A brief article about troubleshooting your QDX.
Remember that QDX firmware is loaded into all QDX boards and they are comprehensively tested prior to leaving our manufacturing facilities. This does not rule out intermittent faults or damage that can have occurred in transit; but generally speaking if the QDX isn't working, it is normally a mistake you made during construction or configuration, sorry.
In the following: for DC continuity checks use a DVM in DC continuity mode, or if it doesn't have one, in its lowest resistance measurement mode; and make sure that the QDX is powered down during the measurement.
Before anything else - I am assuming you have a power supply capable of delivering the required current and at the required voltage. Make sure you know whether you have built your QDX for 12V operation or 9V operation. And make sure you power supply produces a clean regulated 12V or 9V, and can deliver enough current (1A plus or minus). Make sure it is a good quality power supply, does not generate big nasty spikes at power-up, doesn't generate huge RF noise during operation, etc. If it's battery powered, make sure your battery hasn't run flat, under load!
If you do need replacement components, most of the components in QDX are quite commonly and inexpensively available.
- Minor PCB damage such as broken traces can be repaired by a piece of wire and some care and attention
- All SMD resistors and capacitors are size 0603 imperial.
- All through-hole capacitors are 2.5/2.54mm-spaced, NP0/C0G (class-I low loss RF dielectric) ceramic capacitors. NP0/C0G is important for low loss at HF.
- BS170s can be anything from a reputable supplier e.g. Digikey
- Micrometals and Faiirrite toroids can be obtained from http://www.kitsandparts.com
- Most other stuff from Digikey who have a good $4.99 shipping in United States; or your local parts distributor in non-US countries
By the way before delving deeper into anything else - if you visually inspect the PCB under magnification, you might be able to find the problem with no further ado. If you can find a SMD component which has been knocked off, a dry joint, a capacitor installed with the wrong value, etc., or some missed step of the instructions - visually - it's quick and easy to fix.
Please always make sure that you are on the latest firmware revision, which you can download from the QDX page. Installation is very easy and is documented in the QDX operating manual. It requires no programming hardware, installation of no software or drivers, and works on any OS (Windows, Mac, Linux etc). There is almost never any excuse not to upgrade to the latest firmware. And if you don't - how do you know you aren't suffering from a bug that is caused by firmware, is already known and has already been solved?
Firstly, I am assuming you actually NEED to trouble-shoot your QDX, in so far as, it actually genuinely isn't working.
There are various reasons why you might not have a fully operational digital radio station, which have nothing to do with your QDX hardware having any fault at all. I assume WSJT-X though other digi modes software is similar. If your QDX is not receiving properly or not transmitting properly under control of WSJT-X:
1) Did you configure the audio settings to use the QDX transceiver as "USB soundcard"? If not, no audio is going or coming to QDX. See WSJT-X Settings "Audio" tab.
2) Did you configure WSJT-X to connect to your QDX serial port? Which will be named COM-something e.g. COM3 on Windows, and something like /dev/ttyACM0 on Linux. That's the Serial Port setting on the WSJT-X "Radio" tab, and make sure you select Kenwood TS-480 or TS-440 as the Rig.
- If this isn't working then have you chosen the correct COMn or ACMn port? Bear in mind, if you are using other peripherals which also have serial ports, they may be assigned COM or ACM numbers.
- If on Windows OS pre Win10, you will need to install a driver as documented in the QDX manuals and also here http://qrp-labs.com/qlg2 - on Win10 and Win11 this should not be necessary.
- If on Linux, you may need to do other things like issue commands to disable the modem service, or permission the dialout group - refer to the QDX operating manual for details of this
3) Have you got all the audio settings at full volume (100%), no more, no less? This is also documented in the manual. The status LED on the QDX will help you to know.
- If it flashes once for 0.1 seconds, repeating every second, the PC software has put the QDX into Transmit (via the CAT command) but the PC is not sending any audio - most probably because you have not selected the QDX as the output sound device.
