Another USB port question (requests?)
Hello Dick ?
Then I suppose Icom and others must not be properly switched OFF even when they seem they are. It's just sleeping but the USB port still active. QMX on the other hand is properly switched off, by hardware - it is not sleeping, the only current consumption will be nanoamps leakage through the OFF state MOSFET switch.? I could certainly add a command to power down QMX because this is done under firmware control, and it would gracefully switch off QMX including?saving the radio state etc. But powering up QMX again after an power cycle actually for real with your WiFi smart-socket thing, would require that QMX be switched on again by hardware.? I would suggest that this could be quite conveniently done using a small SPST relay, whose switch contacts were in parallel with the left rotary encoder push-button, and whose coil was energized from the supply voltage via a suitably sized capacitor in series with the coil, such that on application of power to the system, the relay coil activates for say half a second, to effectively switch on the QMX.? 73 Hans G0UPL |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Why The EER Method For SSB Generation Works
Hi Ian ?
Thanks :-)?? ?
Nice project, congratulations!? ?
I am sorry I don't have time to study all your code in detail but a few comments or things to think of.? 1) Certainly the '5351 responds quickly enough. Fred?WA7AII did some detailed measurements of PLL settling times for large frequency changes and estimated the PLL settling time to be well under 1us. For small frequency changes as here I would suggest it would be faster perhaps. Either way, it does not seem that this would be a limiting factor. 2) In my code I am updating the PLL feedback ratio to change the PLL frequency, and keeping the MultiSynth divider fixed. I believe this is the best way to do it. It appeared to me that an update of the MultiSynth would produce an instant discontinuity whereas the PLL would more gently slide to the new frequency; though as Fred measured such fast settling time maybe it's still pretty fast! Anyway I prefer fixed even integer MultiSynth division ratio as the '5351 datasheet recommends for best performance re low jitter (lowest phase noise).? 3) Mic and Hilbert transforms tend to emphasise the low frequencies which is the opposite of what is ideal for SSB intelligibility. The low frequencies contain a lot of power but don't help much with readability. You can read all over the 'net that getting rid of everything below 300Hz and even de-emphasis of 300-1000Hz is advantageous to good SSB transmission. So if you haven't done anything, I'd consider that.? 4) I didn't personally?find that a super accurate arctan was needed, so the VE5FP article wasn't useful to me; I think this is because the PA already has phase distortion that can only approximately be compensated for using predistortion techniques. The accuracy of the approximation Guido used is?+/- 0.3 degrees which appears plenty sufficient. Whereas for squareroot (to get the amplitude coefficient) I found it important to use the full precision squareroot not a coarse?approximation.? 5) So you are sampling and processing at 31250 samples per sec and doing the arctan and squareroot there, then averaging down 4x to 7812 for the actual?phase and amplitude modulation. I am not sure if it is mathematically correct, I'd have to consider it a lot more... but it's not the way I'm doing it... I start with oversampling at the microphone at 1.4 Msps? but then do all processing (Hilbert, arctan, squareroot) at 12 ksps which is the phase modulation frequency to the '5351. For the amplitude modulation I do a 32-point linear interpolation and DAC DMA at 384 ksps, the main reason for this was to avoid over- and under-shoots in the amplitude modulator.? 6) It does not matter if the PA is Class C, Class D, Class E, whatever. All that matters is that you feed it a constant full amplitude RF input and you expect to modulate it's amplitude linearly.? 7) I found phase pre-distortion was important: the amplifier DOES have variable phase delay depending on amplitude envelope, and it helped a lot to approximately compensate for that.? 8) Did you sync phase and amplitude updates? This is critically important! When you send a frequency update to the '5351 it only takes effect at the end of the final byte written. Then this is a FREQUENCY change but what your mathematics is really looking for is a PHASE change. And the phase change is effected by the integration over time of the frequency change. So what really happens is that you command the new frequency, and the phase shift gradually occurs in a linear straight line from that time at which you commanded the new frequency, until one whole sample later when the next new frequency command comes. Only at that one whole sample later point, has the commanded phase shift been achieved. You have to synchronize the amplitude commanded to the amplitude modulator, so that it matches the phase modulation. This is somewhat an experimentally derived delay because you don't know exactly when the '5351 update is complete or how long it takes to occur, nor do you know really whether you should sync to the average phase in the middle between samples, or the end phase... or any other number of things to take into account.? 9) If you are serious about it then a good spectrum analyzer is a really critical piece of equipment in my opinion. Which unfortunately is not a cheap item. And much as I love what the TinySA brought the world, it doesn't have a narrow enough resolution for this task.? In conclusion there are a lot of things going on, a lot to think of, it's surprisingly GREAT but I would not claim it is trivial to achieve good EER. If I had to guess what's the problem, I would focus on my points #5 and #8 above.? 73 Hans G0UPL |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Why The EER Method For SSB Generation Works
Hi All
?
