> I also don't know how happy pin PD7 was about being?
> connected to 12V or 9V and then being initialized.
One can debate that, in conjunction with a detailed study of the datasheet, an exercise which to me at least, is ambiguous and inconclusive. Or one can calculate the worst case current through that 100K resistor. Or one can observe the evidential reality which does indicate that the port pin PD7 (and indeed PD4) is not particularly unhappy about the experience.?
Chris:
> In most places the transition between these is managed?
> using additional FETs such as Q101 for LIN_REG_EN?
> and the linear supply, Q106 for the PWM_5V signal and?
> Q107 for the PWM_3V3 signal but in the case of Q111?
> and Q110 this doesn't happen, they can both be on at?
> the same time.
No, you have misunderstood the circuit. There is no scenario where the linear supply and SMPS can both be on at the same time.?
1) PD4 is not yet initialized "high", LIN_REG_EN is pulled towards Vin by 100K resistor R101. Then
a) Q101 is on, pulling the?gate of Q102 to ground, Q102 is ON and the 78M33 is powered.?
b) Q110 is on, putting the 47-ohm resistor R114 "dummy load" across the 3.3V buck converter output.
c) Q111 is off since its gate is higher than its source.?
2) PD4 is initialized "high", LIN_REG_EN is now set to 3.3V and the pin sinks some uA from R101.?
a) Q101 is on, pulling the?gate of Q102 to ground, Q102 is ON and the 78M33 is powered.?
b) Q110 is on, putting the 47-ohm resistor R114 "dummy load" across the 3.3V buck converter output.
c) Q111 is off since its gate is at 3.3V, which is higher than its source - the source is at 0V initially, rising to 3.3V as the SMPS is brought up.?
3) PD4 is set "low" when the processor thinks the SMPS is ready; LIN_REG_EN is now 0V.
a) Q101 is off; the gate of Q102 is pulled to Vin by 10K R102, so Q102 is OFF and the 78M33 is off.?
b) Q110 is off, the dummy load is therefore disconnected.?
c) Q111 is on, since its source is at 3.3V and its gate is at 0V, comfortably exceeding its Vgs.?
At no point can the SMPS and linear regulator both be active at the same time. Diode D103 prevents 3.3V trying to go back into the unpowered 78M33.?
Tony:
> The 1N4148 diode in series with the linear 3.3V output is a?
> necessary evil to prevent backward current flow once the?
> 3.3V SMPS rail comes up. However, it destabilizes the linear?
> 3.3V rail as noted by the variation in voltage as the?
> processor is executing its boot code.?
"Destabilizes" is probably a little strong. It doesn't destabilize anything. It permits some variation in the 3.3V rail, which is harmless and a small band of operating voltage in the context of the processor working fine right down to something like 1.7V. A diode doesn't have a 0.6V drop, it has a drop which depends on the current through it; in this case I believe as the processor current varies as it boots up, the drop is varying and that's why we see some small variation in Vdd during this time. But it's a narrow band and we are far from the minimum operating voltage of the processor.?
> It is advisable to have a substantial capacitance on the?
> processor’s power pins to minimize voltage. This should be?
> at least 47uF - 100uF to ensure that the processor doesn’t
> experience an instantaneous dip below its minimum spec?
> voltage which could cause the boot process to hang or?
> go off into never-never land.
When I make my one-off homebrew projects I often chuckle to myself, "you can never have too many capacitors or too much shielding". However - when you want to make a commercial product which has a very high performance to price ratio - you then consider how much shielding you really need, and how many capacitors you really need.?
I believe the capacitors in the QMX supply (2.2uF and 0.1uF) are consistent with the recommendations of the STM32F446 datasheet and the 78M33 datasheet. No higher capacitance is needed for stability or reliable operation. One could, if one was a follower of Muntz, probably take lots of capacitors out of QMX and it still work reliably and without performance deterioration.?
> What happens on one unit is not representative of 1000’s?
> of units so I hope the processor voltage during bootup is?
> being tested on every unit prior to shipment.?
No! There is absolutely NO need! Everything is fine. The design is not anywhere near marginal.?
Paul:
> The thing I found difficult was that both PS boards must be present?
> (and working) to bring up the microprocessor. (The 3.3 volt supply?
> is dependent on +12 from PCB1...) It would have been nice to be?
> able to bring up the 3.3 volt board, then the 5 volt.
There's really three things going on.?
- Reverse polarity protection and soft power switch
- 5V SMPS
- 3.3V SMPS (and initial 3.3V linear regulator)
I chose to put them on two boards, just because it let me reduce my overall PCB size. 6-layer PCBs are expensive so doing this was a cost-saving, because otherwise when you snapped out a larger SMPS PCB with all the power stuff on it, it would waste a lot of unused space on the PCB panel. Right now all we waste is a little square with the QRP Labs logo printed on it, which I hope you all have on your keyrings!
So the way it worked out, there was best use of space having the soft power switch on the 5V board. The 5V supply doesn't have to work properly, for the 3.3V supply to work and the processor to boot up and allow you access to the diagnostics screen.
> In software, we used to call a unique, elegant, piece of code?
> a "Neat Hack", a sort of backhanded compliment to the coder.
> Hans, this unit is a Neat Hack in the realm of electronic (and?
> software) engineering. Lots of promise here-- You Go Guy! ;-)
Thanks :-)?
> Clearance between Q103/Q104, leading to solder bridges?
> and other challenges as people reworked the board. (I know?
> that was the root cause of?my?problems, anyway!)
Yes. Production problem. Easy to solve in future PCB revisions. And easy to fix now we know about it. I did another 50 boards last night, very quick and easy.?
