F405 & F411 versions of the BluePill

What are you developing?
User avatar
RogerClark
Posts: 7447
Joined: Mon Apr 27, 2015 10:36 am
Location: Melbourne, Australia
Contact:

Re: F405 & F411 versions of the BluePill

Post by RogerClark » Mon Oct 30, 2017 12:32 am

Thanks

racemaniac
Posts: 618
Joined: Sat Nov 07, 2015 9:09 am

Re: F405 & F411 versions of the BluePill

Post by racemaniac » Mon Oct 30, 2017 7:06 am

getting the components on one side isn't that hard indeed, but laying out the traces won't be easy ^^'
thez programming header/crystals/analog rail side was ok to do (btw, that's one difference between your schematic and mine: on one of the 3v3/gnd parts on the pins, i break out the analog 3v3/gnd instead of the regular one, so they are also available on the headers.)
although i have some parts there i still really need to improve.
the other side with boot selection/button/big voltage regulator/... isn't working out for me so far ^^'

User avatar
Squonk42
Posts: 184
Joined: Thu Dec 29, 2016 9:25 am
Location: Bordeaux, France

Re: F405 & F411 versions of the BluePill

Post by Squonk42 » Mon Oct 30, 2017 7:29 am

For BOOT selection, don't use header+jumpers: BOOT1 is useless anyway, and replacing BOOT0 jumper with a tactile switch like the one used for RESET will be much smaller, with the added benefit of providing a USER button if you put a 0R to PB8 in parallel to BOOT0. And using the tiny ALPS SKSGACE010 buttons will save you a lot of space, they are only 3.5 mm x 2.7 mm x 1.4 mm!

Plus, you don't have to put the buttons on the USB side like the original, you can move them on the SWD side if necessary (see my part placement above).

Breaking out the analog VCC/GND on header is a good thing, but it breaks the compatibility with existing Bluepill boards, @RogerClark: what do you think?

In my schematics, I just dropped the LDO providing the analog power supply: this is a luxury IMHO. Better have a good ferrite bead between +3V3 and +3V3A only to filter high-frequency noise, and have (as much as possible on such a small board) separate GND and GNDA only connected at a single point, which is the normal way to handle analog/digital coexistence. Adding a ferrite bead between GNDs is not necessarily a good thing for small signals, as it will create a voltage offset between GNDs that depend on ground currents, better put a trace only.

User avatar
RogerClark
Posts: 7447
Joined: Mon Apr 27, 2015 10:36 am
Location: Melbourne, Australia
Contact:

Re: F405 & F411 versions of the BluePill

Post by RogerClark » Mon Oct 30, 2017 7:37 am

I don't think you can use one of the existing pins for the analog Vcc, as the whole point was to make a BP compatible board

But you could add another pin (through hole) somewhere else, or just a pad on the underside etc

racemaniac
Posts: 618
Joined: Sat Nov 07, 2015 9:09 am

Re: F405 & F411 versions of the BluePill

Post by racemaniac » Mon Oct 30, 2017 7:39 am

i'll keep the regular 3v3 on both sides of the header then.

*edit*
after thinking about it a bit more, aren't both options incompatible?
If i don't put the analog 3v3 on the standard header, anyone feeding their bluepill via the standard header and doing something analog is out of luck, since the analog part won't be powered at all.
at least for my solution, if you provide power via both 3v3 pins, the entire board is powered. now i don't have any bluepill shields/projects that rely on this, so not sure what people who expect pin compatibility are expecting, but i think my solution also has some merits.

the only other option would be a jumper to connect 3v3a with 3v3 (or a small solder bridge you can make, a jumper would be a bit big), and then just keep the regular 3v3 on both pins.

User avatar
RogerClark
Posts: 7447
Joined: Mon Apr 27, 2015 10:36 am
Location: Melbourne, Australia
Contact:

Re: F405 & F411 versions of the BluePill

Post by RogerClark » Mon Oct 30, 2017 9:59 am

@racemaniac

Yes. I guess it would be possible to use one of the Vcc's as analog Vcc.

Are you using a really small regulator for analog Vcc, or is it roughly equivalent to the cheap regulator on the Blue Pill ?

What advantage is there to having the analog vcc available for external circuits?
Lower noise than the main regulated output ?

I suspect people may use the Vcc pins to power a lot of things and existing designs may link them in their PCB, or assume they are connected together internally.

racemaniac
Posts: 618
Joined: Sat Nov 07, 2015 9:09 am

Re: F405 & F411 versions of the BluePill

Post by racemaniac » Mon Oct 30, 2017 10:45 am

RogerClark wrote:
Mon Oct 30, 2017 9:59 am
@racemaniac

Yes. I guess it would be possible to use one of the Vcc's as analog Vcc.

Are you using a really small regulator for analog Vcc, or is it roughly equivalent to the cheap regulator on the Blue Pill ?

What advantage is there to having the analog vcc available for external circuits?
Lower noise than the main regulated output ?