I¡¯m following with interest the discussions on EER. I¡¯ve just posted on QRPTech my project that uses this technique for SSB. It works but not very well. I¡¯m hoping the wonderful people here could please give me help or hints that might improve the audio because I¡¯ve definitely reached the limit of my ability! Before I provide some details I¡¯d like to acknowledge the genius of Hans. I know well how difficult, and how multi-disciplinary this malarky is! The quality of audio that Hans has achieved is amazing and perhaps more so than most realise. Extremely well done Hans, I am in awe!
?
My project is detailed here:
?
It¡¯s basically a simple double sideband transceiver, but with inverse audio. DSP is done with a Pi Pico. It works quite well. I realised that it could also do CW and AM by controlling the amplitude. Seems obvious now. Then it occurred to me that it might also do SSB using EER because of the control over both the amplitude and the frequency. So with the help of the QEX article (James, VE5FP, September 2024) and some code simulations to check I was on the right track I set it up and was surprised to see it works, not great, but still. Basically I¡¯m generating an IQ signal from the mic at a sampling rate of 31250 Hz. The frequency is determined using the atan2 function and the amplitude as usual sqrt(I^2+Q^2). Then an average of 4 samples each is taken so that the amplitude and frequency are updated at rate of 31250/4 = 7812 Hz. The SI5351 seems to accept this rate OK. I modified the si5351 control code by Pavel Milanes so that the PLL is updated with just 4 bytes over I2C with a bus speed of 800 KHz. So far so good but the audio quality is not great.
?
Testing shows that a varying frequency and amplitude signal into the Mic looks fine on my transceiver spectrum display. However voice, whilst intelligible is not of the greatest quality. I¡¯m now wondering if the SI5351 PLL can respond quickly enough ¨C seeing as it¡¯s getting frequency updates at a rate of 7812Hz and it¡¯s the PLL I¡¯m updating and not the multisynth after the PLL? Would it be better to use a fixed PLL frequency and just update the multisynth? Not even sure if I¡¯m on the right track.
I¡¯m not looking for great audio like Hans has achieved, but an improvement would be most welcome.
?
73, Ian
VK7IAN |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: QDX build help - weird twisted sister
Hi Rich All future?kits will be supplied, once current kit batches are exhausted, with only #28 wire (0.33mm). This is because I tested very carefully including power output, efficiency, harmonic content and the thinner wire works just as well in the output transformers (and QMX SWR bridge, and FT37-43 choke) as the thicker 0.60mm (#22). In fact if anything, the #28 wire works slightly better, probably due to it being easier to attain better symmetry. It's also easier to solder (burn off the enamel) and will make our kitting job easier as it's all then only one wire type.? 73 Hans G0UPL On Tue, Oct 15, 2024 at 10:52?AM atkinsonr via <atkinsonr=[email protected]> wrote:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Tune Rate Cursor - QMX+
#QMXplus
Steve, I've had similar issues with the encoders on my QMX. My solution was to fit a couple of small capacitors to de-bounce the contacts and that has completely resolved the problem (until the encoder completely disintegrates). Here's a link to my post about it:?
?
?
Cheers,?
Ronan
MM0IVR? |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Why The EER Method For SSB Generation Works
It is amazing that the "Envelope Elimination and Restoration method" was there alreay in the 1950s despite limitations in means for implemention.
?
