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 I roasted my QMX, do not repeat my mistake
#qmx
|
I had my QMX up and running with expected power consumption, good power output and fine image rejection and RF filter response.
When I prepared for portable use, I took out my QCX-mini power supply: It has two switched mode power supplies, one for 12V and the other for 6V I can switch between the supplies in order to change between 5W outut to less than 1W output This worked fine with the QMX also, but when I switched from 6V to 12V, the QMX just died. On afterthought I realise the the buck-converters in the QMX were running from 6V and with relatively high dutycycles When switching to 12V the dutycycle was much too high, resulting in a voltage spike on both th Vdd and Vcc lines. I found that D109, which protects the Vdd had a short. I removed D109 and QMX came to life... Transmission was fine with same power output as before. Reception seemed to be fine also But power consumption was 150mA instead of 100mA on 12V input. I noted the the linear 3.3V regulator was still active and getting warm (not hot). Despite? the LIN_REG_EN being low. Gate-source voltage of Q102 was about -1.5V, so Q101 has a leak turning on Q102. I connected gate and source of Q102 to turn it off, but it was still on, so Q102 has a short source to drain. I ran the PUTTY diagnostics, it shows nominal operation of all buck converters and nominal Vdd and Vcc But the RF-filter sweeps were terrible as compared before. And image rejection was not existant, it was fine before. I verified this by tuning the receiver 25kHz up and found the same signal with the same strength... I removed T401 and put new windings. No change in image rejection. I concluded that IC402 or IC403 or both were damaged. These ICs are powered by Vcc which does not have a protection Zener. So I gave up on repairing, thinking that more damaged circuits would continue to pop up. I have ordered an new kit, I hope that I will not have to wait too long for it... The existing QMX is operational as it is, but with performance issues and a little high power consumption I also have QCX-minis for 80 and 40 meters that help me survive the waiting time So be careful with sudden power supply changes to higher voltage. If you want to change supply voltage, either make it slow or switch off the QMX during the voltage change Ideal would of course be if Hans can implement power setting in the menu. 73 de SM5EIE / Gunnar |
|
Ouch, Gunnar!
toggle quoted message
Show quoted text
That is a painful lesson indeed! I hope you are able to restore that device at some point. JZ On Sun, Jul 30, 2023 at 2:06?PM <Ugglekatten@...> wrote:
|
|
Gunnar,
toggle quoted message
Show quoted text
This is quite troubling to me. What may happen if a power connector is a bit erratic? What if a line voltage disturbance should occur? These are real-world events. If they have the potential to destroy QMX, that is quite a problem. Hans, what say? JZ KJ4A On Sun, Jul 30, 2023 at 2:36?PM <Ugglekatten@...> wrote:
|
|
Bernie,
toggle quoted message
Show quoted text
The problem would appear to be in the dynamics of PWM regulator response. I think Gunnar nailed it. The 12 volt input probably never hit 15 volts, but it didn't matter. JZ On Sun, Jul 30, 2023 at 2:55?PM Ham Radio <bernard.murphy@...> wrote:
|
|
Thank you for sharing that experience.? That's a great lesson for us all. Glory to God in the highest, and on earth peace, good will toward men. Luke 2:14
On Sunday, July 30, 2023 at 02:06:17 PM EDT, Ugglekatten@... <ugglekatten@...> wrote:
I had my QMX up and running with expected power consumption, good power output and fine image rejection and RF filter response. When I prepared for portable use, I took out my QCX-mini power supply: It has two switched mode power supplies, one for 12V and the other for 6V I can switch between the supplies in order to change between 5W outut to less than 1W output This worked fine with the QMX also, but when I switched from 6V to 12V, the QMX just died. On afterthought I realise the the buck-converters in the QMX were running from 6V and with relatively high dutycycles When switching to 12V the dutycycle was much too high, resulting in a voltage spike on both th Vdd and Vcc lines. I found that D109, which protects the Vdd had a short. I removed D109 and QMX came to life... Transmission was fine with same