¿ªÔÆÌåÓý

ctrl + shift + ? for shortcuts
© 2025 Groups.io
  1. QRPLabs
  2. Topics

#QMX Glitchy encoders? Not any more ? #qmx

 

Hans, this post is why we love and respect you.

Mike Krieger


On Wed, Aug 7, 2024 at 2:48?AM Hans Summers via <hans.summers=[email protected]> wrote:
Hi all

I'm interested to read about this modification.?

In general I don't like hardware debounce of things,?including buttons and rotary encoders which are interfaced to a microcontroller:
  1. Of course it adds parts to the BOM and generally for simplicity?and cost reasons, it's best to optimize as much as possible.?
  2. It's very easy when adding a capacitor to get rid of a bounce but you might?have unintended consequences, the capacitor will slow things down and what can happen then is that fast turning of the rotary encoder misses rotation click events.?
In general where a microcontroller is involved, it ought to be possible to do debouncing in firmware. There's a right way and a wrong way to do this. The wrong way is with timers or filters which effectively emulate what capacitors did in hardware. The right way is with a state machine which eliminates bounce by ignoring repeat state transitions until the next in sequence is seen.

The cost of a few capacitors is very small and certainly much less than the time I will spend on firmware. Adding them to the next PCB revision would be trivially easy and the cheapest solution. But I don't like it because I still think the firmware approach is most elegant. I think I can make improvements to the rotary encoder handling. Most importantly, any firmware solution can be easily applied to all 6,000+ QMX/QMX+ radios already out in the wild. Which after all, is one of the whole points of SDR in the first place.?

Note that the QCX-mini has an RC filter on its rotary encoder signals. However as a point of history, this was NOT done to eliminate encoder glitches. Which is why there are no capacitors on QMX+. It was done for a completely?unrelated reason. There was an audible noise which could not be eliminated, it's something about the PCB layout or compact structure, in QCX-mini. Adding the RC filter eliminated the problem, for which no other solution could be found despite a lot of investigation.?

In summary I think the problem (about debounce) is in my firmware, and the best solution is also in my firmware, not me adding capacitors to future PCB revisions.?

73 Hans G0UPL



On Tue, Aug 6, 2024 at 10:18?PM Daniel Walter via <nm3a=[email protected]> wrote:
Steve,
I found 1206 fits nicely, but so do 0805.?
--
73, Dan? NM3A
 

Hans,
From a cost POV, 8 tiny SMD caps is a trivial addition to the kit in production. And I don't know anything about what other negatives they may bring. But certainly for the thousands of units out in the universe, adding them now is not a trivial endeavor for many hams. So a f/w solution makes sense.?

The issue of skipping menu items, double or triple button pushes, volume or tuning steps in the wrong direction or too fast increase were a minor annoyance, not enough to make a complaint about. But the issues were consistent on all four of my units (rev 1, 3, 4 & rev 1 +) and were there from the earliest on all of them.?
?
So, for me, this h/w 'fix' (is that what it really is?) eliminates all those quirks on all my QMXs. It has been well worth the effort to get it done. So far, I have not seen any ?problems from their addition.?
--
73, Dan? NM3A
 

On Tue, Aug 6, 2024 at 07:33 PM, Steve [AA8AF] wrote:
what size did you use that seems to fit nicely here?
I used 0805 caps as they fit neatly across the pads.
?
 

Hi Hans,
?
Yes, I completely agree that the firmware approach is a wise move and look forward to seeing how that develops. This was purely meant as a stop-gap solution to get my QMX back into a useable state after a fair bit of to and fro with various encoders and the frustration of constantly skipped steps.
?
With the cap values used, the radio keeps up well with fast encoder rotations and doesn't seem to suffer any skipped steps.?
?
On a side-note, I remember making a little tuning add-on for my original QCX "classics" that piggy-backed on the encoder and emulated its rotation at different speeds and directions based on the position of a pot. I was amazed at the time just how fast I could go before things got glitchy.
Here's a video of the box in action:
?
Cheers,?
Ronan
MM0IVR?
?
 

Hi all

I'm interested to read about this modification.?

In general I don't like hardware debounce of things,?including buttons and rotary encoders which are interfaced to a microcontroller:
  1. Of course it adds parts to the BOM and generally for simplicity?and cost reasons, it's best to optimize as much as possible.?
  2. It's very easy when adding a capacitor to get rid of a bounce but you might?have unintended consequences, the capacitor will slow things down and what can happen then is that fast turning of the rotary encoder misses rotation click events.?
In general where a microcontroller is involved, it ought to be possible to do debouncing in firmware. There's a right way and a wrong way to do this. The wrong way is with timers or filters which effectively emulate what capacitors did in hardware. The right way is with a state machine which eliminates bounce by ignoring repeat state transitions until the next in sequence is seen.