- If it flashes twice, each flash lasting 0.1 seconds, separated by a 0.2 second gap, and repeating every 1 second, then the PC is sending audio but the volume of the audio is not sufficient. For QDX, all volume controls should be set to 100%, no more, no less.
- If the LED undulates in a pulsating (sinewave current flow) pattern at approx 1Hz, then it is transmitting normally, it means the PC volume and device selection is all correct.
Bear in mind that audio volume is typically (on both Windows and Linux, for example) controlled in THREE places:
- The output volume slider at the bottom right of WSKT-X named "Pwr"; make sure this is all the way at the top
- Specific gain applied to the particular sound device in the sound manager (whatever sound manager applies on your system) - make sure the device is at 100%
- Master Volume control - make sure it is also at 100%.
NOT more than 100%. 100% is correct. No more, no less.
4) If you are producing power output and seeing noise and/or stations on receive but no decodes then:
- Are you on the right frequency for the modes you want to use?
- Have you selected the correct operating mode in WSKT-X (e.g. FT8)?
- Are you have chosen a band which is open, propagation conditions will allow a contact?
- Is your PC time accurately synchronized? Remember, many digi modes require precisely synchronized timing and your PC needs to be set up with a time service that automatically uses internet time servers to sync the real time clock (or some other method e.g. a connected GPS receiver).
If none of the above... then proceed to Hardware debugging next.
There are always three phases in debugging a transceiver such as QDX or QCX.
- The thing is alive and talking. In the case of QDX, that means, the status LED lights up, and you can connect to it via a Serial terminal emulator.
- The transmit side is working and producing the correct output power levels.
- The receiver side is working correctly.
See, you have to take each of these in turn, and not jump ahead. EVER. Taking them next one by one.
QDX is alive and talking
Alive and talking means:
- QDX is connected to your power supply which is switched on and isn't freaking out
- There is no smoke
- The red status LED on the QDX front panel flickered for 5 seconds at power up then went steady
- Current consumption is normal (around 150mA for a QDX Rev 3-onward in Receive)
- You connected a PC via a USB cable and fired up a terminal emulator and saw a screen like the following:
Now if these conditions are NOT met and you have an ongoing problem with one or more of them, see if any of the following are helpful:
1) Your QDX consumes an enormous amount of current, as much as your power supply can deliver, or deliver before hitting any applicable current limit you have set?
a) Check for any obvious short-circuits such as a solder splash, that could be shorting +12V to ground
b) You might have wound the T1 output transformer (binocular core) incorrectly; it is surprising how easy it is to mess this up. Or mess up its installation by getting the wrong wires in the wrong holes. If you incorrectly have the some wires mixed up such that current can flow through L14 right through the binocular windings to ground, then you will indeed draw a massive current. The solution is to remove T1, rewind it carefully (refer to the manual) and try again. You do NOT need exactly the same 0.60mm (#22) wire as used in the kit, something similar will work as long as it fits the core and the holes.
2) Your status LED is not lit?
Did you install it the correct way around? Remember LED's have an anode and a cathode. It has to be installed the correct way around otherwise it will not light. Refer to the appropriate assembly manual step.
3) PC cannot connect serial port
a) Surprisingly again, how many times it happens that a USB cable is defective! These cables are often produced for minimum cost, not maximum quality.
b) If you have a Windows PC on Win 7, 8.0, 8.1 etc., anything pre Win10., then you need to install a driver; please refer to the operating manual.
c) If it is Linux, have you disabled the modem manager, and allowed dialout group permission? Please refer to the operating manual.
d) Are you using the correctly numbered COMn (Windows) or ttyACMn (Linux) port?
e) Any other connection or driver problem - please ask on the QRP Labs discussion forum at http://groups.io/g/qrplabs where you will get lots of help.
