Looks like my last post has material that's already covered in this thread.
Hans is now recommending 220k instead of 100k in parallel with the 22k at R112 and R113, for a resultant value of 20k.
And Hans is correct, you lose the rather nice power-on/power-off scheme.
?
I agree with Chris that the power supplies should be revisited, though understand that the board won't be changed anytime soon as Hans has bigger fish to fry.? That's fine with me, it works well enough for most of us.? I'm convinced that the long term goal should be for the QMX to use commercially available IC's in the SMPS's.
.
I was able to implement linear regulators by plugging a new daughterboard into JP105 of the QMX+.
In addition, I had to make a few mods to the main board, the mods I did could be reduced to:
Remove Q106 and Q107 (to disable the 3.3v and 5v PWM drive) and short across Q111 (so ADC_3V3
sees the new 3.3v regulator).? And of course, add 220k resistors across R112 and R113.
I also disabled the 78M33 regulator, though that is probably not necessary.
We could avoid all of those mods and just short from Q101's gate (LIN_REG_EN) to Q101's source (Ground) if the firmware were slightly adjusted:
In that post I state: "Shorting Q101's gate to source is still necessary, of course."
By which I meant to short those pins if the firmware is updated to accomodate my hack, that being the only hardware mod required.
But with currently available firmware you should not short Q101's gate to source, as that grounds the STM32F IO pin driving LIN_REG_EN.
?
This post also pertains to this issue:
?
Jerry, KE7ER
?
?
On Sun, Dec 22, 2024 at 11:42 AM, Chris wrote:
I produced my linear PSU board because I wanted a reliable and robust supply to provide the required voltages and survive mistakes such as a probe slipping.? I am prepared to live with higher current demand.
?
On Sun, Dec 22, 2024 at 07:33 AM, Hans Summers wrote:
Note:
- You need a 220K parallel to the top 22K resistor for Vdd supply also. Otherwise if it works, it's just working by chance (component tolerances). You need that to convince the firmware that it's SMPSs are OK and continue with the rest of the power up sequence.
Yes, I know, and said this in my notes.? Mine is just about OK.
?
- You lost reverse polarity protection?
Yes, I considered adding a diode in series or crowbar diode protection but decided I would be careful.? I routinely use polarised plugs for supplies so only need to get it right once.? People are fee to add a diode, a 1N4001 would do the trick.
- You lost the soft power on/off. You can still use long left button press to save the QMX operating state (frequency, volume etc) and shut down, it just won't really switch off the power.?
Yes, I know about the long button press and I can plug and unplug the power supply.
?
- I can't see a future problem now but I suppose it can't be guaranteed that future firmware versions would be compatible with it.
That's up to you, I don't think that you will change the firmware just to break this.
?
- Doesn't apply to QMX+ after Rev 1 because QMX+ has the SMPS integrated on board.?
I think you made a mistake doing that.? You have a power regulator circuit that's been a constant source of trouble, that attempts to protect voltage critical components using a crowbar Zener diode which regularly fails, destroying itself, and sometimes other components in the power supply. You rely on the zener failing short circuit and the voltage drop shutting the MCU down.? I think that relying on a failure mode to protect voltage critical components is folly.? Are the PCM1804 failures down to momentay voltage spikes?
?
When this was all on a cheap, easily replaceable board this was sort of OK, users could buy replacement boards and no more construction expertise was need that to build the kit in the first place.? Now when the power controllers fail the user has the option of replacing the entire main board or replacing SMD components - if the board survived.
?
The original use of this circuit was fine, to produce the TX bias, because it doesn't connect to any voltage critical components.? That's actually a current regulator isn't it?
?
And the justification for this - that the PWM frequency could be adjusted to avoid birdies - has never been implemented.? The PWM frequency is fixed and people are reporting birdies.
?
I must admit that I'm considering getting a QMX+ in spite of the board mounted power controllers.? If they fail I can work round it - put linears on the prototype board - but would really have preferred that the power controllers were still on a separate board.
?
Chris, G5CTH
