|
My QMX released the magic smoke yesterday and I quickly traced the smoke to a crater in D108, the 5.6V zener on the 5V PSU board.?
Having recently read Hans’ excellent SMPS analysis ( ) it got me wondering what could have caused the fault. I had been powering my QMX from a USB-C battery and this 12v cable (). I then remembered that USB-C PD adapters typically start with the output at 5V and then shortly afterwards increase the voltage to that negotiated by the device (in this case 12V). This would generate exactly the voltage pattern Hans has told us to AVOID. I took a trace of the voltage power-up sequence, and this is the result (no QMX’s were harmed in creating this trace).
Output power starts at 5V for 120ms before increasing to 12V, perfectly creating Hans’ disaster scenario and causing D108 to sacrifice itself to protect the rest of the QMX. Thankfully the rest of my QMX appears to have survived unharmed, thanks to D108.
Conclusion: don’t use a USB-C battery to power your QMX, unless you’ve confirmed your adapter behaves differently!
Cheers,
Dave M0JTS
?
|
Ouch, Dave! Yup, a direct hit... That is a valuable learning experience for all of us. You mentioned the PD negotiation protocol. No such thing is in QMX!? Maybe it should be in future products from QRP Labs? USB PD batteries, bricks, etc. are becoming ubiquitous and could be an excellent addition to the QRPer's?toolkit.
JZ KJ4A
On Fri, Aug 11, 2023 at 5:08?AM david mugridge < dave@...> wrote: My QMX released the magic smoke yesterday and I quickly traced the smoke to a crater in D108, the 5.6V zener on the 5V PSU board.?
Having recently read Hans’ excellent SMPS analysis ( ) it got me wondering what could have caused the fault. I had been powering my QMX from a USB-C battery and this 12v cable (). I then remembered that USB-C PD adapters typically start with the output at 5V and then shortly afterwards increase the voltage to that negotiated by the device (in this case 12V). This would generate exactly the voltage pattern Hans has told us to AVOID. I took a trace of the voltage power-up sequence, and this is the result (no QMX’s were harmed in creating this trace).
Output power starts at 5V for 120ms before increasing to 12V, perfectly creating Hans’ disaster scenario and causing D108 to sacrifice itself to protect the rest of the QMX. Thankfully the rest of my QMX appears to have survived unharmed, thanks to D108.
Conclusion: don’t use a USB-C battery to power your QMX, unless you’ve confirmed your adapter behaves differently!
Cheers,
Dave M0JTS
?
|
The cord you have linked from Amazon has the wrong sized connector. I wonder if that played a part in this? Intermittent connection could cause a 12v to? 0 to 12v jump anytime the cord is moved...
-mike/w1mt
Ouch, Dave! Yup, a direct hit... That is a valuable learning experience for all of us. You mentioned the PD negotiation protocol. No such thing is in QMX!? Maybe it should be in future products from QRP Labs? USB PD batteries, bricks, etc. are becoming ubiquitous and could be an excellent addition to the QRPer's?toolkit.
JZ KJ4A
On Fri, Aug 11, 2023 at 5:08?AM david mugridge < dave@...> wrote: My QMX released the magic smoke yesterday and I quickly traced the smoke to a crater in D108, the 5.6V zener on the 5V PSU board.?
Having recently read Hans’ excellent SMPS analysis ( ) it got me wondering what could have caused the fault. I had been powering my QMX from a USB-C battery and this 12v cable (). I then remembered that USB-C PD adapters typically start with the output at 5V and then shortly afterwards increase the voltage to that negotiated by the device (in this case 12V). This would generate exactly the voltage pattern Hans has told us to AVOID. I took a trace of the voltage power-up sequence, and this is the result (no QMX’s were harmed in creating this trace).
Output power starts at 5V for 120ms before increasing to 12V, perfectly creating Hans’ disaster scenario and causing D108 to sacrifice itself to protect the rest of the QMX. Thankfully the rest of my QMX appears to have survived unharmed, thanks to D108.
Conclusion: don’t use a USB-C battery to power your QMX, unless you’ve confirmed your adapter behaves differently!
Cheers,
Dave M0JTS
?
|
Dave,
Great catch - thanks so much for sharing!
