Currently, when operating CW with fldigi and QMX+, if you set the fldigi CW modem set for CAT keying, fldigi uses the CAT 'KY' command to send the characters - it does not try to use PTT to turn the carrier off and on for dit and dah.? This seems to work fine with QMX, but there is a bit of a trick to it: you need to set the fldigi CAT keying to use 'Elecraft', not 'Kenwood'.? [Note that you leave flrig radio set for QMX, only the CAT keying to Elecraft.] ? For some reason the 'Kenwood' setting fills each KY command with one character followed by many spaces; this fills the QMX buffer with spaces and does not work correctly.? But the Elecraft setting sends one character per KY command, and works for me.?? It would be nicer if fldigi sent several characters per KY command, and just let QMX keyer do the timing.

?

I have not done extensive testing with this, but it seems preferable to having QMX assert carrier with each TX, and is already in the firmware.

?

Stan KC7XE

I received my kit for a high-band QMX this Tuesday, and have been working off and on over the course of the week to assemble it. I've gotten to the point where it's fully assembled, and after several close inspections under magnifying glass and microscope, I can't find any significant defects that would lead to the problems I've been seeing, outlined below

I'm powering the QMX from a 10V, 250mA current limiting DC power supply. The core issue is that the board (That I have never programmed) is drawing far more current than the assembly document suggests it should at this stage)
The different cases I've run:

Main board connected to power, no additional boards attached: 0mA drawn
Main board connected to power, PS #2 board attached, all others detached: 0mA drawn
Main board connected to power, PS #1 board attached, all others detached: Initially current limited to ~8-9V @ 250mA. After an inrush period, stabilizes to 10V @ ~100mA
Fully assembled QMX: Current limited to ~6V @ 250mA. LCD screen turns on with solid black squares displayed. Appears as a usb device immediately upon plugging in to a computer, without pressing the potentiometer button.

I've been focusing my troubleshooting efforts on the interaction points between the main board and PS #1, and haven't found any obvious shorts. Checking the pin header connection to PS #1 with the board attached, I didn't see any shorts to ground. I also double checked and ensured that there weren't shorts around the power amp transistors, and around X201 where I saw VCC being used.

I've attached photos of the back side of the main board, and front side of PS #1 as installed under a magnifying glass below. If any other pictures are required let me know and I can provide.
Note that there is some debris near the PA transistors in the second photo. I cleaned that and tried again, and the board still enters the current limited condition.

?

Thanks and 73 de VA3KBY

?

I received my 20m-10m QMX kit and am currently assembling it. Unfortunately, the kit I received was missing the three 33pF and the 100pF capacitors (C401, C404, C521, C516). It will probably take a while until I receive the missing parts and I wonder if it is still possible to finish the build and start testing the device (basic tests like: will it turn on etc). Or should I just let it sit and wait?

?

73,?Dennis DM4NTZ?

Audio levels problem solved ! It was due to incorrect parameters in the AGC settings. I swear I did not do it....think these defaults came with v1.02 ? The AGC threshold was set to 15...should be 8. Also sample blocks was 2....should be 5. The audio levels on all modes are now roughly the same. Audio attenuation setting I have now increased to -100dB, seems I have very sensitive headphones.

de VK6XT?

?

On Thu, May 22, 2025 at 10:03 PM, Curt wb8yyy wrote:

Whether the via is needed depends upon which or both have connections.?

Drilling is not an acceptable rework method for boards with plated-thru holes.? This is a 6-layer board with internal layers you should never go any where near it with a drill bit lest you ruin the board.

73, Don N2VGU

?

On Thu, May 22, 2025 at 09:44 PM, Stan Dye wrote:

That's why I recommend to carefully scrape or burn/sand off the enamel before soldering the coil wires, which avoids much of this issue.

I think that the assembly instructions should recommend this approach, I think folks are poorly-served by encouraging use of the burn-and-pray method, it just fails too often.??

Perhaps it works OK for the QRP Labs elves, who do it all of the time and learn the technique, but we keep hearing the same cries of woe about open transformer connections ad infinitum.

?

Hans, please change the instructions!

?

73, Don N2VGU

Hans explains it so well, I almost understand it.?

I'd like a solder pot, but it will have to wait until I have a bigger workspace and that day may not come. My bench is so cramped that I think a pot of melted lead would be a rather dangerous thing to have on it! As is, I'm always tripping on cables or catching wires with my sleeve. So I scrape, sand and tin.

?