4) If the QDX cannot remember any configuration settings, and always powers up in firmware update mode, but then you copy in the firmware and it appears to work normally
This generally means there is a problem with the 24C64 EEPROM memory chip IC6. You should check all the pins visually to make sure that none have come loose, dry solder joints etc. If the processor can't talk to the EEPROM, this is why it boots up in firmware update mode every time, and why it can't remember any configuration settings.
Next we come to transmitter debugging. Generally this is easier than Receiver debugging. The best way to test the transmitter hardware is by using the "Transmitter test" utility, see below. Press the T key to transmit for 0.5 seconds or keep your finger on the key to keep transmitting continuously. Use the + and - keys to move up and down between the configured bands.
1) No output power on any band
a) Check that you have correct continuity across the 10-turn FT37-43 choke L14. A common error is failure to scrape off the wire enamel, which means no electrical connection, no current flow through to PA, no power output.
b) Similarly check the toroid enamel on the wires of the output transformer T1
c) The T1 transformer could be incorrectly wound - the assembly manual contains details of DVM continuity checks that you can do. You may need to remove, rewind and reinstall T1.
d) Check the Band Configuration screen to make sure that the band is enabled
e) Make sure you are running the latest firmware
f) If you have an oscilloscope, check for a 5V squarewave during transmit at the gates of the four BS170 transistors
g) Check enamel on the six LPF toroids
h) Check that the 47uH axial inductors are all correctly installed and have continuity
i) Check that the six 1N4007 diodes are installed and with the correct polarity (white strip aligns with the stripe shown on the PCB silkscreen; also check photos in the assembly manual
j) check for DC conductivity through the 47uH inductor L1 (sometimes these fail open-circuit during installation)
2) NO power output on a band
You need to understand which toroid belongs to which band. For the common 80 / 60 / 40 / 30 / 20m model of QDX, L2 + L6 are for 80m, L3 + L8 are for 60/40m, and L4 + L10 are for 20/30m.
a) Check the toroid wire enamel to make sure you have a good electrical connection, no dry joints, and so there is DC continuity through the toroids.
b) Check that there is DC continuity through the 47uH inductors L6, L7 or L9 (band depedant) and L1.
3) Low power output on all the bands
a) This can mean that the output transformer T1 was incorrectly wound. For example, the primary is supposed to be 1.5 + 1.5 turns. The center tap is supposed to be, well, exactly that, a CENTER tap. If you accidentally wound 0.5 turns + 2.5 turns then this would severely mess everything up and result in low power output across the bands.
b) You might have counted turns incorrectly when winding T1. Remember that "1 turn" means that the wire goes through BOTH holes of the binocular toroid.
c) You might have wound the incorrect number of turns; for example, you wanted to build for 9V and so you should have wound T1 as 3:3 but you accidentally read the wrong thing and wound it 3:2 (as for 12V); then obviously at only 9V the power will be reduced.
d) Check with an oscilloscope to see if you have good power output at the pad connected to the center pin of the BNC connector, If so, it could mean the BNC connector itself is defective.
3) Slightly low power output on one or more bands
Low power output is a common complaint and it could just be a matter of tweaking the QDX for best performance. You need to understand which toroid belongs to which band. For the common 80 / 60 / 40 / 30 / 20m model of QDX, L2 + L6 are for 80m, L3 + L8 are for 60/40m, and L4 + L10 are for 20/30m.
Generally if you squeeze together the turns on the appropriate one of L2, L3, L4 (for the band in question) then you will increase the power output. You should NOT go above 5W. It is tempting to go for "maximum smoke" but... if you do so, then will significantly increase the risk of literally seeing maximum smoke. But to go from 3.5W to 4.5W by squeezing the turns is fine. L6, L8 and L10 turns should generally be spaced evenly around almost the entire core circumference.
4) Low power output on one or more bands
If you have really low power output like 1W, 2W etc., and if you have an oscilloscope and the output waveform looks weird, not nice and sinusoidal - then this can be an indication that the band switching is not working properly. If more than one Low Pass Filter is enabled, that really can mess everything up.