power output as before. Reception seemed to be fine also But power consumption was 150mA instead of 100mA on 12V input. I noted the the linear 3.3V regulator was still active and getting warm (not hot). Despite? the LIN_REG_EN being low. Gate-source voltage of Q102 was about -1.5V, so Q101 has a leak turning on Q102. I connected gate and source of Q102 to turn it off, but it was still on, so Q102 has a short source to drain. I ran the PUTTY diagnostics, it shows nominal operation of all buck converters and nominal Vdd and Vcc But the RF-filter sweeps were terrible as compared before. And image rejection was not existant, it was fine before. I verified this by tuning the receiver 25kHz up and found the same signal with the same strength... I removed T401 and put new windings. No change in image rejection. I concluded that IC402 or IC403 or both were damaged. These ICs are powered by Vcc which does not have a protection Zener. So I gave up on repairing, thinking that more damaged circuits would continue to pop up. I have ordered an new kit, I hope that I will not have to wait too long for it... The existing QMX is operational as it is, but with performance issues and a little high power consumption I also have QCX-minis for 80 and 40 meters that help me survive the waiting time So be careful with sudden power supply changes to higher voltage. If you want to change supply voltage, either make it slow or switch off the QMX during the voltage change Ideal would of course be if Hans can implement power setting in the menu. 73 de SM5EIE / Gunnar |
|
Gunnar,
VCC does indeed have a protection zener ( D108 ). Check it. We may be looking at a problem here. I hope not. The MCU is quite sensitive to the smallest and shortest of voltage spikes. When it dies then repairs are not possible. It would be instructive if detailed boot sequence measurements were make on the VCC and VDD voltages. Like what the maximum currents occur through D108 and D109 and their duration. A maximum-reading DMM may show the voltages reaching the rated voltages of D108 and D109 and they probably should not if the switch-mode regulators are doing their job. Just to be certain. 73, Don |
开云体育Doesn't D108, a 5.6V Zener 'protect'
VCC?
Chris, G5CTH
On 30/07/2023 20:39,
Ugglekatten@... wrote:
Well I guess D109 heroically sacrificed itself to save the Vdd bus
|
|
Depending on how the regulator circuit was designed, there could well have been a brief interruption in an SPDT switch that let the voltage rise to maximum for an instant, which is all it takes.
toggle quoted message
Show quoted text
Yes, we’d all design it for 12V and switch in another shunt for 6V, but who knows what the actual designer did or what his control loop parameters look like. 73, Willie N1JBJ On Jul 30, 2023, at 2:06 PM, Ugglekatten@... wrote: |
|
Imagine your QMX is connected to a power supply. You flip the switch on the supply and its output begins to rise.? At some point when the supply voltage is high enough and the QMX linear regulator Vdd supply is high enough the processor fires up and the PWM regulators start to operate. When the processor sees that ( internally to the PWM circuits) Vdd and Vcc are ready, the processor throws a big switch and now the PWM regulators are in charge of delivering Vdd and Vcc to QMX 's circuits. Here is what I worry about. If the external supply voltage continues to rise after the handoff occurs, and does so at a rate faster than the slew rate of the processor PWM control loop, output spikes will take place at Vdd and Vcc. The Zeners there may be inadequate to safely absorb that blow and downstream damage may occur. I believe this is what happened when Gunnar switched away from his 6V power supply and onto a 12V supply. Gunnar gets kudos for understanding this first. I would love to be convinced that this cannot happen. JZ KJ4A? On Sun, Jul 30, 2023, 5:00 PM William Smith <w_smith@...> wrote: Depending on how the regulator circuit was designed, there could well have been a brief interruption in an SPDT switch that let the voltage rise to maximum for an instant, which is all it takes. |
开云体育The way to keep voltage spikes from occurring on a power line is to place a largish capacitor across the line. It will absorb any spike and I hope give other circuits time to respond.Dave On Jul 30, 2023, at 17:43, John Z <jdzbrozek@...> wrote:
|
|