> The impact of rapid changes (e.g., voltage spikes) in input
> voltage as power comes up on V_IN. Maybe the best fix?
> for this would be to wait until the input supply stabilizes for?
> a bit, before engaging the switching supplies?
I don't think there's a problem during power up. There has been a problem for a couple of people when there was a sudden large increase in voltage during normal operation. This is really the ONLY problem with the concept. But even this can be substantially mitigated by firmware, and is really quite unusual operating scenarios.?
Kees:
> That tells me that if you power ON again about 1-2 seconds?
> after powering OFF, the 5V VCC rail is still at about 1.7V.?
> Question is does the subsequent Power ON try to drive?
> VCC above 5V ?
No. When you power on, the firmware looks at the voltage it is seeing on the 5V ADC sensing input. If the voltage is above 5V it will not enable PWM. If the voltage is below 5V, but above zero, such as your 1.7V example, then the rail will just be gently brought up to 5V from there. No problem.?
Stephan:
> I don't think that "it's a single source part" is a strong?
> argument in the context of a product that contains an ARM?
> CPU, which I can personally attest to being frustratingly?
> unobtainable during the semiconductor shortage. I also?
> see an ADC from Texas Instruments and a DAC from?
> Cirrus Logic, also not generic parts.?
Yes, all that is correct... these are all non-generic parts, yes. And yes during the semiconductor shortage STM32 were hard to get (at no point, unobtainable, at least the things I needed). For me, the semiconductor shortage was a reason to TRY to avoid non-generic parts where possible. When finding STM32 and such was a fight, it seemed inadvisable to take on more unnecessary fights if possible. At the time when I did the design for the buck converter in the QDX PIN diode supply circuit, my first thought was certainly to look at SMPS ICs. However it was frustrating that so many were out of stock and unobtainable everywhere, and furthermore many are designed for higher current capability and don't work very efficiently at 50mA. So in the end I thought it would be fun and practical with discrete components!
> As far as BOM cost, if every dollar of BOM adds two dollars?
> to retail price (this is my formula for things I build and sell),?
> I'd have paid an extra $20 for a QMX with that extra $10?
> of BOM in the power supply... And spent several fewer?
> hours troubleshooting my dead radio. That time is worth?
> so much more than $20 to me.?
Yes, I understand; but your view is molded by your particular experience and hindsight in your specific case. Had you happened to have been in the fortunate majority who didn't suffer this type of failure then your view about increasing the kit price $20-$30 might be very different.?
To frivolously take it further - if someone made mistakes because the board is so compact, they might say they would have preferred a slightly larger radio and been happy to carry the extra grams and pay the increased materials costs on the PCB and enclosure. If someone had trouble with the connectors to the SMPS they could say they would prefer a larger board and it all on one board. If someone had trouble winding toroids, they could say they want a larger toroid in the kit and would be happy with the larger size and costs. Many people, if they had a problem, would have been happy paying more to have avoided it; many other people have other views about how the kit could have been a little different, to suit their own desires and circumstances better. In the end, it has to be acknowledged that you can never please all the people all the time, and have to try and strike a balance somewhere. Which is often very difficult indeed.?
> > Of all the QMX I have yet seen, other than the Q103/Q104 Drain?
> > short (manufacturing problem) I have yet to see a failure that is?
> > not attributable to shorts, damaged components or other >?
> > construction errors.
>
>?I do remember reading about at least one other person who?
> killed their QMX by toggling their power supply?
>?between 6 and 12 volts, in order to toggle their transmit?
> power. This is similar to my own scenario, where I?
>?forgot to raise the current limit on my power supply,?
Yes you are right, it is not only constructor?error. There have been a few cases such as this, that exposed a vulnerability. I think with my firmware changes I was able to very substantially mitigate that vulnerability but the risk is still there if extreme voltage fluctuations suddenly take place.?
> I don't think there is a single DC-powered device in my house
>? that would be harmed by subjecting it to voltage transients?
> within its specified input range. So I would have to disagree?
> that this phenomenon should be categorized as a?
> "constructor error."
Yes, I agree; these improbable scenarios are not constructor error.?
On the one hand, in my defence, there's always the question of how far one should go in order to protect the device. All the QRP Labs transceivers include reverse polarity protection. This means they are not harmed even by voltages OUTSIDE their specified input range. On one view, which has even been expressed in this group, you should be able to connect anything you like, to any port of a device, and it should not harm it. But to accomplish that would really increase circuit complexity and cost significantly. And one could argue that as radio hams, we are supposed to be somewhat technically competent, so we can perhaps be expected to take more care over not shorting the PTT output for example, or providing a reasonable SWR and antenna match. So, how far you go with protection is an interesting and difficult question.?
On the other hand, I am, and should be, always looking for ways to improve the designs... and we can see in other designs such as QDX, there were minor alterations like the addition of the AO3407 reverse polarity MOSFET, then when that turned out to be a little more delicate than its datasheet suggested, it was changed to AOD403. Etc. IF possible, it's always nice to try to make the transceivers more robust. Use of a SMPS IC would be one way. But maybe there are also other ways that can keep the current low cost, "neat hack", discretes version! Minor tweaks that could cost nothing or nearly nothing, but improve the robustness.?
Summary:
- All the 'scope screenshots in this thread, I consider normal.
- The 3.3V rail narrow band of voltage fluctuation during boot up is also, I consider, normal and far from any marginal operating scenario, there are no problems here.
- The capacitors choice and placement is I believe, compliant with the STM32 datasheet recommendations and 78M33 recommendations.
- The operation of the LIN_REG_EN signal (starting at 7.5V, then 3.3V, then Grounded) is also correct, and there are no scenarios where the linear and SMPS are switched on at the same time.?
73 Hans G0UPL