You need to understand which toroid belongs to which band. For the common 80 / 60 / 40 / 30 / 20m model of QDX, L2 + L6 are for 80m, L3 + L8 are for 60/40m, and L4 + L10 are for 20/30m.
You can switch between bands using the + and - keys when in the Transmitter Test screen, without pressing T to transmit. For the low pass filter that is active, the voltage on any of the LPF toroids for that selected band LPF should be low (around 0.6V DC). You can measure this with a DVM or an oscilloscope. For the non-selected two band positions, the voltage should be near 12V (for a 12V powered system, for example).
For 80m, if this is the band with low power output, also see the next section.
a) Do you have DC continuity across each of the 47uH axial inductors? These can sometimes break internally during installation or soldering, and it's invisible from the outside. But the DVM will show the break (no DC continuity).
b) All toroids installed correctly with enamel properly scraped/burned off so that there is a good electrical connection and DC continuity through the toroids.
c) Sometimes, one of the switching transistors Q3, Q4, Q5 could have become damaged (destroyed). So in the last resort - if you don't see the correct voltage here then one or more of these may need to be replaced.
d) CHECK that the correct capacitor values have been installed in the correct locations (refer to assembly manual), it is surprisingly easy to mix them up.
e) Check that all the diodes are the correct 1N4007 type and they are all orientated correctly.
5) Low power output on 80m band but the other bands seem to work normally
a) Check the above section "low power output on one or more bands" to see if there is any mistake in the LPF components.
b) PCB Rev 4 and above: check the 470uF capacitor, 330uH inductor and 22uF capacitor which are in the switched mode power supply circuit for the PIN diode forward bias current, these are the components on the top side of the board, next to the binocular transformer T1. If one of these components has been damaged or the connection broken, a dry joint or whatever, then it means the high forward bias current doesn't get generated; higher bands are unaffected by this but for 80m a higher transmit forward bias current is required than the 4mA or so provided by R1 and R2 always (even on receive). Otherwise there will be sub-optimal attenuation which will reduce the output power.
A quite good place to start is the RF Sweep for 40m. Maybe you have something like this:
Maybe the peak doesn't occur exactly at 7MHz, whatever. That is a matter of adjustment. But for now look at the number in the top left of the display. We generally see 0, 1, -1 or -2dB at this point (top of the y-axis). If you see significantly less than that, you could have a problem. This number is independent of the gain settings in the Band Configuration screen so it should be the somewhat similar for all units. Something like the above trace is the desired, correct outcome. Typically:
20m sweep: top is 0 to -4dB
30m sweep: top is around -10dB +/- a few dB
40m sweep: top is around 0 to -4dB
60m sweep: top is around 0 to -4dB
80m sweep: top is around 0 to -4dB
For 60m and 80m the peak of the filter is normally not near the 5 or 3.5MHz desired band center frequency respectively; this is normal and does not indicate a problem with the QDX. Such high sensitivity and performance is anyway not a requirement on 60m and 80m due to high ionospheric noise.
Now run the Image sweep tool on 40m and you should have something like this if all is working correctly:
In this example the measured image rejection is -50.4dB but this number varies quite wildly from unit to unit. You should have at least -35dB but it could be as high as -60dB or more. It doesn't matter. But if it is less than -35dB, perhaps much less (for example -5dB), then you have a problem. Also note the y-axis top in my example is -2 and this is consistent with what we saw in the RF Sweep and at the 7MHz position so this is as expected; if it was much lower, it could be a problem.
