BluepillF4 PCBs

I would just buy the minimum order of PCBs

IMHO with such a compact board,there is a high lightlyhood of there being a small but critical error, or perhaps a tolerance which is too tight for the cheap manufacturing plant

I worked in the electronics industry quite a few years ago, and virtually every board that was designed had at least one problem.
Sometimes this was minor and we could compensate in the firmware etc, but often we had to slightly modify the PCB at least once.

I know in this case , we have many very talented people looking at the PCB, but as it’s Sooooo compact and uses a processor we don’t use that often, I still think there is a chance the boards won’t fully function correctly.

Of course, I agree with you Roger, same experience!

The thing is, 5 boards cost $13. 55 boards cost $20.
If it works or it doesn’t cost about the same. If I order 55 at once, and it does work, then there is 50 more to send around.

[victor_pv – Sat Feb 03, 2018 10:32 pm] –
The thing is, 5 boards cost $13. 55 boards cost $20.
If it works or it doesn’t cost about the same. If I order 55 at once, and it does work, then there is 50 more to send around.

I know there was a good deal of interest originally and I was even interested in assembling one. But that like 90 days ago. Today, I am distracted with other interests and if I were to procure a board, it would likely never get assembled. Time changes everything.

I think the design effort for the board is fantastic and I think a completely assembled board would be highly desirable if in the $5 range, but with the ESP32 pricing now under $5 and the rPi_Zero-W at $5 (sale) here in Atlanta, I just do not see myself taking the time to 1) source and order the uC and 2) assemble the board.

I have really enjoyed reading the ongoing discussion and decisions made in the long main thread; enlightening stuff.

Ray

You can add the Onion Omega 2 to the list of cheap boards.

But regarding all other competitive boards, the situation has not changed much in 90 days: maybe the ESP32 is now a little bit cheaper, but it was already in the <$7 game then. I know this one and the smaller ESP8266 very well (I helped to support the Xtensa CPU in GCC), but the cons are the number of GPIOs and high power consumption. Great board though, I am also using it a lot!

As for the RPi0W, the situation is not that simple: if you can get some at MicroCenter for $5 (lucky you!), for most of us here in Europe, it is difficult to get one below $20 with shipping

I agree the STM32F4 only fills the needs for a niche market between these boards and the cheaper Arduino mini/micro and BluePill boards in the <$2 range.

But with a 168 MHz / 210 DMIPS Cortex M4 with single precision FPU, DSP instructions and its plethora of peripherals, the STM32F4 addresses applications requiring high-processing power and real-time control, such as flight controllers in drone applications, where they dominate all other MCU types by far.

Of course, assembling a board from scratch takes time, but it is fun and rewarding, and it is up to everybody to see an interest in it!

Assembling such a board is fun for us who work with 0603/bga/qfn and are equipped with tools and with skills
99% of people who want to pay and play are unhappy when ie. even the pin-header is not soldered in already..
This board to assemble at home would require some experience, however.
Thus for US market you have to offer a fully assembled board for $5, as the Arduino Company and Raspberry Pi Foundation do :P

Yes, most users will want assembled boards. This version is still a prototype to be tested even for its most basic functions (e.g. turn on power), so this is different. We will be able to deliver some PCBs to volunteers, but expect tears, sweat and blood. No guarantee whatsoever that it will work for application X, Y or Z.

Then once we have a working prototype, we can make plans to have them manufactured and shipped pre-assembled (W/ or W/O pin headers soldered) directly from China, as this is the only effective distribution channel: for Rpi0W, the main online distributors are charging a lot for shipping, and small online shops tend to bundle this with junk to get a minimum amount order of at least $10, then you get additional $5 shipping fees.

But the first step is debugging the initial prototype board!

Oh, my, my regrets for causing this thread to go off topic … I seem to be a lightning rod for such things.

But with a 168 MHz / 210 DMIPS Cortex M4 with single precision FPU, DSP instructions and its plethora of peripherals, the STM32F4 addresses applications requiring high-processing power and real-time control, such as flight controllers in drone applications, where they dominate all other MCU types by far.

