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
Search
QMX: Requiescat in Pacem
|
My QMX just passed away - requiescat in pacem !
You might remember, that I built it in August (/g/QRPLabs/message/107523). I did only a handful of FT8 QSOs with it, to validate, that it is working and was not very happy with the instability of the CAT connection to WSJT-X/JTDX. I did several firmware updates up to the current 014 firmware, but the CAT instabilities did not go away and other usage than for FT8 was impossible due to the missing AGC. Today I wanted to use it, to record the output of a QSE (<-10dBm output), coupled through a 50dB attenuator to the input of the QMX. Checking the signal through a headset was fine and then I connected the QMX USB connector to my laptop, to be able to record the QMX audio. Most probably this killed the QMX, which was connected to 12V from a battery!?? Display of the QMX was off. I connected it to a lab power supply limited to 500mA and tried to switch the QMX on with the left button. Current limit immediately applied and the QMX did not switch on. Even with a reduction of the supply voltage to 6V, the current limit still applied, when trying to switch the QMX on. I have to say, that I completely lost the trust into the CPU controlled switching regulator in the QMX. My guess is, that under certain situations during the handshake between the QMX and the PC via the USB port, the timing of the switching regulation loop is negatively impacted and the output voltage might raise uncontrolled. I don't think, that I want to spent a huge amount of time to find out, what has been killed QMX internally, given the risk, that it might be the CPU... -- 73, DB2OO -Joerg |
|
Hello Joerg ? I have to say, that I completely lost the trust into the CPU controlled switching regulator in the QMX. In any firmware version from?1_00_009 onward, the switching regulation loop runs in the highest priority interrupt process. NOTHING can stop it. It certainly doesn't get delayed by any USB handshaking between QMX and PC. The USB handling actually happens in the lowest priority interrupt, it is the least critical thing and everything else is prioritized above it. In any system fault that crashes (resets) the processor, the PWM outputs float, which considering they are pulled low in hardware, means the SMPS converter switches are OFF. So the hardware defaults into a safe state. The SMPS are one of the complex and daring parts of the QMX and it's too tempting to blame this part of the circuit if things go wrong. I believe (contrary even to my own initial expectations) that the actual reliability of this circuit is very high, much better than we would intuitively feel; our intuition is ruined by a fear of its complexity.? I wonder if you might have had ground potential differences between multiple power supplies. This is a quite easy trap to all into (don't ask me how I know). In your case it sounds like you had three power supply domains. The 12V battery powered QMX, your laptop, and whatever power supply was powering your QSE. Your QMX "ground" wouldn't have been your shack ground, and most probably your laptop ground wouldn't either (laptop power supplies normally "float"), and I don't know about what was powering your QSE. I've seen situations where I had an ungrounded lab variable supply and half mains voltage appeared?on its "negative" terminal. Sure, at very high impedance, it's not like you feel anything when you touch it. But I do wonder if you might have inflicted some kind of discharge when you connected these various power supply domain grounds together and thereby forced them all to the same level? I don't know, just speculating. 73 Hans G0UPL |
|
Even if it is not yet Easter my QMX stood up from the deaths yesterday:
Based on Hans' feedback to my dead QMX as was pretty sure, that the CPU would have been killed due to a voltage difference between
Yesterday I disassembled my QMX from the enclosure and pulled out both power supply (SMPS) PCBs. I then added jumper wires for GROUND and VDD (3.3V) to the main board PCB#2 connector and supplied 3.3V (JP104 Pin 4 and Pin 2 GND) from a current limited lab power supply and slowly increased the voltage while watching the current. At 3.3V I saw a current consumption of approx. 150mA --> Hooray, CPU not shorted, most probably OK! I then checked the PWM_3V3 and PWM_5V signals on the main board, that are directly coming from the CPU --> 116.4kHz rectangle signals --> CPU is working. As a next step I put in the SMPS PCB#1, to create the 5V and pass through the 12V, that I connected through the power supply connector with 3.3V still being supplied directly to the main board. Again I had high current consumption on 12V --> problem must be on SMPS PCB#1! I pulled out SMPS PCB#1 and put in SMPS PCB#2 for 3V3 supply. I fed 12V to the main board 12V connector of PCB#1 (JP101 pin 4) and saw 3.3V coming out of SMPS PCB#2 and a nice 116.4kHz signal on PWM_3V3: I was now suspecting, that the 5.6V Z-diode D108 on SMPS PCB#1 had a short, desoldered it and soldered in a new one. Unfortunately a 3W 5.6V Z-Diode, that I had ordered, was too big to fit into this place, so that I had to put in a glass 5.6V Z-diode. Put in PCB#1, connected 12V to the power connector of the PCB, slowly raised to voltage and again hooray, QMX was working again!! I updated to the firmware from 1_00_012 to 1_00_017 without any problems and the QMX seems to be working fine. That D108 obviously took up its protection role and died with a short is for me a strong indication, that the SW based switching regulator still has problems... I think my use case to provide power to the QMX from one power supply and have a laptop conencted with a different power supply with at least one ungrounded is not an common scenario, if you run FT8. -- Vy 73, DB2OO -Joerg |
|
The correct sentence is "Requescat (third person, indicative) in preposition requiring ablative case answering the question where) pace ablative case object of preparation.? This is the end of your unwanted Latin grammar lesson. On February 21, 2024, at 23:27, Steve Roberts <robers97@...> wrote: Thank you for this post.? I am working through an early build failure and this help me isolate it to the 3v3 circuit.? Well really it just gave a great guide on how to work through PON issues. |
Leland Lannoye, W9HIF
to navigate to use esc to dismiss