1) Receiver RF sweep peaks are much too low on every band
a) Check the soldering of the BPF inductor L12, make sure all the enamel is burned/scraped off so that there is a good electrical connection. You should be able to use DVM probes (with QDX powered off) to measure DC continuity between pins 3-to-9, 4-to-9, 5-to-9 and 6-to-9 of IC3. Additionally the voltage (DC voltage with a DVM) should be within a few 10'ths of a volt from 2.5V. If not then check the soldering and/or enamel again.
b) Trifilar transformer T2 is a common cause of problems. If any of the wires are in the wrong holes, or there is any short between wires, or if the enamel is not burned off completely, then the receiver won't work correctly; you can remove T2, check the winding continuity (to understand which wires go in which holes) and reinstall. Refer to the assembly manual.
c) If you have a Rev 3 or Rev 3a board, these can suffer op-amp lock-up - you can remove each of C71/C72/C73/C74 and replace them with a short (for example, just solder blob across the pads of each one to short out the capacitor position).
d) PCB Rev 1-4: Check for a solder bridge from C72 to the trifilar toroid T2 which is quite close nearby
e) If you have PCB Rev 4, the capacitors C71-4 should already have been removed and bridged by a solder blob; but check that one of the blobs didn't get accidentally removed when you installed T2.
f) Check that none of the QSD capacitors C39, 40, 41 or 43 are shorted; sometimes there could be a solder whisker for example. A DC continuity check across these (QDX powered down) should show no continuity.
g) If this is a Rev 3a board, did you remove the C41 via short? Refer to the Rev 3a Assembly manual.
2) Receiver RF sweep peaks are normal on 20m, 30m perhaps, but very low on 40m and 80m
a) There could be a soldering problem at the L12 band pass filter inductor taps, due to a dry joint or enamel not being burned/scraped off properly
b) When installing L12, if the toroid is attempted to be repositioned after some wires are soldered, it could physically snap one wire under the toroid; for example, in my example if 40m and 80m have low receive peaks, it could be that the wire has snapped just after the 30m tap.
3) Receiver RF sweep peak is low on one band
a) Most probably, there is a soldering problem at the L12 band pass filter inductor taps, due to a dry joint or enamel not being burned/scraped off properly
4) Receiver RF sweep peak is not normal on one or more bands
a) Check that the capacitors C28, C29, C30 and C31 are installed correctly: correct values in the correct holes.
5) Receiver RF sweep peak is just weird, not quite right on all or most bands
a) Check resistor R19; it is nearer to the board edge so evidently could be prone to damage; the resistor could be still installed but check with a Loupe, if the resistor is scratched or damaged, it could be open-circuit. This is the resistor that feeds the Si5351A Clk2 signal generator output back into the receiver front end. If the resistor is missing, you get just signal leakage picked up by PCB parasitics. In this case the sweeps are weird. But not terribly low, just weird. Not quite right. So check R19, even with a DVM on a suitable resistance range to measure 100K.
6) Image rejection is far too low (significantly worse than -35dB)
a) Check trifilar transformer T2 again. If all the wires are not correctly connected (enamel burned/scraped off, properly soldered, and with wires in the correct holes) then this leads to poor image rejection.
b) If you have a Rev 3 or Rev 3a board, these can suffer op-amp lock-up - you can remove each of C71/C72/C73/C74 and replace them with a short (for example, just solder blob across the pads of each one to short out the capacitor position).
c) PCB Rev 1-4: Check for a solder bridge from C72 to the trifilar toroid T2 which is quite close nearby
d) If you have PCB Rev 4, the capacitors C71-4 should already have been removed and bridged by a solder blob; but check that one of the blobs didn't get accidentally removed when you installed T2.
A log of QDX repairs (by Hans G0UPL)
Here at QRP Labs HQ we have a set of busy elves who produce assembled QDX and QCX-mini and QCX+ kits for a small extra fee, which is an option when ordering. Now, the great majority of these work first time and without issue, particularly once the elves are no longer as green, and have plenty of practice with how to wind toroids well, solder well etc. But there are always a few where someone's concentration lapsed, or even in some cases there could be a defective component. These cases fail our rigorous testing and come to me to be repaired.
I began keeping some brief notes on the failed QDX units and their repairs. These note may be useful to you if you have similar faults.