I suspect people may use the Vcc pins to power a lot of things and existing designs may link them in their PCB, or assume they are connected together internally.
for analog i'm using the small ap2112 (the one i originally used for the non analog part too, but that was found to be too light).

and there is no very specific reason to have the analog 3v3 also available on the pin headers (other than it at least being reachable/connectable if you're not using a 5v supply).

it seems we've found another design decision to make:
- keep the original pins (both just 3v3, the analog supply then maybe on other pins) -> not quite a fan of this honestly. taking up quite some extra space for this...
- my solution of having the 3v3a & gnda on the pins next to the programming header -> probably also not ideal, depends on how the original bluepills are used (if they power via both 3v3 pins, it's compatible (as both grounds are tied together via a ferrite bead), otherwise they'd have to make sure to power the other 3v3 pin too)
- keeping the original pins, but offering a jumper/solder bridge that can be made to tie the 3v3 to 3v3a -> very compatible, but it's less elegant if you want to provide your own analog 3v3 rail (you can connect your 3v3a to the appropriate pin of the jumper/solder bridge, but the analog ground isn't really broken out in this case...)
- probably some other options i'm not seeing yet ^^'

each choice has its advantages & disadvantages, kepping 100% compatibility is impossible if we now have a separate analog 3v3, so we'll have to pick our compromise :).

User avatar
Squonk42
Posts: 184
Joined: Thu Dec 29, 2016 9:25 am
Location: Bordeaux, France

Re: F405 & F411 versions of the BluePill

Post by Squonk42 » Mon Oct 30, 2017 5:00 pm

As I said previously, there is absolutely no need for an extra AP2112 LDO for VCCA: to get good small-signal ADC performance, we only need a correct filtering through a ferrite bead, and a clean AGND connected to GND at a single common point without ferrite bead to avoid ground offset depending on ground currents, like all dev boards I know of, and like what ST recommends in its ANs and implements in its own dev boards.

From AN2586 "Getting started with STM32F10xxx hardware development":

Code: Select all

Power supply schemes
The circuit is powered by a stabilized power supply, VDD.
● Caution:
– If the ADC is used, the VDD range is limited to 2.4 V to 3.6 V
– If the ADC is not used, the VDD range is 2.0 V to 3.6 V
● The VDD pins must be connected to VDD with external decoupling capacitors (one
100 nF Ceramic capacitor for each VDD pin + one Tantalum or Ceramic capacitor (min.
4.7 µF typ.10 µF).
● The VBAT pin can be connected to the external battery (1.8 V < VBAT < 3.6 V). If no
external battery is used, it is recommended to connect this pin to VDD with a 100 nF
external ceramic decoupling capacitor.
● The VDDA pin must be connected to two external decoupling capacitors (100 nF
Ceramic + 1 µF Tantalum or Ceramic).
● The VREF+ pin can be connected to the VDDA external power supply. If a separate,
external reference voltage is applied on VREF+, a 100 nF and a 1 µF capacitors must be
connected on this pin. In all cases, VREF+ must be kept between 2.4 V and VDDA.
● Additional precautions can be taken to filter analog noise:
– VDDA can be connected to VDD through a ferrite bead.
– The VREF+ pin can be connected to VDDA through a resistor (typ. 47 Ω).
And correct me if I am wrong, but the original BluePill should be powered from +5V, either coming from the USB connector or from the headers, and the RT9193-33 LDO output is providing the +3.3V, so all +3.3V pins on headers and SWD should be considered as outputs only... There is no guarantee that this LDO will accept to be fed with current on its output without a reverse protection. Looks like it is working as the board can be powered from +3.3V SWD only, but this is just pure coincidence :)

With our new design, we provide the ability to power the board from +5V like before, and down to +3.7V (LiPo battery) because of lower drop-out voltage AP2114AH LDO, but the +3.3V pins should still be considered as outputs only.

So the question to provide VDDA and GNDA instead of VDD and GND on some existing header pins is irrelevant: these pins will provide a filtered analog supply instead of a noisy digital supply from the original BluePill, and in the case the user connects noisy digital circuits to these pins, it will bring us to the same (bad) ADC performance as the original BluePill => Not worse for existing designs, much better for new designs knowing this improvement.

racemaniac
Posts: 618
Joined: Sat Nov 07, 2015 9:09 am

Re: F405 & F411 versions of the BluePill

Post by racemaniac » Mon Oct 30, 2017 5:32 pm

fair enough, and since we're using an LDO that can handle voltages well below 5V, always using the "5V" input is a nice option with this board :).

User avatar
RogerClark
Posts: 7447
Joined: Mon Apr 27, 2015 10:36 am
Location: Melbourne, Australia
Contact:

Re: F405 & F411 versions of the BluePill

Post by RogerClark » Mon Oct 30, 2017 8:05 pm

I will need to doible check one of my Blue Pill boards, but last time I tried to use the Vcc to power an external analog input amplifier, I found the Vcc very noisy.

Well beyond what a small ferrite choke could clean up.

The amplifier was for a HB100 doppler radar module, and I have since found a better design for the amplifier which uses its own regulator to minimise the noise.

IMHO, I think before the analog Vcc regulator was removed, we would need to make sure its definitely not needed, by doing some practical validation on real hardware.

Post Reply