My QMX powered up a few times, then died. I haven't started trouble-shooting yet, but I bet that's what happened.
Assuming the negotiation happens when the cable is plugged into the power bank or wall-wart, then maybe I happened to plug in the USB-C end first a few times and was OK.?
I really like using the power banks.? I even have one of those high-capacity car jump-starter ones that also has a PD port.? It's always in the car, so with a trigger cable it's a backup for the LiFePO4.
Cheers,
Chuck
|
Good catch. In practise, the slightly larger connector makes it ever-so-slightly more snug, not less so. Also, Hans' tests indicate that an intermittent 0-12v connections is okay. A 5v-12v jump is the killer.
|
Hans, can I suggest a possible work-around for this USB-C PD scenario?
On initial power up, can your firmware wait (say) 500ms to allow the input voltage to stabilise before starting the SMPSU? That would neatly avoid any power spike being generated.
That would allow the surprising number of us USB-C PD fans to continue using our batteries without fear.
Cheers,
Dave M0JTS
|
The connector is the other way round, a
2.1mm centre pin on the QMX and a 2.5mm male socket on the plug.?
I've seen this a lot on other things and the equipment gets an
undeserved reputation for unreliability.
Chris, G5CTH
On 11/08/2023 12:33, david mugridge
wrote:
toggle quoted message
Show quoted text
Good catch. In practise, the slightly larger connector makes it
ever-so-slightly more snug, not less so. Also, Hans' tests
indicate that an intermittent 0-12v connections is okay. A 5v-12v
jump is the killer.
|
2.5mm plug on a 2.1mm jack is not snug at all, hence the problem.
-mike/w1mt
|
That jump in voltage is the expected behavior for USB-PD. USB always starts at 5V, then the device (the power sink) negotiates with the source for the desired voltage and power. In your case, the "device" is a USB-PD sink that is built into the cable; cables and standalone USB-PD sinks allow a USB-PD source to power a dumb device that does not implement USB-PD negotiation, and that may not even have a USB port.
You should be able to safely use a USB-PD sink with the QMX by connecting the power cable first, THEN turning on the QMX. That sequence will cause the 12V negotiation to happen before the QMX powers up. Test your sink to make sure that it outputs a steady 12V even before it sees a load; some might not do the negotiation until something that requires power is connected, though the few I have personally tried deliver the requested voltage with no load.
As somebody pointed out, disconnecting and reconnecting the power bank from the USB-PD sink will cause the voltage negotiation to happen again, which means that the catastrophic voltage increase from 5V to 12V will recur. So don't do that; make certain the connections are secure. Using a cable with the wrong size of plug could cause disconnection to happen, so using a cable that is intended for a device with a 2.5mm?center pin with a QRP Labs rig is definitely NOT recommended. Get a cable with a 5.5x2.1 plug.
The other important thing is to make sure that your power bank actually delivers 12V. That voltage is no longer a required part of the USB-PD spec; it was required in an earlier version for sources that can deliver 12V or more, but support is now optional. Another option is to build your radio for 9V operation and use a 9V USB-PD sink; 9V is still a required part of the spec for sources capable of delivering 9V or more.
The reason for the qualification of "capable of delivering" is that the USB-PD spec covers a wide range of sources and sinks. A source that can only deliver 5V, or that can only deliver 5V or 9V, can still be USB-PD sources that comply with the specification. Such a source is even?useful because it can be asked to?deliver more than 0.5A -- up to 3A, and up to 5A at 20V and up. Cell phone chargers are often in the category of only delivering 5V or 9V, though phones that can charge at a rate higher than 27W will require a more capable charger to reach maximum speed.
The USB-PD spec also includes PPS (Programmable Power Supply), an optional extension that allows the sink to ask for ANY voltage in the range from 3.3V to 21V from the source in 20mV steps. (Since the starting voltage is always 5V, all USB-PD sinks must be able to tolerate 5V.) The EPR (Extended Power Range) extension allows selection of any voltage from 15V to 48V in 100mV steps; those higher voltages are used by sources that can deliver more than 100W. Most power banks don't implement PPS; you're more likely to find it in wall chargers.