SUCCESS RATE: 24 OUT OF 25 UNITS REPAIRED SUCCESSFULLY
1) Deaf receiver
Testing initially done on 40m. There was a largish sinewave-ish at pin 9 of IC3 indicating the Quadrature Sampling Detector mixer was unbalanced. Further investigation of all components in the QSD determined that C40 was shorted out by a solder whisker. It wasn't visible but the DVM measured DC continuity across capacitor C40. C40 was desoldered hten reinstalled. Then the receiver worked correctly (on 40m and all other bands).
2) Very low power output on all bands
An oscilloscope connected at the BNC center pin pad on the PCB showed normal power output. The problem was therefore determined to be the BNC connector itself. Further investigation revealed that the ground pin connecting to the cylinder of the BNC connector, was not in fact connecting to the cylinder. It was possible to poke in a soldering iron and some solder to make a connection (and melt a little plastic, a,k,a, collateral damage), and remediate the situation, without needing to replace the BNC connector.
3) No power output on 20/30m bands
L9 (47uH inductor) was found to be open circuit and was replaced.
4) Deaf RX (very poor image sweep)
Broken wire was found after the 30m tap on the BPF inductor L12, meaning signal could not reach through to the 40m and 80m taps; this was repaired with a component wire offcut then all sweeps were normal.
5) Huge current draw and no other signs of life, on switching on
Current hit the 5A power limit on the PSU. Cause was a miswound O/P transformer causing short from +12V to ground. Additionally this was a 3A board and there was the C41 shorted via problem, which had not been cut with a knife; so there was also a deaf receiver problem. O/P transformer T1 was rewound and reinstalled; the C41 short via was cut, and there were no other issues.
6) Deaf receiver with no image rejection
Multiple errors found in the receiver due to poor construction:
- Rev 4 board with missing C74 solder bridge, and missing copper tracks between C74 and other components; tiny wires were used to repair the broken track and short out C74.
- Poor installation of trifilar transformer T2 had shorted two of the wires, with a large solder blob on the top side of the PCB
After the mistakes were repaired, the QDX tested perfectly.
7) Very low transmitter output power, deaf receiver and no image rejection
- Low output power was caused by a miswound output transformer T1, which was removed, rewound and reinstalled.
- The receiver problem was caused by the Rev 3a C41 via short, which was cut
Tested successfully after repairs.
8) Deaf receiver
In what appeared to be a PCB assembly manufacturing error, by visual inspection it was found that R17 was displaced by one component-length and therefore only one end of it was soldered to a pad. R17 is part of the potential divider that sets the midrail voltage in the BPF switch and QSD; without it the voltage was wrong and this drastically impairs performance of the subsequent amplifier.
9) Somewhat low output power on 40 to 80m
This was a case of messily wound toroids which simply required tidying up and some optimization to increase the power output to the acceptable level.
10) High receive current, approx 300mA
Normally receive current should be about 150mA for a recent board revision (Rev 3 and above). Lower for Rev 1/2. 300mA is way too high. It was found to be caused by one defective BS170. I always replace all BS170 together to ensure if possible, that they are all from the same batch. Necessary or paranoia I don't know but they are not expensive compared to the headaches so I do it.
11) Low power output, but non zero; 80m very high transmit current
Investigation showed the output waveform was also far from sinusoidal, varied significantly from one band to another, as did the power output. On checking the voltages at the toroids when switching bands (+/- keys in the transmitter test screen) I found that voltage at L4 & L10 was not correct (correct is +12V when 20/30m bands not selected, 0.6V when selected). This was found to be due to a defective Q5 transistor. Q5 was replaced and all became normal.
12) Weird receiver sweeps; incorrect peak levels, same for Image Sweep
Cause was a broken R19; this is the resistor which feeds the Si5351A Clk2 output to the receiver input. When broken, all you get is leakage due to PCB stray parasitic component values. R19 was replaced and all was well.
