Page 1 of 1

Choosing between powering options.

Posted: Sun May 03, 2020 9:06 pm
by kesterlester
I am building a pyboard-D based device which will have ample access to mains-derived power which will almost certainly have to be delivered either as:
  • 5V to VUSB, or
  • 3.3V to VIN, or
  • something else if none of those above are thought to be a good idea for my application in the long term.
In both cases that power will come from a small embedded switched mode power supply something like this: using ether a 3.3V version or a 5.0V version.

I am trying to work out the pros-and-cons for long-term-reliability of those two main powering options given that my device will mostly be in a (~40deg C?) environment for ~10 years.

Among other statements, the instructions on powering the D-series:
You can power the PYBD via the VUSB port (see X-positions). The voltage on this port must be between 4.8V and 5.2V. This port contains a protective diode and a 1A fast/0.22Ohm fuse.
You can power the PYBD via the VIN port (see Y-positions). The voltage on this port must be between 3.2V and 4.8V. ... For best performance provide 3.4V on VIN (3.3V is also ok but 3.4V will benefit from LDO filtering).
3.3V on VIN pros and cons

So far as I can see, the 3.3V option has the benefit that it will not waste power as heat, which could lead to my board not running unnecessarily warm in what will already be a warm environment. However I am alarmed that 3.3V sits at exactly the level at which we are told LDO filtering will no longer happen, and only 0.1V away from the recommended absolute minimum lower power voltage limit of 3.2V. Who knows: with a small amount of derating it would not be impossible to imagine that the embedded PS could give a little less than its rated 3.3V at times ... How sensitive is that 3.2V lower limit? If I am right up against it, and in a somewhat noisy environment (coupling from mains) will I be almost ensuring that my pyboard has unpredictable glitches once per month and works unreliably? Does anyone have any experience with that?

5.0V on VUSB pros and cons
This has almost the reverse pros and cons of the other option. Brown outs from marginal undervoltage are presumably very unlikely, but correspondingly I assume that the pyboard regulator will get hot over time -- though perhaps not hot enough to cause me problems??

What sort of temperature difference do people see for 5V vs 3.3V powering schemes? Anything noticeable or basically imperceptible?

Other options
If (as the notes say) "3.4V is best" then I will have to do more design work as I can't get any off the shelf power supply that would suit me at voltages between 3.3 and 5V. And I'm not very good at desiging power supplies, so I don't want to go there unless I really have to.

Re: Choosing between powering options.

Posted: Tue May 05, 2020 2:33 pm
by chuckbook
Power requirements are defined by these components:
  • Wireless Module (Murata 1DX)
    The Wireless Module uses an internal linear regulator for the wireless transmitter. See Murata datasheets for the specs.
  • ADC Reference
    ADC performs best with a stable +3.3V reference voltage which is provided with a proper input voltage to the LDO. (We observed dropout voltages less than 150mV @ 300mA for temperatures below 85°C)
  • USB Operation
    USB Operation will stop with Vdd (LDO output) below 3.0V.
  • MCU
    MCU works fine with Vdd (LDO output) above 2.7V. Lower voltages may also be used with reduced SYSCLK and internal flash write modes.
  • A mains AC/DC converter should be carefully selected as some of them don't provide well regulated output voltages.
  • Applications with high average processor loads and/or high Wifi throughput benefit a lot from carefully selected input voltages whereas light load apps with reduced SYSCLK and occasional wireless transfers obviously don't.

Re: Choosing between powering options.

Posted: Tue May 05, 2020 2:45 pm
by pythoncoder
I ran a Pyboard D for 8 weeks from a good quality USB wall-wart without issue. It ran code using WiFi/MQTT. Rock solid stable, unlike some Espressif devices.

The linear regulator consumes a little power but does perform a useful role in rejecting noise and brown-outs. The WBUS-DIP28 offers other powering options.

Re: Choosing between powering options.

Posted: Tue May 05, 2020 4:41 pm
by tve
If I were you I would probably feed 5V but choose a slightly more powerful power supply. At 5V the PYBD overall will dissipate 1.5x more power than at 3.3v, so yes, it's more, but not a ton. I wouldn't be very concerned about the regulator chips on the PYDB, it looks to me like they have low thermal resistance to the PCB. The temperature of the PCB is probably of greater concern given how small it is, depending on how you mount it you may be able to conduct heat away to the PCB it's mounted on. But first you need to assess the overall power consumption, it's very different if you use it to stream video footage to wifi vs. send some sensor value every minute.

WRT the Recom power supply, the 3.3v version is rated at 500mA and the 5v version at 400mA, that could be plenty or be tight depending on the use you're making. It may be smart to upgrade to the 3W version...

Re: Choosing between powering options.

Posted: Wed May 06, 2020 9:16 pm
by kesterlester
Thank you, all, for the various pieces of advice.

Fortunately the board is not going to be streaming anything via wifi. It will be controlled/configured by wifi, but only very occasionally (e.g. to receive updates to parameters, or to file some hourly reports on sensor vales, etc). I expect I can therefore configure the thing to only turn the wifi on every so often.

Most of the time the board will just sit in a control loop monitoring some io pins at no more than 1Hz and and toggling others in response to condition changes. The board will be vastly over-spec for its task, but as I have been using other early-series pyboards for other tasks for a while, I fancied putting a D in this one as an exercise in getting to know it.

In short, therefore, it's probably on average a low power application. Still, I can imagine upgrading what it does later on to use some ADCs to monitor some secondary analog sensors. Given that, and given the feeling I get from the responses that the D is unlikely to get really on run-of-the-mill tasks (I don't have the patience to spend ages learning how to make it super low power, though I am aware that experts could easily make it so!) I will probably therefore go for the higher-margin 5V supply.

Which means this forum has probably been useful. Had no-one replied I would probably have gone for the 3.3V option. Thank you.