The connector is the other way round, a
2.1mm centre pin on the QMX and a 2.5mm male socket on the plug.?
I've seen this a lot on other things and the equipment gets an
undeserved reputation for unreliability.
Chris, G5CTH
On 11/08/2023 12:33, david mugridge
wrote:
Good catch. In practise, the slightly larger connector makes it
ever-so-slightly more snug, not less so. Also, Hans' tests
indicate that an intermittent 0-12v connections is okay. A 5v-12v
jump is the killer.
|
Mike, Chris, Apologies - my mistake, doh. You're absolutely correct - the 2.5mm is not snug. Time for me to buy the correct cable.
|
It's a bit of a worry that the link from the OP shows a Goshyda cable
that does not mention a "PD induction chip" while some other cables
from them do.
?
JZ
toggle quoted message
Show quoted text
On Fri, Aug 11, 2023 at 8:57?AM david mugridge <dave@...> wrote:
Mike, Chris, Apologies - my mistake, doh. You're absolutely correct - the 2.5mm is not snug. Time for me to buy the correct cable.
|
I have one of these??in my kit and use it to monitor the voltage from my power banks/USB trigger cables so I can ensure that it has stabilized before turning on /plugging in.? It measures the current and charge (mA-hours) also and it even allows setting voltage and current limits although I have not used that feature until I can put an oscilloscope on it and understand the system's behavior going into and recovering from a fault.? The load resistors and clip leads are handy additions.
Plugging the cable into the power source first, then into a radio, should be a safe process as long as a reliable connection is made to avoid power (and voltage) cycling due to intermittent contacts.
73, Don N2VGU
|
On 2023-08-11 08:37, Shirley Dulcey KE1L wrote:
The other important thing is to make sure that your power bank actually delivers 12V. That voltage is no longer a required part of the USB-PD spec; it was required in an earlier version for sources that can deliver 12V or more, but support is now optional. Another option is to build your radio for 9V operation and use a 9V USB-PD sink; 9V is still a required part of the spec for sources capable of delivering 9V or more.
?
I have an Anker 65W USB-PD wall wart, with two USB-C and one USB-A outputs:
https://www.amazon.com/gp/product/B09C5RG6KV
Strangely enough, 12V is ONLY supported on the USB-A output, so I had to search for (and found) 12V trigger cables with USB-A connectors.
73,
-- Dave, N8SBE
|
So after a month using my QMX with great success. However, it's now broken ;-( ... likely due to this current discussion. It was powered up, not transmitting, and the power cord going to a 12V trigger USB-C device got bumped when I turned my PC a bit. So I suspect it took a 5v to 12v power hit.
I've not opened it up yet, but the QMX powers up, menu works, hardware diagnostics on the terminal all look good including showing power from the 3.5v and 5V supplies in the hardware diagnostics as expected. But I just get static noise on the audio output, and all the RF filters sweeps which used to look great, are all garbage. Paddle input does not work in Practice Mode like it used to.
Would these symptoms point to that D108 Zener, or do I have something else gone wrong?
I'll be digging into it later this week to see what I can see.
Thanks!
Loved this little QMX while it worked, so am anxious to get her back on the air!
73
de Roy - KI0ER
|
Forgot to mention that I am getting about 5W out when I transmit ("T" key) from the hardware test Diagnostics screen. And I also did a Factory Reset (for the first time), and that has not helped; still get nothing but static on audio.
Ideas on what I may have fried?
Roy - KI0ER
|
Brainstorming. Don't have the schematic in front of me... How
about if something were added to rob the SMPS of drive, when that
rising voltage transient occurs?
On 9/2/23 16:59, Roy - KI0ER via
groups.io wrote:
toggle quoted message
Show quoted text
So after a month using my QMX with great success. However, it's
now broken ;-( ... likely due to this current discussion. It was
powered up, not transmitting, and the power cord going to a 12V
trigger USB-C device got bumped when I turned my PC a bit. So I
suspect it took a 5v to 12v power hit.
I've not opened it up yet, but the QMX powers up, menu works,
hardware diagnostics on the terminal all look good including
showing power from the 3.5v and 5V supplies in the hardware
diagnostics as expected. But I just get static noise on the
audio output, and all the RF filters sweeps which used to look
great, are all garbage. Paddle input does not work in Practice
Mode like it used to.
