New APM32F103CB

Post here all questions related to STM32 core if you can't find a relevant section!
Squonk42
Posts: 16
Joined: Sat Dec 21, 2019 8:35 am

Re: New APM32F103CB

Post by Squonk42 »

blue-man wrote: Sat Mar 28, 2020 8:19 am I only have quoted blaatshaap.be
The APM32F103CB is listed here as fake: https://www.blaatschaap.be/?p=120
The stop quoting and make your own mind;-)
blue-man wrote: Sat Mar 28, 2020 8:19 am But it is my opinion that selling undocumented functions that cannot be used can be seen as a fake functionality.
Here you can read what is praised by the manufacturer: https://www.apexmic.com/en/newproduct/apm2/16
System & Architecture

ARM®Cortex®-M3
Frequency 72MHz
Built-in AHB and APB
Support FPU

Do you want to support such business conduct?
It is your opinion, mine is that it is just typical bad Chinese documentation ;-)

When I started to work on a free gcc compiler for the ESP8266 MCU more than 5 years ago, there was only poor partial Chinese documentation and an illegally locked-down gcc-derived compiler from Tensilica which required installing a Chinese Windows XP VM and rolling back the RTC because of an expired licence to be able to compile a firmware :shock:

So yes, I supported and continue to support this "business conduct": just look now how Espressif has grown up and is now a total supporter of Open Source Software and how great their documentation now has changed!

Again, this is not the chip manufacturer that is to blame here, but the shops that are assembling board with erased chip marking. I suspect that at least some vendors on AE are not even aware of this traffic.
User avatar
blue-man
Posts: 51
Joined: Mon Mar 23, 2020 5:02 pm

Re: New APM32F103CB

Post by blue-man »

I can agree with what you have written.
But i have already written about my own mind. ;)

Your story with the gcc compiler is interesting.
From my understanding you are not supporting this "business conduct" - you support a hole in the documentation. :D
Of course you are doing this in a very positive way, because you enable other people to use this product.
i am asking me what would have happened, if you would have explained this dependencies to the manufacturer?

Honestly - at least i am simply disappointed that an product claim is not fullfilled.
Not more or less.
Squonk42
Posts: 16
Joined: Sat Dec 21, 2019 8:35 am

Re: New APM32F103CB

Post by Squonk42 »

Just to give you an idea about what needs to be achieved: imagine yourself having to write a documentation in Chinese for your brand new chip ;-) Most Chinese engineers are not speaking and often not reading anything but Chinese. Given their huge domestic market, they seldom put an effort in providing an English documentation because they just don't see the point.

Semiconductor industry in China is very different from the "Western" standards. Considering Chinese manufacturers as only copycats is strongly misleading: Chinese semiconductor industry is now innovative enough to provide disruptive chips like the ESP8266/ESP32 and in pass of becoming leaders in design in many applications.

To continue to be positive, this MCU seems to have an FPU attached, let's try to find some documentation, at least describing its peripheral registers. Maybe ask Robotdyn? They may be able to provide some help? Or skim the Chinese forums for FPU application examples in order to port them to the Arduino core?
User avatar
blue-man
Posts: 51
Joined: Mon Mar 23, 2020 5:02 pm

Re: New APM32F103CB

Post by blue-man »

O.K. Continue to find better documentation. 8-)

Your statements let me search for chinese documentation.
And it's true - there is a newer version from 2020/3/27.
https://www.apexmic.com/cn/newproduct/apm2/16

https://www.apexmic.com/uploads/tool/AP ... V1.1.0.pdf
On page 14 for the FPU - it is the same short text.

But there you can read that the FPU will support the IEEE 754 standard.
https://en.wikipedia.org/wiki/IEEE_754

This is standard: https://en.wikipedia.org/wiki/IEEE_754# ... operations
This is optional and not sure: https://en.wikipedia.org/wiki/IEEE_754# ... operations

Additional there is an update of the "Apexmic model guide", so it seems that this MCU is still under developement.
Hopefully better documentation will appear.
User avatar
blue-man
Posts: 51
Joined: Mon Mar 23, 2020 5:02 pm

Re: New APM32F103CB

Post by blue-man »

It would not be surprising if they used something like this: https://opencores.org/projects/fpu :D
Squonk42
Posts: 16
Joined: Sat Dec 21, 2019 8:35 am

Re: New APM32F103CB

Post by Squonk42 »

Here is the (Google) translation for page 14:
3.10. Floating Point Unit (FPU)
The product has a built-in independent FPU floating-point arithmetic processing unit, it supports the IEEE754 standard single-precision floating-point arithmetic.
As it is memory-mapped on the AHB bus (0x40024000-0x400243FF), it is probably not the standard ARM FPv4-SP core-integrated that can be found in the M4 or M7 cores.
Squonk42
Posts: 16
Joined: Sat Dec 21, 2019 8:35 am

Re: New APM32F103CB

Post by Squonk42 »

blue-man wrote: Sun Mar 29, 2020 8:17 am It would not be surprising if they used something like this: https://opencores.org/projects/fpu :D
No, generally, Chinese manufacturers are not confident in Open Software / hardware, they prefer to buy a commercial IP for these kind of purpose.
ag123
Posts: 1655
Joined: Thu Dec 19, 2019 5:30 am
Answers: 24

Re: New APM32F103CB

Post by ag123 »

my thoughts are that while apm32f103 seem to be a 'compatible' implementation of stm32f103, as mcus are complicated stuff. there is a chance things varies from a stm32f103. stm32 is popular in part due to the rich documentation which ST releases. but for 3rd party 'compatible' implementations, it would take one some risks to literally test the implementations to see that it works as one expects.
apparently the datasheet is as much docs as you can find about it in the open and while it is assumed to be 'compatible', functionalities may actually differs from what is stated in the datasheet ,it takes tests to find it out and one can complain if it works different, but it is hard to see if they would correct that unless you are a 'big' client!
there are a lot of arm based mcu implementations in which the manufacturers would request a NDA just to get the manuals and they may even charge for it or even refuse to provide it for 'small' clients
Last edited by ag123 on Sun Mar 29, 2020 11:28 am, edited 1 time in total.
User avatar
blue-man
Posts: 51
Joined: Mon Mar 23, 2020 5:02 pm

Re: New APM32F103CB

Post by blue-man »

Squonk42 wrote: Sun Mar 29, 2020 10:17 am As it is memory-mapped on the AHB bus (0x40024000-0x400243FF), it is probably not the standard ARM FPv4-SP core-integrated that can be found in the M4 or M7 cores.
If it would be the "standard ARM FPv4-SP core-integrated" it would not be peripheral.

I think the goal is to
a) save license cost
b) avoid stress with ARM and other manufacturers, because they don't want features of bigger MCU's in the smaller low cost one
c) have an selling argument for this F103CB MCU
ag123
Posts: 1655
Joined: Thu Dec 19, 2019 5:30 am
Answers: 24

Re: New APM32F103CB

Post by ag123 »

for the FPU, my guess is one would need to just ask for it (the documentation) from the manufacturer expressing interest.
if they seem 'uncooperative' it would diminish the popularity of even the chips itself given it is an 'advertised' feature
my thoughts are that it would likely be a 'simple' FPU implementation, and may not implement many 'hard to make' FPU features.

i used to remember the intel x87 fpu costing like $200 and has a big chip surface (i.e guess that is in the 1um days) and that is for a 'proper' implementation
Post Reply

Return to “General discussion”