13) Deaf receiver
Reinstalled trifilar transformer after checking and confirming the correct wire continuities; after that the receiver worked normally. So there must have been a dry joint or wire mixup in the transformer T2.
14) Deaf Receiver (40m RF sweep, 40m image sweep)
Cause was, L12 BPF toroid wire was snapped after the 30m tap, so no signal was reaching the 40m and 80m taps. It was repaired with a piece of component lead cut-off and then all was well.
15) Huge current draw and no other signs of life, on switching on
Same as #5 above
16) Deaf Receiver
Trifilar transformer T2 had been forgotten completely, the constructor simply forgot to install it at all! Doh!
17) Low output power on all bands
The waveform of the output was also far from sinusoidal on any band. Investigation revealed the DC voltages at the LPF toroids were incorrect during Receive, when using the +/- keys in the Transmitter test screen to change band. For any given band, the voltage should be 0.6V when the band is selected, or +12V otherwise. But these voltages were somewhere in the middle. It was decided that likely two of the switching transistors were defective, for the two bands (of the three sets of LPF components) that showed incorrect voltages; accordingly Q4 and Q5 were replaced, then normal transmitter power was seen on all bands.
Next issue, the receiver was found to be deaf; reinstalled trifilar transformer after checking and confirming the correct wire continuities; after that the receiver worked normally. So there must have been a dry joint or wire mixup in the transformer T2.
18) Huge current draw and no other signs of life, on switching on
Same as #5 above
19) No output power on 20/30m bands
Simple case of poor soldering on the 20/30m LPF toroids, lumps of solder, dry joints, and enamel not properly burned off. Maybe someone was drunk on the job or asleep on duty.
20) No output power on 80m band
DVM continuity testing showed continuity through the 80m toroids (L2 and L6) but no continuity through the 47uH inductor L5 even though physically it looked fine. L5 was replaced and normal operation resumed.
21) Weird receiver sweeps; incorrect peak levels, same for Image Sweep
Same as #12 above
22) No sign of life whatsoever. About 100mA current draw.
Not an issue of a miswound output transformer. Furthermore, the output voltage of each of the two voltage regulators correctly measured 5V and 3.3V respectively. The LED was not lit and no connection was possible via the USB connector (the QDX did not show up as an audio device nor a Virtual COM Port serial device). I installed a 4-pin header connector at the ISP header pins and tried to connect via an STLink2 programmer, but could not connect (which is expected as the STM32 chip is locked). The onboard LED was dimly lit. The 25MHz TCXO and buffer were working correctly. Since I could not connect to the board in any way, and it showed no signs of life, I could not proceed further and had no choice but to consign this unit to the spare parts bin.
23) Low power output on the lower bands
C49 (470uF) capacitor had become disconnected. After repair, power levels were normal.
24) Low power output on all bands, and weird waveforms
Probing with an oscilloscope showed the DC voltage at the left side of D1, D2 and D3 was zero; it should be about 0.5 or 0.6V. Testing with a DVM in continuity mode showed a short from the top side of L1 (relative to the schematic) to ground. Closer inspection revealed a tiny solder bridge on the top side of the PCB, between the non-RF side of L1 and the nearby grounded pin of C11. Once the solder bridge was removed, everything was fine.
25) 20m full power output, but power output dropping progressively, too low on 80m and a weird waveform
As is usually the case when there is a low power output on 80m but not on higher bands, it was caused by insufficient forward bias current through the PIN diodes, which was caused by a broken trace under 470uF capacitor C49.
Forced firmware update
Apparently, if the firmware file gets somehow corrupted, so that it neither works nor can be put into firmware update mode to redo it, there is a trick which lets you force it into firmware update mode. The trick is to hold the I2C_SDA line low (pin 5 of the EEPROM chip) during power-up. Once the front LED starts flashing, then remove holding the pin low. This prevents the EEPROM from being read, which puts the QDX into firmware update mode so you can copy in a fresh firmware file via the USB port. Thanks to Kevin K3TL for this tip...