Those that follow my rants know that I am a big proponent of using high performance, low cost microcontroller boards in conjunction with RTOS higher level products for providing near-real-time data aggregation.

My intent on posting was to clarify why I wanted a board to assemble back at the start of Autumn ’17 and now mid-way into 1Q18, I am no longer interested. My decision certainly should not reflect in any manner on the usefulness of the design. Many members seemed interested in having an initial board and in doing the evaluation, I am just removing my name from that desire so someone else can participate.

Ray

Ray, and I think you make a good point, things move fast.
I decided to open this post because another forum member sent me a PM saying he would be happy to get some PCB and try to assemble to join in the effort of developing the prototype.
As long as people understand this is a prototype and the effort required, I think there is value letting more people test it, so we may move along faster, so I decided to open the thread to let people willing to join the effort know I can ship some bare PCBs to them, and if I have the time, I may be able to solder the SMD components in a few boards. While I can only work in a board at a time, if 5 other people is testing them at the same time I am sure we will move faster in finding any correction needed, or if there is no flaws in testing GPIO, peripherals, clocks, RTC, etc etc.
We are not yet at the stage of mass producing them, but perhaps we can get to that point soon.

No problem Ray!

No, this is not off topic, it is always useful to see the whole picture when you are focused on a subject.

There are a lot of low-cost boards available now for tinkering with, and this is a good thing. Among the inexperienced users, the low-cost Arduino micro/mini clones are the most used. For more power users seeking the lowest possible cost, the BluePill boards are a must.

OTOH, there are software-only people who are looking down to hardware stuff, and their favorite platform is then the RPi, because it is relatively cheap and you can find a lot of online resources. You may find more adventurous people looking into Onion Omega 2 or C.H.I.P., far less in BeagleBone Pocket.

Then you have the language-centered developers that feel more comfortable when finding a small board supporting their favorite language (Javascript, Python, Basic) for hardware-related projects.

A cheap, pre-assembled STM32F4 may find its purpose as a high-end MCU/FPU/DSP with lots of IOs, for applications like flight or 3D printer controllers, I think.

And in our case, the most interesting is not the destination, it is the journey! I enjoyed this project because I learned a lot while working on it, and I hope this is the case for others, too.

For what it’s worth, I’d be interested in attempting assembly and test. I’ve done manual SMT work before, it’s tricky to say the least, but not outside of the realm of possibility. The trick with a QFP is to drown it in flux, and allow shorts to happen, but draw solder away with the capillary action of the iron. The smaller parts are not quite so bad. I think if you have got this far, there is no turning back.

[electrobling – Sun Feb 04, 2018 10:15 pm] –
The trick with a QFP is to drown it in flux, and allow shorts to happen, but draw solder away with the capillary action of the iron. The smaller parts are not quite so bad. I think if you have got this far, there is no turning back.

I have done the same trick with QFP before, and didn’t have any problem. I think the worst for me is holding small components in place when they are all slick from the flux…

I agree the the previous postings about theoretical cost of other boards, vs what most of us actually have to pay for them.

Locally, to get a RPi zero, prices are nearer $15 USD
e.g. This is probably the cheapest I can get hold of
https://core-electronics.com.au/raspber … eless.html

ESP32 are around $10 USD direct from China (including shipping)

Boards like the BP and MM (and the ESP8266) etc are the only ones I can get in the $2 – $3 range

Victor

I’ve ordered almost all the parts for 5 boards. So if you haven’t ordered them yet then please put me down for 5 or 6.
These are on a slow boat from China so they won’t be here anytime soon! I will order the remaining ones from Mouser this week. Let me know if you would like me to get some for you. The prices for the AP2114HA-3.3TRG1, LQW2BASR47J00L and MCP73812T-420I/OT seemed better there and the shipping can be covered by another order I am placing. Only the first of these is required but we will need to test the optional parts.

