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+ V2 - I've killed it
|
I'm sufficiently unhappy about the power controllers on the QMX that I'm running my QMX using a couple of linear regulators and accepting the increased current and dissipation.? And, when the QMX+ was changed to have the power controllers built into the main board decided that I wasn't going to get one.? I might get a high bands kit for my QMX, I might even get some replacement controllers for it, but I'm not going to get something that has these controllers built in.
?
Chris, G5CTH
? |
|
On Sun, Oct 20, 2024 at 10:25 AM, Chris wrote:
I'm sufficiently unhappy about the power controllers on the QMXInteresting.? I am very happy with the QMX power controllers.? This nicely thought-out design allows QRP operation for many hours from a tiny battery because of their excellent efficiency.? And the way this is accomplished while avoiding switching noise in the RF is excellent. ?
And I have never had a problem with one failing.? Nor have most of the many thousands of QMX/QMX+ users.? And I currently use a USB-C PD power-bank for my power source, with a simple cable that switches it to 9V for my QMX.
?
Yes, the design implementation is a bit fragile to unexpected external power fluctuations.? I have read the many posts from the relatively few who have had failures, and a very high percentage of the failures are the D109 zener acting as a fuse to protect the rest of the board from an external over-voltage condition - that is why D109 is in the design.? Perhaps D109 should have been sized at a higher max current, so it wouldn't fail under the most common over-voltage conditions which have been reported here - but that's an easy mod, just put in a bigger zener, and keep the benefits of the switched power design.? The larger zener would clamp the voltage and sink the current to avoid damage until the (hopefully momentary) overvoltage condition resolved itself.?
?
The other common PS failures are from mishaps during kit assembly and test - these are avoidable, and readily fixable.? The most common of these are solder bridges to the connectors, and accidental breakage of the tiny coil wires during handling.? The newer rev 2 QMX+ design with the PS integrated on the main board should eliminate many of those construction issues - but yes, when there is a failure it may be harder to fix.? But a bigger D109 would eliminate most of those failures.
?
So my vote (or perhaps my mod) would be to install a higher-current D109.? Going back to inefficient linear regulators to avoid the problems is (in my opinion) a poor choice.
?
Stan KC7XE |
|
Hi Chris, Stan I'm interested in why Chris objects to the SMPS design so much? Stan, FYI you can't just change to a higher wattage 3.6V zener. I did try that in the second batch of QMX, back over a year ago. It failed and in the end we manually replaced all the 3.6V 5W zeners with 3.6V 500mW zeners.? The problem is that low voltage zener diodes, below 5.5V or so, aren't the same as higher voltage ones. Above 5.5V the mechanism responsible is "avalanche breakdown" and it's quite sudden, by which I mean the current vs voltage curve gets suddenly very steep. So in the case of the 5.6V zener on the 5W SMPS board, whether you put in a 500mW diode or a 5W diode, the IV curve is practically idencial, the only difference is that the diode will take a lot more abuse before it fails. Changing it from a 500mW to a 5W device was harmless.? Diodes below 5.5V rating use the actual real "zener breakdown" effect. It's a much gentler shallower curve, unfortunately. When a diode is specified as "3.6V zener", read the datasheet, this voltage is specified for a given current flow, a certain point on the curve. The specificarion current flow is very different for higher wattage devices than low wattage ones.? Dumb me, I didn't know any of this, and so I happily ordered 3.6V 5W zener diodes thinking that it's an easy substitution like using a 10V 470uF tantalum instead of a 6.3V one. Then found that even at the correct 3.3V operation voltage of the SMPS a substantial current flowed through the diode, making it very hot and raising the current consumption of the whole QMX by something like 30mA. It took me a while to figure out what was going on, and a lot of experiments, research and reading. 3.6V 500mW zener means the voltage is 3.6V at some much larger current flow (I forget exactly); and even at 3.3V there's considerable current flow.? I attached one of my measurement plots from those days. Notice the 500mW and 5W 5.6V zener IV curves rotallt overlap over on the right hand side of the chart. But the 3.6V curves are all over the place... Because "3.6V" only has a specific meaning at a specified current; it's a point on a very gently sloped IV curve.? Conclusion is that zener diodes with breakdown voltages under 5.5V are evil. Any possible hope I have of avoiding them for the rest of my life, I'll take... Every effort to design out low voltage zeners from any projects! So please don't be tempted to change out the 3.6V 500mW zener for a higher watt rated one... It's a good fuse, keep it that way ;-) 73 Hans G0UPL On Sun, Oct 20, 2024, 21:18 Stan Dye via <standye=[email protected]> wrote:
|
|
Thank you, Hans, for that chart and for the education on the low-voltage Zeners.? Very different from what I expected in my naive thoughts.
So we will keep the nicely-fusing D109.??? Which doesn't fuse for most of us, because we take care of the external voltage we supply.
And thanks again for this amazing QMX design - it is such a fun rig to use and show off!
