tiny stm32 F4 boards

Limited support for STM32F4 Discovery, Nucleo and custom F4 boards
palmerr
Posts: 58
Joined: Tue Jan 31, 2017 6:21 am
Location: Melbourne, Australia

Re: tiny stm32 F4 boards

Post by palmerr » Mon May 15, 2017 3:09 am

Ollie,

EXTI on ADC or EXTI generally?

@DanielEff and I have been talking about some neater ADCinit() code for his STM32GENERIC implementation under https://github.com/danieleff/STM32GENERIC/issues/6

To use the existing HAL (and LL) code requires a lot of intimate knowledge of the ADC and DMA registers. We're looking at functional (rather than register setting) code.

ADC EXTI triggering is one of the areas where special treatment will be required.

I agree that the HAL code tends to try to deal with every possible eventuality (like Linux drivers) rather than simply ignoring the flags that are irrelevant for a particular application.

I've had some success with Mixed HAL/LL code for the ADCs. Initialise with HAL (slower, but simpler), then attach interrupts to LL code for fast end of conversion handling, etc.

Ollie
Posts: 161
Joined: Thu Feb 25, 2016 7:27 pm

Re: tiny stm32 F4 boards

Post by Ollie » Mon May 15, 2017 3:49 pm

Richard,

My reference was for the minimal EXTI usage. My expectation was that it can be done completely in HAL, but I didn't find all the required functions. I had to use LL or direct register manipulation.

In the real life applications you are referring, in addition of ADC, EXTI, and DMA, you have often TIM and NVIC involved. The use cases have so many variants, that it is quite difficult to have Arduino IDE library to support most of them.

palmerr
Posts: 58
Joined: Tue Jan 31, 2017 6:21 am
Location: Melbourne, Australia

Re: tiny stm32 F4 boards

Post by palmerr » Tue May 16, 2017 1:10 am

I see your point - and thanks for the advice.

We'll just have to work on something general enough to be useful, but allow easy customisation.

The NVIC isn't so much of an issue, as the interrupts are pretty specific for a particular task, even though there are quite a few of them.

DMA is relatively straightforward, for most cases - the unpacking of values is a little messy for the multiple ADC modes, but is mainly done in user code.

TIMs are a little more difficult as there are 14 options and these are spread across both peripheral buses, and have a mix of 16 and 32 bit counters!

EXTI will be yet another adventure...

michael_l
Posts: 334
Joined: Mon Aug 24, 2015 6:11 pm

Re: tiny stm32 F4 boards

Post by michael_l » Sat May 20, 2017 1:50 pm

Received this small F407VET6 board. Looks like it is well manufactured. The size is pretty much 2x BluePill. I think this must be the smallest I have seen with this amount of GPIO's available. There's also EEPROM and SD card slot.

Interesting thing is that it has 25Mhz ceramic resonator.

Image

Image

ag123
Posts: 634
Joined: Thu Jul 21, 2016 4:24 pm

Re: tiny stm32 F4 boards

Post by ag123 » Sat May 20, 2017 3:42 pm

oh i got that recently as well, nice photo :D
i'm procrastinating on soldering that huge number of pins for the headers, the extreme of which some of the silkscreen labels would be covered by the headers. i'm thinking about 'inverting' the 4 debug pins so that all the headers would be 'below'

the sheer compactness is staggering, i placed it side by side with a maple mini (about the same size for blue pill i'd think), the length is shorter than blue pill/maple mini and the breadth is about double that of blue pill / maple mini, the stm32f4vet chip itself takes up the lion's share of the board

and for what's worth, it seem the board makers shipped with those 'legendary' 6pf Epson crystals, not the 'cheapo' ebay 3 cents one (but strange couldn't find the fitting part numbers)
http://www5.epsondevice.com/en/products/
http://www.stm32duino.com/viewtopic.php ... 582#p27582
http://www.stm32duino.com/viewtopic.php ... 409#p27407

off topic, japan still holds an old trophy for being a master at quartz crystals
https://www.watchreport.com/the-most-ac ... the-world/
http://www.livemint.com/Sundayapp/ipfLC ... world.html
but obviously, we probably can't expect the same as those 'grand seiko' luxury watches for these boards
for what is worth even touching the pins and adding some extra pf capacitance to the crystal would throw the oscillator seconds off the benchmark
http://www.stm32duino.com/viewtopic.php ... 409#p27426
:lol:

Ollie
Posts: 161
Joined: Thu Feb 25, 2016 7:27 pm

Re: tiny stm32 F4 boards

Post by Ollie » Sat May 20, 2017 6:04 pm

I have the same board. Indeed, there are a lot of pins, but I am using most of them in my application.