While waiting for my QMX+ kit to arrive, I bought a small crochet hook but my first attempt at using it was a disaster and since my wire supply was limited, I wound with the normal method. I still screwed another toroid up, but caught the mistake before soldering it in...I still ended up with about 13" of spare wire when finished. Which is still sitting on top of my washing machine and no doubt driving my wife crazy.?

On Thu, May 22, 2025 at 09:35 PM, Jerry Gaffke wrote:

Could be process variation between different manufacturing lines for the BS170?

Could be variations in cap and inductor impedances for the LPF?

?

Probably more likely different load impedances and/or latent failures due to electrostatic discharge or poorly-regulated or excessive supply voltages.

73, Don N2VGU

It has already the AGC board, see just below the "s" of Labs.

73 - Pierre - FK8IH

One person's experience:

?

I use a current and voltage limited supply when first testing kits. However, after I confirm that all is working properly, I go to battery power thereafter. All of my subsequent updates and testing is done with full power.?

I have not had a failure of finals in 12 QRP Labs tcvr kits. I have had various early failures (infantile part failures?) but none after those were addressed and fixed.?

Excess power and poor RF matching can certainly attribute some of the failures, but I don't think running on battery or any full voltage p/s explains failures.?

--

73, Dan? NM3A

Yes Stan, that is the scenario. When listening on headphones to relatively weak signals in, for instance LSB mode...to change to another mode you can only click forward passing through Digi mode. In my case and the same for others posting to this group, audio is painfully loud. I don't believe that I have changed any settings other than the band configuration overall audio gain. I do not support the use of AGC for digital modes either. The difference in headphones levels is the problem. If the levels for LSB, USB and CW were higher, the end result would be the same as by reducing Digi level.

de VK6XT?

The price of success!

?

My Factory built QMX hiband can be in your hands in a week or so. 12- volt build, enclosure, .027 installed. Works perfectly.

Email sfab43@ho**ail.com for more info. Steve JS6TMW/AI6KX

Enjoy

73 DE muhsin TA1MHS

?

Randy, a couple of comments that I can give.

1) I have 2 QMX+ and 2 QMX (one low-band rev1, one mid-band rev3).? Each of these units had quite different looking plots for some bands of the calibration, some similar to what you describe.? And some bands didn't finish the plot during cal.? I was able to get most of the bands to 'finish' by repeating, sometimes changing other parameters.? I didn't really keep track of whether the QMXs performed substantially differently than QMX+.

2) I did have one band with the "change over 30 degrees" issue.? This may be a firmware bug.? Check and see if the "change over 30 degrees" was really a rollover from near 360 to near 0, which is really a 1 or 2 degree change.? When I investigated mine, I found that the firmware rollover logic was working in one direction, but not the other.? I reported this in a previous post late in the beta test period.? Then later I discovered I had a bad dummy load connection on that QMX+, and repeated the cal without encountering that same error - but I don't remember Hans saying he addressed the rollover issue in the release firmware.

3) I also see widely varying Y-axis values in the IMD cals.? Most bands dip to low values, but some are in the millions, like you describe.? And for a few bands there isn't a dip in the middle, it is still dipping when the cal maxes out at value 74.? Hans said finding the 'low value' or 'dip' is the key, and that the vertical axis value was not important.? The vertical axis is, though, from his description, some measure of IMD products - so maybe those bands have a higher total IMD, or maybe it is just a sensitivity to his measurement mechanism.

?

In the release firmware, Hans made a change so that if a calibration didn't complete, he filled the rest of the data structure with 0s or valid data.? Previously it was uninitialized, and could result in distorted speech signals.? But now it should be fine.? So test your SSB output on the different bands - you will likely find it is just fine, regardless of the strange cal data.? I would check carefully on the bands with that 'over 30 degrees' issue, in case the error there prevented his logic that does the fill.

?

My QMXs appear to still work fine with SSB, regardless of the strange-looking cal data on some bands.? So you may be fine.

Stan KC7XE

?

I raised the matter with the supplier (a very well-known company then). They apologised and sent me a full set of correct value resistors.

73 John. F5VLF

Sounds like you lost some connections to your front panel/display board.? Try re-seating the front panel in its connectors.? (Display and backlight are powered by 5V supply, and I think top row 'blocks' typically mean the processor can't talk to the display and initialize it.

?

And regarding your 3.3V: in normal operations, the 78M33 is only used for about 1/2sec during power-on, then the processor activates the 3.3V smps to provide Vdd.? If you are also powering with 3.3V from the Drok, it could conflict with the SMPS, and confuse the processor (as it reads feedback on regulating the voltage), depending on how you hooked it in, and whether its voltage is a bit higher or lower than smps.? Did you connect it in via the diode that the 78M33 feeds (D103)?

?

Stan KC7XE