Stan KC7XE |
|
It doesn't fuse for me, even though my QMXs go through a lot of abuse. Including supply rail abuse. Yesterday I set up about a 50m or so long wire, connected to my QMX via an LC match. I forgot to connect anything at all to the ground terminal of the LC match. So the ground was flowing through QMX from right to left, then back out to the power supply and eventually the house wiring ground. Yuck! Neither the QMX nor the power supply was happy. The LCD baxighr was flickering and the power supply voltage and current jumping all over the place. But I realized my error and connected a wire from the antenna ground terminal to power supply ground, and everything was happy again. It collected around 3,300 spots on 80m WSPR overnight.? 73 Hans G0UPL On Sun, Oct 20, 2024, 22:03 Stan Dye via <standye=[email protected]> wrote:
|
|
Thanks Jerry, interesting and informative - I was using the latest 027 firmware, by the way.?
The event happened when I nudged the rig and the power cable. While that seems to have been the issue, I also had speaker ?and key cables plugged in - so I can’t rule out that the little board for these connectors might have shorted against the case (the clearance there was less than it should be due to a small gap between it and the main board). Yes - builder error - while soldering the board I allowed some solder to flow between the boards causing a small gap.?
?
All that said - if it’s just one very inexpensive zener which is easy to replace…
?
Steve |
|
I have had my issues with high SWR, antenna flubs, and just stupid mistakes and my QMX and QMX+ have never failed me. They have been used into my beam, qtr. wave vertical and EFHW with well over one hundred cw contacts. The QMX has worked FT8 ?for several hours with no problems. Both run on 12v Talentcell packs. They are rugged little rigs. I try to always check my SWR before transmitting. Mike Krieger On Sun, Oct 20, 2024 at 3:44?PM Hans Summers via <hans.summers=[email protected]> wrote:
|
|
Steve,
?
I doubt the speaker or key had much to do with this
unless the connectors somehow shorted out the 3.3v or 12v rails.
?
I doubt this happens often enough to recommend that everyone
hack their rig to use a different voltage regulation scheme.
Errors in doing such a hack would also bring on failures.
?
But I think a change should be considered if the board is ever updated.
?
Jerry, KE7ER
?
?
On Sun, Oct 20, 2024 at 12:55 PM, Steve MW0KST wrote:
|
|
Hi Hans, Just a point about the Zener curves you plotted. The 3.6V 1/2W Zener (D109) does indeed have a very shallow overvoltage curve. On the damaged SMPS boards Jeff sent me, I measured the voltage across these Zeners at currents up to 100 mA and in all cases, the Zener voltage was in the range of 4.0 to 4.2V. The diodes held up fine at 100 ma although they got very hot. I didn’t go higher than 100 ma to avoid damaging the good boards in the batch he sent. However, from your plotted curves, it’s clear that the Zener voltage can go considerably higher than 4.2V until it eventually fails if the SMPS isn’t clamped by firmware under overvoltage conditions. For the sake of the chips operating from the 3.3V rail, let’s hope the Zener failure mode is a short and not a fuse. Under overvoltage conditions, the Zener fusing would allow the 3.3V rail to shoot far above 4.2V and potentially damage chips on that rail. I have one SMPS board here where the Zener shorted but I haven’t checked the rest of them yet. The more important point is that the 1/2W zener is only marginally protective for chips that can’t tolerate their 3.3V rails being at 4V or above. I don’t know how much current the 3.3V SMPS can provide under overvoltage conditions, but I assume you’re limiting the 3.3V rail’s excursion via firmware. Tony On Sun, Oct 20, 2024 at 1:45?PM Hans Summers via <hans.summers=[email protected]> wrote:
|
|
In my experience it’s very difficult to fully protect low voltage rails against overvoltage without an active sensing circuit that (depending on the nature of the risk ) interrupts the supply or crowbars the rail. ??
TVS diodes often fail after the most sensitive connected devices, ? Zeners can be a bit laid-back, leaking early ?or letting too high a voltage pass for too long. ?
Their main benefit is their tendency to fail short which at least leaves a helpful hint as to the problem? ?
There’s definitely a market for a small fast and cheap 3 or 4 terminal protector for the common supply rail voltages and hopefully they arrive soon? |
|
I agree Bruce. I’m a bit confused as to how the 3.3V SMPS can go so far overvoltage when it’s being monitored by software via the voltage divider. I’d think that if the divided voltage exceeded, say, 1.8V, software would see this and turn off the SMPS control. The Zener would be considered a backup if all else failed. There are power rail monitoring chips that disable the rails if they go out of a specified range. Tony On Mon, Oct 21, 2024 at 11:43?AM Bruce Akhurst via <bruce=[email protected]> wrote:
|
|
Hi Tony Indeed the zeners are only a backup. Firmware takes care of all normal operation and quite a bit of abnormal operation too. If all else fails the zener can be a final backup.? 73 Hans G0UPL On Mon, Oct 21, 2024, 20:09 Tony Scaminaci via <tonyscam=[email protected]> wrote:
|
|
Tony says:
"I’m a bit confused as to how the 3.3V SMPS can go so far overvoltage when it’s being monitored by software via the voltage divider. "
?
I believe the issue is the size of the coil (330uH) of this buck mode switcher.
It's relatively large for a switcher because the STM32F doesn't check 3.3v very often.