I wrote up a little overview of the methods, called the "Fourth Method of Generation and Detection of Single-Sideband Signals", here https://la3za.blogspot.com/2024/10/a-fourth-method-of-generation-and.html
--
Sverre
LA3ZA |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Hans, Researching and filing a patent is a very expensive proposition for a company. When I was working, the typical cost of patent searches, attorneys, and filing fees was in excess of $10,000 so filing a patent application wasn¡¯t something individuals did very often, if at all. Even for companies, many awarded patents never were used or licensed so it¡¯s a risky proposition from a financial standpoint. Trade secrets are a cheap way to keep the technology in house and vital if the secret is the bread and butter of your operation (your firmware if course)!? Tony? On Tue, Oct 15, 2024 at 12:01?AM Hans Summers via <hans.summers=[email protected]> wrote:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Tony That's good to know, and well deserved. Glad to hear Motorola have good policies.? Around 15 years' back I was a critical part of the world's largest commodity derivatives transaction which in the end netted my employer a profit of $450m (yes 'm' is for million). My own additional benefit from this was a nice round single digit: nothing at all.? But these be the rules. You work for an employer, you make money for the employer, sometimes a lot, and if he decides to be nice he can give you a little taste, if not, tough luck. On the other hand you have a relatively more secure and relatively low risk regular income, Your employer made the investments and took the risk. And in the end if you don't like it, leaving is an option and do your own thing - like QRP Labs! Where I can't make someone else $450m but at least if I make a $ I can keep some of it. Pro's and cons! 73 Hans G0UPL On Tue, Oct 15, 2024, 07:41 Tony Scaminaci via <tonyscam=[email protected]> wrote:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hans, you¡¯re correct. Motorola benefited from our patent filings but as employees, there were a few tiers of monetary awards. The first tier bonus was awarded if Motorola decided to file a patent. Another larger monetary award was paid to the patent contributors if the patent was actually awarded. Dan had at least two monetary awards from his detector patent. Tony? On Mon, Oct 14, 2024 at 12:49?PM Hans Summers via <hans.summers=[email protected]> wrote:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: QDX: Troubleshooting high output power
Ok, rewound it using the "WTST for 9V transceivers (¡°Weird Twisted Sisters Transformer¡±)"
?
And then I realized I am kind of slow.? I was thinking my oscilloscope was showing 40v peak, not 40v peak to peak (even though I posted the correct reading).? And that is correct.? So I didn't even need to rewind T1.?
?
The other thing is it seems that it only transmits in the test screen if you hold down the "t" button.? I was pressing it momentarily (as I thought it was a toggle), and that's why it was doing what it was doing.
?
I still have an issue where the RF sweep is off, but at least the transmitting chain seems happy.
?
Thanks!
? |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Tune Rate Cursor - QMX+
#QMXplus
I think that the "problem" has been resolved. I have added a jumper from the encoder GND side to a known GND (see pg 48 of the manual + photo). All has been stable since.
?
Thanks to Joe, VE7BFK, for the suggestion.
?
--
73, Rick
VE7TK Website: |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Why The EER Method For SSB Generation Works
I was one that dismissed and still do the uSDX not that it could not work but it was
very clear after working with it the classic issue of not enough CPU or bits to do it
well.? Getting a ATMEGA328 to do that was pretty amazing but it was not just quite
there.
?
The problem was basic.? The math was perfect,? the micros needed to be able to do
it with enough precision and fast enough.? That they appeared is not new or a surprise
as us that have been in tech could foresee that.
?
I still have the brassboard I made up to separate the RX functions from the TX so I could?
study it and really probe deep.? I came very close to using a STM32F4xx but round that
time there were other things in my life demanded attention.? I thrilled that my initial
thoughts of more and better a/d as well as d/a and cpu needed were proven.
?
One remarkable thing that came from the brassboard was double sideband AM
that was only hampered but the lack of bits (A/D and D/A).? All the micro had to
do was drive the 5351 for a base carrier.
?
As to SGC 500w amp it also failed they made some and they worked but never
really caught hold.? As an amp its easy to limit the RF for analog phase data/drive
and extract amplitude from a already produced SSB signal.? I suspect the amount
of analog power hardware was part of the issue.
--
Allison ------------------ Post online only,? direct email will go to a bit bucket. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: Think there is an 40 meter intermod like issue in QMX+?.
Performing the ADC test the results didn¡¯t look anyway near the graph in the manual.
After doing more scope probing I discovered there isn¡¯t a CLK0 on pin 2 of IC403; but there is CLK2 on pin 14. As I study the QMX+ ?schematic and learning how a Taylor detector works, new to me, I would guess that the I and Q signals are then messed up. Not fully understanding this yet I think the missing CLK0 is the culprit. Can anyone confirm that there should be a CLK0? This then would be caused bu a defective IC204, MS5351M chip. Any thoughts?
?
thanks,
?bob
? |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Re: QMX 3d printed cover
Hi Nicolas,
?
Thanks for making your QMX cover design public - I printed one and am very pleased with it and mentioned it in a post to this list some weeks ago. I solved the problem of the radio slipping inside the cover by placing a rubber band around it lengthwise, so that the outside part of the band contacts the desk when used as a stand, and the inside part stops it slipping when used as a cover. Simple, effective and enhances a great design.
?
73? David? VK3KR |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ON7DQ Luc
Rick Williams - VE7TK
to navigate to use esc to dismiss