Would these symptoms point to that D108 Zener, or do I have
something else gone wrong?
I'll be digging into it later this week to see what I can see.
Thanks!
Loved this little QMX while it worked, so am anxious to get her
back on the air!
73
de Roy - KI0ER
|
There is such a solution, please refer to /g/QRPLabs/message/108946
With this mod, the QMX will be immune to input voltage steps, and I believe it will not confuse the software regulator.?
I have sent for the parts and will modify my QMX next week.?
This is such a smart fix and should be a candidate for the rev 2 board!?
73 de SM5EIE /Gunnar?
|
|
|
Hi Gunnar, all
A few comments/questions...
1) this circuit has been proposed, but has anyone actually tried it for real, and then verified that on the step change from 6V to 12V it does protect the circuit without killing the zener?
2) As far as I can see the transistor will be shorting the processor's GPIO pin directly to ground. In other words the transistor will be fighting whatever transistor is inside the microcontroller. Does this worry anyone??
3) More fundamentally... Does it work at all, even theoretically? So the voltage input suddently jumps from 6V to 12V. The SMPS is then running at a too high duty cycle and the voltage at the SMPS therefore jumps. Ordinarily the 5.6V zener is supposed to eat this spike by passing the current to ground. It has to do that for up to 2ms until the processor control loop detects the situation and acts to fix it. Apparently that couple of milliseconds is a bit too much for a 0.5W zeners and there is a possibility of killing it. Which brings me to my question. Now that same current is going through the zener and though the base-emitter junction of the transistor. It could be for a much shorter time because it will (hopefully) win the fight with the GPIO pin and shut down the PWM; still, by that time substantial energy has been stored in the 330uH inductor and now that has to dissipate as a current pulse onto the base-emitter junction of the transistor. Why is the transistor going to survive this?
|
Hans, The proposed circuit has problems but the concept shows promise.? I didn't look at it very carefully. I saw the issue of "?the transistor will be shorting the processor's GPIO pin directly to ground. In other words the transistor will be fighting whatever transistor is inside the microcontroller." That could be fixed by adding a resistor in series with the GPIO output pin to limit the GPIO pin's current when the transistor turns on.? I haven't looked at the other more difficult questions you bring up or whether this fixes the input voltage jump issue at all or is it not a robust fix.? I do though like the idea of a "dead-man switch" so no matter what the software is doing, the circuit should have some hardware protection if it can be implemented at very low hardware cost with relatively few components.? OMG the density, the density of that wonderful 3D art packaging.? One can't appreciate it until you get the plugging all the boards together in the final assembly steps!
-Steve K1RF
------ Original Message ------
Date 9/3/2023 7:05:52 AM
Subject Re: [QRPLabs] #qmx Don’t Use USB-C PD to power your QMX?!
Hi Gunnar, all
A few comments/questions...
1) this circuit has been proposed, but has anyone actually tried it for real, and then verified that on the step change from 6V to 12V it does protect the circuit without killing the zener?
2) As far as I can see the transistor will be shorting the processor's GPIO pin directly to ground. In other words the transistor will be fighting whatever transistor is inside the microcontroller. Does this worry anyone??
3) More fundamentally... Does it work at all, even theoretically? So the voltage input suddently jumps from 6V to 12V. The SMPS is then running at a too high duty cycle and the voltage at the SMPS therefore jumps. Ordinarily the 5.6V zener is supposed to eat this spike by passing the current to ground. It has to do that for up to 2ms until the processor control loop detects the situation and acts to fix it. Apparently that couple of milliseconds is a bit too much for a 0.5W zeners and there is a possibility of killing it. Which brings me to my question. Now that same current is going through the zener and though the base-emitter junction of the transistor. It could be for a much shorter time because it will (hopefully) win the fight with the GPIO pin and shut down the PWM; still, by that time substantial energy has been stored in the 330uH inductor and now that has to dissipate as a current pulse onto the base-emitter junction of the transistor. Why is the transistor going to survive this?
|
Shirley Dulcey KE1L
Donald S Brant Jr
Paul - AI7JR