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 Fail on power up
|
?Built 2nd QMX. Initial continuity checks good. Powered with current limited supply- no current at first. Left encoder push gave 155 mA steady draw. Powered down.
I made the bad assumption that all was well. (I did not follow my own advice.) My next power up was with my usual 12V battery supply. Left coder push, loaded 005 firmware. This appeared to go well, but QMX did not reboot and turn on LCD light. I sequenced power but no response to left encoder push. A second or two later C107 and Q108 self destructed. No obvious damage elsewhere, but something must have drawn high current with 5V p/s turn on. All boards were in place when this happened. Any ideas where to start? Dan de NM3A at mail dot com -- 73, Dan? NM3A |
|
Did D108 also fail? JZ On Tue, Aug 1, 2023, 2:57 AM Daniel Walter via <nm3a=[email protected]> wrote: Built 2nd QMX. Initial continuity checks good. Powered with current limited supply- no current at first. Left encoder push gave 155 mA steady draw. Powered down. |
|
OK Dan. I asked because C107 probably failed due to overvoltage. If so D108 would have been taking a beating as well. That might happen if the PWM regulator control loop failed. If overvoltage at the 5V output did take place there could be a lot of downstream damage. The entire Receive chain and the driver for the finals would have been under stress, maybe more. JZ On Tue, Aug 1, 2023, 4:08 AM Daniel Walter via <nm3a=[email protected]> wrote: Don't think so. High reverse resistance on D108 with DMM; 190 ohm forward voltage. DMM uses about 0.75 V for testing.? |
|
Backwards almost never happens in an automated assembly process. In a
rework/repair station maybe, but even that is low likelihood. JZ On Tue, Aug 1, 2023 at 1:43?PM Daniel Walter via groups.io <nm3a@...> wrote:
|
|
I am slowly working through troubleshooting. 5V bus reads 450 ohms. Powering 5V bus through external power supply shows 90 mA, which seems reasonable. I want to try to power the 3.3V bus with an external power supply. Is this reasonable?
|
|
Dan, that may be a little too close for comfort. Ten volt 220uF
tantalums are available in the same package size. Stack two? Put one on the PS board, one somewhere else on the 5V net? JZ On Wed, Aug 2, 2023 at 5:12?PM Daniel Walter via groups.io <nm3a@...> wrote:
|
|
Maybe you should drag your feet a little bit on this repair until more
toggle quoted message
Show quoted text
is understood about root cause and corrective action. JZ On Wed, Aug 2, 2023 at 5:28?PM John Zbrozek <jdzbrozek@...> wrote:
|
|
My concern, Dan, is that replacing the cap with a sturdier cap without understanding and correcting why it failed in the first place may just move the next failure to another, and costlier, place. As Allison said "When Hans sees a problem, he fixes it." I think that this has his attention. I would go slow here. If you decide to move forward now maybe the best thing is to replace with an identical cap. Better it should be the failure point than some fancy piece of silicon. FWIW,? JZ KJ4A? On Wed, Aug 2, 2023, 6:39 PM Daniel Walter via <nm3a=[email protected]> wrote: John, |
|
Agreed, John. Original part is also a 6.3 V tantalum, so replacing it would be an identical swap. Whether that was the original failure or not, I think it's failure fried more than the obvious parts on the 5 V p/s board. So likely the active devices?should probably be replaced and possibly?the entire 5 V p/s board . However, at this point, I am going to limit my evaluation to checking resistances and voltages. I don't want to risk swapping my good QMX parts on this one yet.? -- |
|
Dan,? I think the failure scenario went something like this: On power-up, after handoff from the 3.3 v? linear regulator to the buck converter, a disturbance on the power input, or a rapid continuing rate of rise of the input, overwhelmed the PWM control loops resulting in a serious overshoot of the 3.3 volt and 5 volt rails. Zener diodes present there heroically tried to limit the overshoot but became damaged or destroyed in the process and the voltage continued to rise. The tantalum capacitor on the 5 volt rail now saw a voltage well above its rating and it failed as a short. The elevated current flow resulting from that created an overvoltage Ldi/dt response that exceeded the 30 volt rating of Q108 (an AOD403 PMOS FET) and it too failed. It is unlikely that Q108 failed directly from over-current as it is rated to handle 70 amps! The other two identical transistors in series ( Q103, Q105) would have had to handle the over-current as well. Those don't appear to have failed on your board. The root cause here would appear to be the dynamic response of the two buck converter processor controlled PWM loops. This architecture is a rather ambitious design approach.? As the MCU currently samples and cycles these loops on a 1mS schedule, bad stuff can happen before it has a chance to react. A solution may be found in firmware change. As Hans has mentioned in a couple of posts here, there are protective measures in place and firmware knobs and levers are available to him to address this. If you have escaped additional silicon damage on the 3.3 volt and 5 volt distributions, consider yourself lucky so far. I think your 5V board is repairable but also advise against doing too much with your QMX just yet. If you can, scope your power source and see how it behaves on start up. Use a representative load other than QMX! The control loops were disturbed by something and it may be your power supply itself. The loops need to become more robust against such input problems. That's the full story as I see it presently, Dan. 73, JZ KJ4A? On Thu, Aug 3, 2023, 5:48 AM Daniel Walter via <nm3a=[email protected]> wrote:
|
|
I am still not convinced that the cause for failures is in a ratty supply. My MCU failed from a VDD (3.3 volt) failure. If the supply was indeed interrupted due to a connector then we are in for a rocky road if we are to take it to the field! How many times have we experienced a loose power coaxial connector while outdoors in the wild? For me, a lot. All of my other radios are quite tolerant of this sort of interruption and I have never incurred damage from it. My QMX death, however, was in tightly controlled conditions on a test bench with a very expensive (and reliable) power supply feeding the QMX through a tight connection.