Being a mathematician, I don’t have the experience to comment on the layout but I’ll try to review the schematic.
I don’t have a reflow oven but do have a cheap rework station and stereo microscope so I think assembly will be fine … but slow.

I have been mainly playing with some baremetal and HAL/LL/CMSIS programing on the STM32L4. So I should look at the F4 boards I have more seriously. Is the idea to work with the STM32GENERIC core ?

Eric

[victor_pv – Sun Feb 04, 2018 10:48 pm] –

[electrobling – Sun Feb 04, 2018 10:15 pm] –
The trick with a QFP is to drown it in flux, and allow shorts to happen, but draw solder away with the capillary action of the iron. The smaller parts are not quite so bad. I think if you have got this far, there is no turning back.

I have done the same trick with QFP before, and didn’t have any problem. I think the worst for me is holding small components in place when they are all slick from the flux…

For a small component, I apply the least amount of solder I can to only one pad. Then I place the component with tweezers and heat that end until it sticks. I go to the other end and make a proper solder join. Then I return to the original end and fix the bad join with a decent amount of solder.

[electrobling – Mon Feb 05, 2018 11:27 am] –
For a small component, I apply the least amount of solder I can to only one pad. Then I place the component with tweezers and heat that end until it sticks. I go to the other end and make a proper solder join. Then I return to the original end and fix the bad join with a decent amount of solder.

+1 .. also good tweezers are a must.

I don’t necessarily mean expensive ones, simply ones that have well defined tips that are parallel to each other and small enough not to be too clumsy to handle the component. I have filed away the tips of some cheap tweezers and carefully hammered them flat and parallel to make them more serviceable.

They tend to need some adjusting as time goes by due to the heat from the soldering iron.

Having the eyesight of a 20 year old would also help, but I’ve checked on ebay, and nobody appears to be selling this.

I would be interested in trying to build several boards, but I do have the problem of parts. I always order absolutely everything from China, as the shipping costs from the U.S. to Canada are astronomical because the companies like Mouser decline to use the post office. Also the Chinese New Year is upon us soon, and no orders to N.A. will be shipped for an entire month. That might mean a 2 month delay to get parts, assuming that I can source them.

[ahull – Mon Feb 05, 2018 11:42 am] –

[electrobling – Mon Feb 05, 2018 11:27 am] –
For a small component, I apply the least amount of solder I can to only one pad. Then I place the component with tweezers and heat that end until it sticks. I go to the other end and make a proper solder join. Then I return to the original end and fix the bad join with a decent amount of solder.

Having the eyesight of a 20 year old would also help, but I’ve checked on ebay, and nobody appears to be selling this.

Thank god i got my eyes from the good side of the family XD
my sisters have to wear glasses, while i’ll probably have eagle eyes till i’m 60 ^^

[racemaniac – Mon Feb 05, 2018 12:46 pm] –

[ahull – Mon Feb 05, 2018 11:42 am] –

[electrobling – Mon Feb 05, 2018 11:27 am] –
For a small component, I apply the least amount of solder I can to only one pad. Then I place the component with tweezers and heat that end until it sticks. I go to the other end and make a proper solder join. Then I return to the original end and fix the bad join with a decent amount of solder.

Having the eyesight of a 20 year old would also help, but I’ve checked on ebay, and nobody appears to be selling this.

Thank god i got my eyes from the good side of the family XD
my sisters have to wear glasses, while i’ll probably have eagle eyes till i’m 60 ^^

All bets are off when you hit 50
I had pretty much perfect vision till I was in my late forties, I now use reading glasses and complain about the writing on packaging, but I can still spot an eagle at 1000 yards.

i almost have an impulse to join a group purchase, e.g. the boards
on the record i’ve 2 castellated nrf51822 boards
https://www.ebay.com/itm/NRF51822-YJ-14 … 2396174917
after receiving them i’m dumbfolded as i’m not sure how i’m going to solder them, smt really takes some skill, it certainly isn’t as simple as those 0.1 inch through hole boards