An inductor has a sort of electrical inertia, when you put a voltage across it the current
through the coil slowly ramps up, when you remove the voltage the current slowly ramps down,
so it takes awhile for the current out to the 3.3v rail to fall to zero after the STM32F sees?
a too high 3.3v rail.? ?Perhaps the algorithm should also be checking the 12v supply,?
and anticipate trouble when there's a jump.
?
Bruce says:
"There’s definitely a market for a small fast and cheap 3 or 4 terminal protector for the common supply rail voltages"
?
I'm not familiar with SCR specs, perhaps someone more familiar with them
can say if this is appropriate for 3.3v rail protection.
?
Mouser? ?511-TN805-600B-TR, under $0.50
Can sink 70A for 10ms.
?
It can also deal with 600V across it, I assume it conducts fine at 3.3v.
?
Gate trigger voltage Vt is spec'd at 1.3v max, threshold Vto at 0.85v max.
So should be able to trigger it with an PNP transistor, emitter tied to 3.3v,
collector to the SCR trigger (through a resistor), a resistive divider driving the base.
Triggers the SCR when the base is more than 0.6v below the 3.3v rail.
?
If Vt is tight enough, might get by with just the SCR plus the resistive divider
driving the trigger directly.
?
Either way, that could be a very simple board that's added to the QMX
with wires to 3.3v and GND.?
?
Jerry,? KE7ER |
|
From my reading of the posts about these D109 failures, they happen for two primary reasons:
1) assembly issues that result in shorting of the 12V or 5V rail to the 3.3V rail
2) rapid large step voltage increase on the external supply, too fast for the controller to react to (Jerry, your explanation is interesting and may be spot on)
But I don't think either of these merits a design change, because they can both be readily avoided.
For assembly issues, I would rather have a 'fused' response that makes me find the problem, rather than having an intermittent internal short that is clamped away so I never see and fix it.
Perhaps the case of USB-C PD switching could be operationally problematic with some implementations of the PD interface cable - but my power source is a small PD power-bank, and I don't have a problem. ? I have checked for any switching transients, and don't see them.? I use a simple inexpensive 9V USB-C to power cable, and it puts out exactly 9V when I plug it in. (I built my QMX for 9V).? So at least one case-in-point that PD can be used safely.
So I don't see an operational use case for a design change here.? Do you?
Stan |
|
Hi Jerry ?
It can't. Unless there's something wrong, some wiring or soldering fault etc.?? ?
The algorithm IS doing exactly that?already.? Whilst the size of the inductor limits the speed at which something can be done about a positive excursion, it also limits the speed at which such an excursion can occur, making it possible to fix it before damage can occur. Inertia slows down everything. The 470uF capacitor helps with this too.? 73 Hans G0UPL |
|
On Mon, Oct 21, 2024 at 12:43 PM, Bruce Akhurst wrote:
There’s definitely a market for a small fast and cheap 3 or 4 terminal protector for the common supply rail voltages These are $1.56 single quantity, under a dollar at 25 pieces (DigiKey). ?
Overcurrent (2%) , over/undervoltage (3%), thermal protection, Vo dV/dT limiting, etc.? They even have a design calculator tool spreadsheet, just fill in the blanks with your specs. ?
I used them on a MIL SATCOM portable earth terminal power board I designed, among others.? There are many like it, higher voltage, higher current, lower RDSon, etc.
The internal FET makes it a small, bulletproof solution.
73, Don N2VGU
|
|
Hans,
?
So why is it that a step from 5v to 12v on Vin can blow D109?
?
I'm guessing you have a PWM rate up around 1 MHz,?
but only inspect Vout in firmware perhaps once every millisecond.
That works for constant Vin and constant load current, but not otherwise.
For example, a sudden 2x load on Vdd (and especially the sudden unload).
?
It really should be more robust.
I think long term that means using a standard SMPS IC,
but you have plenty of other things to do first.
?
?
Short term solutions include:
?
A small board with a couple TPS54160/260/360? chips sync'd to a 2.5mhz oscillator
?
Linear regulators.
?
Something like that SCR clamp I suggested recently:
?
?
?
I haven't had time to build my QMX+ kit yet, need to do that first.
?
At work using standard SMPS IC's for a 5v rail, we shipped thousands
of boards to customers who plugged into whatever supply they had from 5 to 30vdc.
Never saw SMPS failures except some guy who swapped his power leads,
but you have that issue covered with the pFET's at Q103,Q105.
?
Jerry, KE7ER
?
?
?
On Mon, Oct 21, 2024 at 12:30 PM, Hans Summers wrote:
|
|
Jerry, I *think* the switcher frequency is 116.7 KHz from what I’ve seen on the terminal screen shots. Maybe I’m wrong on this, if so, what does the 116.7 KHz reading mean? Tony On Mon, Oct 21, 2024 at 3:42?PM Jerry Gaffke via <jgaffke=[email protected]> wrote:
|
|
Tony, yes that is the PWM frequency...JZ On Mon, Oct 21, 2024, 5:18?PM Tony Scaminaci via <tonyscam=[email protected]> wrote:
|
Bruce Akhurst
Donald S Brant Jr
to navigate to use esc to dismiss