I agree with the increased sampling suggestion. With that and the current filter capacitors it should not be a problem. I do question the high operating voltages. Is the target VCC voltage actually 5.4 volts and the VDD voltage actually 3.6 volts as the FDIM presentation slides suggest? What would be the consequence of just running at 5.0 volts and 3.3 volts as the target SMPS potentials? How about below those levels? Usually performance is guaranteed on most logic devices quite a ways below those listed. By not crowding the absolute upper limits of the components then the safety margin will be immensely increased. What would suffer? I would gladly forfeit a dB of performance or 20% increase in battery drain if it meant that my transceiver didn't die irretrievably at the drop of a hat. With no warranty and with no chance of replacing the MCU ever, reliability becomes really a BIG deal. Those decreased voltages can be set in the firmware quite easily, I believe. Might even run cooler. Am I correct? Or has it been tried and rejected? 73, Don? |
|
Hello Don You're misreading the FDIM presentation slide. What you're reading is the Vpp which is 3.670V on the 3.3V rail and 5.445V on the 5V rail; the Vpp measurement being the difference between the minimum and maximum voltages observed in the sample set, the little negative spike at the switch-on is the cause of that elevated value. The actual regulated values are indeed 3.3V and 5.0V. If you look at the traces on the right hand side of the screen, and take into account the grid is 1V/div, you will see that the target and observed values match (3.3V and 5.0V resp). 73 Hans G0UPL On Thu, Aug 3, 2023 at 3:39?PM Don, ND6T via <nd6t_6=[email protected]> wrote: I am still not convinced that the cause for failures is in a ratty supply. My MCU failed from a VDD (3.3 volt) failure. If the supply was indeed interrupted due to a connector then we are in for a rocky road if we are to take it to the field! How many times have we experienced a loose power coaxial connector while outdoors in the wild? For me, a lot. All of my other radios are quite tolerant of this sort of interruption and I have never incurred damage from it. My QMX death, however, was in tightly controlled conditions on a test bench with a very expensive (and reliable) power supply feeding the QMX through a tight connection. |
|
One thing to consider.? I took a look at the 470uF 6.3V tantalum capacitors C107 (on PCB #1? for the Vcc (5V) supply and C106 (On PCB #2) for the VDD (3.3V supply.) C107 might not have adequate voltage de-rating.? Tantalums typically require a 50 to 60% voltage de-rating.? (See link below for Vishay's guidelines on tantalums.)? In addition, in aerospace work I was involved with, tantalums had to be surge rated (which most are not) to prevent potential failures under high current surge conditions, which typically occur at power supply turn-on.? ?Just a thought on possibility for at least some failures.
- Steve K1RF ------ Original Message ------
From "Don, ND6T via groups.io" <nd6t_6@...>
Date 8/3/2023 8:38:59 AM
Subject Re: [QRPLabs] QMX Fail on power up I am still not convinced that the cause for failures is in a ratty supply. My MCU failed from a VDD (3.3 volt) failure. If the supply was indeed interrupted due to a connector then we are in for a rocky road if we are to take it to the field! How many times have we experienced a loose power coaxial connector while outdoors in the wild? For me, a lot. All of my other radios are quite tolerant of this sort of interruption and I have never incurred damage from it. My QMX death, however, was in tightly controlled conditions on a test bench with a very expensive (and reliable) power supply feeding the QMX through a tight connection. |
开云体育Where are the PWMs started? In the boot code or the user installed firmware? If in the boot code then it can only be changed by Hans at his desk. Chris, G5CTH
|
Don, ND6T
to navigate to use esc to dismiss