In this application the normal connectors would not work very well. It would be hard to remove the board without bending the pins. My solution was to use the machine pins which do require less force to insert and remove the boards. At the same time, the electrical connection with the round pins/holes is better that with the traditional square pins/holes.

ag123
Posts: 634
Joined: Thu Jul 21, 2016 4:24 pm

Re: tiny stm32 F4 boards

Post by ag123 » Sat May 20, 2017 7:32 pm

i'm doing things a little different, i tend to use wires with single pin connectors rather than those full row multi pin connectors (worse if those are double row connectors). of course, this result in a 'mess of wires' but it would be more flexible for me to re-define the usage rather than full row connectors, i'm thinking of going on to wire wraps as well similarly leaving a 'mess of wires' :lol:

Ollie
Posts: 161
Joined: Thu Feb 25, 2016 7:27 pm

Re: tiny stm32 F4 boards

Post by Ollie » Sat May 20, 2017 11:48 pm

@ag123,

Have you tried those small proto boards with through holes. There it is very easy and fast to solder sockets for multipin components and solder the two-leg components directly. Based on some recommendations from Uncle Google, I did use the wire wrap wires for soldered connections. The stripping and soldering was not a problem, but the insulation didn't tolerate the hot soldering iron.

Now I have returned back to enamel wires for all connections - thin wires for signals and thicker for power. The striping is a major challenge. The solution was to find the right wire types that can be stripped with very hot soldering tip. I have some enamel wires that have so tough cover that no amount of heat and soldering paste flux can strip them. I did use sandpaper for the stripping, but now I have given up to use them.

It is a little bit slow to change between the normal temperature and the stripping temperature, but it is still faster and more reliable than any other method I have used. When I am lazy, I am not even changing the temperature, instead I do the soldering with the "too high" temperature.

If and when you make the wiring mistakes or change plans, it is very easy to take the old wires completely out or reroute them.

I have been using the mesh of single wires - especially with the breadboards - but that is not good for robots or for the youngsters playing with the robots. The soldered enamel wires are very robust. With soldered wires, you can avoid the breadboard problems in lost female contacts and bad male wiring pins.

ag123
Posts: 634
Joined: Thu Jul 21, 2016 4:24 pm

Re: tiny stm32 F4 boards

Post by ag123 » Sun May 21, 2017 12:32 am

thanks Ollie, soldered enamel wire sounds like an interesting thing to try too :D
i've seen this http://elm-chan.org/docs/wire/wiring_e.html
it is quite inspiring/awesome, just that i'm not too sure about my own soldering skills to be able to do just that, but i'd think it's nevertheless something good to try :lol:

palmerr
Posts: 58
Joined: Tue Jan 31, 2017 6:21 am
Location: Melbourne, Australia

Re: tiny stm32 F4 boards

Post by palmerr » Mon May 22, 2017 1:19 am

Mine arrived in the post on Friday.

Has anyone looked inside the EEPROM to see if it's just a resource, or has any magic pre-installed?

Also, I tend to substitute long pin headers on my boards, rather using than the shorter ones supplied, so that I can attach logic analyser probes or put several wire wraps on one pin. Where there's enough space I use long-pin female headers - which provide an extra connection option. It's really useful for debugging with the logic analyser as "Dupont wire" (rainbow cables with individual header pins or sockets) pins are much more likely to stay in place than test clips.

Embedding the board in a piece of EVA foam (non-conductive) keeps the bottom pins out of trouble when they're not needed.

Image

It was really hard to solder the pins next to the debug header. I did the side ones first (which as a good idea) and then the cross piece. I insert the header pins into a socket, while soldering, to keep them straight. BTW, the poor solder joints in the front row have been fixed!

Next time, I'll insert the header pins from the top, solder on the bottom and then strip the black plastic former, so that the silkscreen is fully exposed.

Richard

Post Reply

Who is online

Users browsing this forum: No registered users and 5 guests