Another wrinkle to this power supply story. As the Zener diodes at the PWM outputs are heroically trying to keep the output voltage near the set point of the control loop, they deny the creation of a feedback error signal that proportionally reflects how much the PWM pulse width needs to be reduced. In the worst case with the Zener voltage at its spec minimum (e.g. a 3.6V Zener showing an actual Vz of 3.3 volts) there may be no feedback error signal at all. Ugh! JZ On Sun, Jul 30, 2023, 5:43 PM John Zbrozek <jdzbrozek@...> wrote:
|
|
In the QMX schematic I see an input to the processor labelled ADC_BATT. It connects to the 12V rail via a resistor voltage divider and 0.1uF bypass cap. This looks like it could be useful to monitor the incoming power supply voltage and enable self-protection mechanisms in QMX. The time constant of that RC network is 1 mS though and that might not be fast enough for effective protection. JZ On Sun, Jul 30, 2023, 7:14 PM John Zbrozek <jdzbrozek@...> wrote:
|
|
Hello JZ, all The zener diodes are on both the 3.3V supply (D109, 3.6V) and 5V supply D108 (5.6V). These diodes are not intended to be part of the circuit in normal operation and use, even at power-up. In other words the circuit would work perfectly fine without these diodes but for reasons of paranoia, and the possibility to protect perhaps against some kind of short term issue, I included them. They are a short-term protection, intended to protect the downstream circuits to some degree if there is any spike for example. In the event of a serious problem they wouldn't last long - but this would apply even to much larger high power rated devices too.? Control loops are a very deep topic. Probably a lot of time could be invested to try to find ways to improve it. We don't really know what spike conditions existed in Gunnar's circuit. I think this is a very difficult subject and designing a SMPS hardware circuit or a software feedback loop as in QMX that is resilient to any and every kind of spike abuse would be a rather tough target.? As I think I've mentioned here, there IS an attempted software protection system in QMX, which continuously monitors the supply voltage and applies a "performance envelope" limit to the PWM; the idea is that the QMX knows what its typical current consumption should be, and it has a built-in curve which kind of a representation of the attached measured curve, plus a margin (for component value variation). It limits the PWM so that if things go wrong or the supply voltage changes suddenly, it should provide some protection.? So there ARE protective measures in hardware (the zeners) and in software. But clearly in Gunnar's case the protective measures have not been enough. Well not completely enough, since something has survived but not everything.? 73 Hans G0UPL |
|
Hans, Thank you for that response. Yes, protecting against every potential threat is an impossibility and it is good to gain a better understanding of the protections you have built in. On the forum we have now seen a couple of Zener failures and an MCU failure none of which seem to be attributable to the slip of a probe or some other known fault. While these are small numbers, QMX is also still very young. I think this class of failure bears close following and scrutiny, as your power supply solution is novel and there is limited experience with it. I hope there are no additional surprises like that which presented itself to Gunnar. As always, Regards and Best Wishes, JZ On Mon, Jul 31, 2023, 1:48 AM Hans Summers <hans.summers@...> wrote:
|
|
Just for the sake of completeness, I made recordings of the transition of my portable power supply from 6V to 12V operation
There are some spikes in the beginning coming from contact bouncing of the SPDT mechanical switch The spikes look a little different each time, the shown recording is the worst out of 10 recordings. The rise from 6 to 12V is linear and takes a little less than 3 milliseconds The portable supply contains four 18650 Li-Ion cells The 12V output is driven by a step up/down converter (boost/buck) and the 6V by a step down (buck) , There is a 1000 uF capacitor between the positive and negative power output. (Obviously not so good at filtering out spikes...) This was included to prevent the QCX-mini from restarting when switching beteen the two power levels SM5EIE / Gunnar |
Manuel; DL2MAN
Don, ND6T
to navigate to use esc to dismiss