The cost of a few capacitors is very small and certainly much less than the time I will spend on firmware. Adding them to the next PCB revision would be trivially easy and the cheapest solution. But I don't like it because I still think the firmware approach is most elegant. I think I can make improvements to the rotary encoder handling. Most importantly, any firmware solution can be easily applied to all 6,000+ QMX/QMX+ radios already out in the wild. Which after all, is one of the whole points of SDR in the first place.?

Note that the QCX-mini has an RC filter on its rotary encoder signals. However as a point of history, this was NOT done to eliminate encoder glitches. Which is why there are no capacitors on QMX+. It was done for a completely?unrelated reason. There was an audible noise which could not be eliminated, it's something about the PCB layout or compact structure, in QCX-mini. Adding the RC filter eliminated the problem, for which no other solution could be found despite a lot of investigation.?

In summary I think the problem (about debounce) is in my firmware, and the best solution is also in my firmware, not me adding capacitors to future PCB revisions.?

73 Hans G0UPL



On Tue, Aug 6, 2024 at 10:18?PM Daniel Walter via <nm3a=[email protected]> wrote:
Steve,
I found 1206 fits nicely, but so do 0805.?
--
73, Dan? NM3A
 

Steve,
I found 1206 fits nicely, but so do 0805.?
--
73, Dan? NM3A
 

Ronan,
I'm going to order some 10nF SMD caps - what size did you use that seems to fit nicely here? ?1206?
Steve
AA8AF
 

Thanks for this suggestion. I have never had problems with the QMX "Tune" encoder, but continue to have volume glitches on the Volume encoder: /g/QRPLabs/message/123019?p=%2C%2C%2C20%2C0%2C0%2C0%3A%3Arecentpostdate%2Fsticky%2C%2CVE7EA%2C20%2C2%2C0%2C106355942
?
I hope to try this capacitor addition soon to see if it helps my volume control problem.
?
Larry VE7EA
 

Good luck, Jon. Hope it works for yours too. It might be worth posting back here with your results.?
?
 

Glad that's worked for you, Walter. Job well done, Sir. The buttons on my rig behave well so I didn't add any caps to those.
?
As for the encoders; They do deteriorate with age as the contacts become dirty/worn and the wipers lose their pressure. Some are sub-par from the factory and it sounds like that's what you've got there.?
?
Cheers,?
Ronan
MM0IVR?
 

Thanks for posting this modification. One of the main reasons I haven't been using my 1st rev QMX is because the glitchy encoders really got on my nerves!
?
I'll be adding the caps shorlty and very much look forward to the improvement :-)
?
Thanks,
Jon G4IVV
?
 

Select and Exit caps placed. Can't see any double presses any more, so looks like more success.?

Photos below of button caps on a Rev 4 QMX and encoder caps on a QMX+.
--
73, Dan? NM3A
 

Ronan, Manfred,
I don't think age has anything to do with it. My early (May 2023) Rev 1's had the glitch from the beginning. My Rev3, Rev 4, and QMX+ Rev 1 all have the glitch. It seemed to be worse with the earliest firmware.
?
I have just put the 10 nF caps in and the glitches are completely gone! Before this mod, scrolling through the Menu steps were frequently skipped semi-randomly, especially when turning fast. Occasionally it would even go backwards.?

I have also seen unintentional double clicks with the 'Select' and 'Exit' buttons, so I wonder if caps on these might also stop these double clicks. I think I will try it.?
--
73, Dan? NM3A
 

On Sun, Aug 4, 2024 at 07:46 AM, Manfred DL9XX wrote:
But I'm surprised that this standard measure isn't included in the original design.
The software seems to handle most things well so I suspect it's not really an issue until the encoders start to age a bit
 

Thanks, Roman. My QMXs suffer from the same issue. Looks like an easy fix.?
?
73, Mike KK7ER?
?
 

Great that you found out and shared it.
But I'm surprised that this standard measure isn't included in the original design.
Software hides many hardware problems, but it doesn't fix them.
Keyword contact bounce
 

Ronan,
Fantastic! It's been an annoying problem for me too. I am going to do this with mine too.?
--
73, Dan? NM3A
 

On Sat, Aug 3, 2024 at 09:27 PM, Wei AG6AQ wrote:
will freeze the QMX randomly.
Wei, that sounds like a different problem but it seems you've solved it one way or the other. ?
 

QCX rot encoder resistors and capacitors
 

Hi, Ronan

Thank you for sharing this,

I remembered that QCX has 10k pull-up resistors and capacitors to ground like on image attached.

It is quite interesting solution for QMX. May be chip has resistors internally but maybe some experience add more details.

On the other side, contact bouncing can be solved in software part, but adding capacitor and resistors definitely add more accuracy at least for VFO encoder.

1 - 20 of 22