Arm China STAR-MC1 processor

[ali_asadzadeh]

Today I was reading keil release notes and I see this  China STAR-MC1 processor!
Do you have any info on that?

[ataradov]

It is not a processor, it is a core IP based on the ARMv8-M architecture. It is strange, since it is not listed anywhere on the official site, yet it is mentioned in a few places in erratas. It looks like this is a part of the whole story where ARM China went rogue. Or may be it was just designed for the internal market and some targeted customers.

Anyway, I don't think there are any actual devices using this core in the wild.

[SiliconWizard]

A few details: https://infigbt.com/arm-china-launched-xingchen-processor-and-has-30-authorized-customers/

[Simon]

If I vaguely remember correctly when we had a certain member and rows abounded about china versus the west it was that china decided that it could do as it pleased and having got an ARM subsidiary simply went rogue, and stole the designs, am I over paraphrasing or over simplifying?

[koendv]

MindMotion MM32F5277.  "no US IP, no problems".

[bidrohini]

Here is a related content too.
https://www.firstxw.com/view/263536.html

[ataradov]

So, yes, this is the part where ARM China went rouge and started to do their own thing.

That article is just a random word salad, even excluding the translation issues.

And peripheral set on MM32F5277 looks like a typical generic clone of ST. Meh, but I guess it is not going to target western market at all. Although 5 V supplies are curious.

[DC1MC]

As a 3V3 STM32 clone is not impressive, but as a full 5V capable (not just "5V tolerant") MCU is a game changer, at least in the Arduino/legacy replacement space, I think is the most powerful 5V MCU around at this price (please feel free to correct me if there are any other).
I've attached the brief here and I will try to get some Taoao modules via the cumbersome path of reshippers, if anyone knows some easyer way to get some in EU, please let me know.

 Cheers,
 DC1MC

[GromBeestje]

There are several Chinese STM32F103 clones whose datasheets claim to be able to run at 5 Volts out there. Such as the BLM32F103, HK32F103 and MM32F103. They're been around for years, so, a game changer, I doubt it.

[DC1MC]

There's a comparison here:

https://www.blaatschaap.be/32f103-comparison-part-1/

 and the MM part comes on top, also the others are not officially 5V, but "5V tolerant"and the documentation and software availability is pretty low. Of course this could also be a false start, but if it's really working reliably at 5V is worth every penny. I'll let you all know if the modules are ever coming form TaoBao.

 Cheers,
 DC1MC

[GromBeestje]

There's a comparison here:

https://www.blaatschaap.be/32f103-comparison-part-1/

 and the MM part comes on top, also the others are not officially 5V, but "5V tolerant"and the documentation and software availability is pretty low. Of course this could also be a false start, but if it's really working reliably at 5V is worth every penny. I'll let you all know if the modules are ever coming form TaoBao.

 Cheers,
 DC1MC

That's my website

Be aware, you can't use USB at 5 Volt, as the USB I/O voltage is 3.3.

[PCB.Wiz]

And peripheral set on MM32F5277 looks like a typical generic clone of ST. Meh, but I guess it is not going to target western market at all. Although 5 V supplies are curious.

Nothing 'curious' about 5V at all    - I've already mentioned that's a growing trend in small MCU Vendors.
The newest MindMotion MM32L small parts are 1.8~5.5V
3V limited parts will the the orphans of history.

With 120MHz, 256KB Flash storage, 192KB SRAM, that MM32F5277 is larger than 'small', and the  2.0V to 5.5V range joins some Nuvoton M4 parts  M451/M471 which are 2.5~5.5V and 72/120MHz

[ataradov]

To me it is curious because I don't get that 5V trend. It makes no sense to me for mass market. I guess it does not hurt if you can make it work without significant price increase. But there are fewer and fewer peripherals that are 5V. To me it seems like having level shifters to the outside world is more optimal design in most cases.

[DC1MC]

That's my website

Be aware, you can't use USB at 5 Volt, as the USB I/O voltage is 3.3.

Wut ?!? Where is this written that to use USB you have to use 3V3 all-over, doesn't have an internal regulator for that?

[ataradov]

Wut ?!? Where is this written that to use USB you have to use 3V3 all-over, doesn't have an internal regulator for that?

Right in the datasheet. VDD - USB operating voltage range is 3.0V - 3.6 V.

This is what happens when you simply slap 5V I/O ring on the exiting device without doing too much work.

[DC1MC]

Wut ?!? Where is this written that to use USB you have to use 3V3 all-over, doesn't have an internal regulator for that?

Right in the datasheet. VDD - USB operating voltage range is 3.0V - 3.6 V.

This is what happens when you simply slap 5V I/O ring on the exiting device without doing too much work.

Do you have a better MCU datasheet than the brief I've posted, that you can share, or can you kindly please show in the brief where is it written, I seem to have a blind spot and I don't see it .
It may be very well possible and it will be a pithy, such an powerful MCU operating fully at 5V AND with a working USB would be an exceptionally useful device. Even without USB is still very good, but USB will be perfect.

Thanks

 

[ataradov]

Not for the new MM32F5277. This comment was about the old MM32F103.

There is a chance they fixed this in the new one, of course. USB pins are not marked as 5VT in the multiplexing table, so they may be on a different I/O structure. Although in the old datasheet they are marked as "-", so who knows.

It looks like "-" in the old DS corresponds to "TC" (standard IO) in the new DS. So, the overall structure is the same.

[PCB.Wiz]

That's my website

The MM32F103 you have tagged as 168MHz but the data says 96MHz -40°C - 85°C  ?

Their newer value families
https://www.mindmotion.com.cn/products/mm32mcu/mm32f/mm32f_value_line/
seem to tag 72MHz, but they also include -40°C - 105°C

The new MM32F5277 tags 120MHz, 2.0V~5.5V, and up to -40°C to 105°C.

[ataradov]

Rev1.13 (which is searchable by google, but is not on the official site for some reason) of the datasheet states:

– Standard operating frequency is up to 96MHZ
– Maximum operating frequency is up to 168MHZ

  They don't clarify much, but it looks like the core and the closest buses can run much faster.

[DC1MC]

So actually MM has a nice English language site, but hell, it is well hidden 

So our "victims" are here, in the Performance Line section and they seem to have been spit in two products:
https://www.mindmotion.com.cn/en/products/mm32mcu/mm32f/mm32f_performance/

The clock is given as 120MHz, I'm attaching the Data Sheet here to spare you the pain of slooowly downloading it, indeed there is a chapter where USB is is speccd at 3V§ but I'm really curious what will happen if the USB is active while povered at %V, will it blow up, will destroy the connected devices or will it work, and the mention is only for Vbus ?!?!

It seems that there is a nice evaluation kit, that costs 50$, but nowhere to be found in EU

[ataradov]

indeed there is a chapter where USB is is speccd at 3V§

So, it is basically the same, which is expected to be fair.

but I'm really curious what will happen if the USB is active while povered at %V, will it blow up, will destroy the connected devices or will it work, and the mention is only for Vbus ?!?!

The device would be fine. It will use 5V for signalling. This will for sure violate USB spec and potentially damage host device.

This is the same thing as various V-USB implementations do. If you really want it, you can add 3.6 V zeners on USB data lines. It is still going to be crap as far as compliance goes, but would not be as damaging. 

Although it should not cause real damage, since USB spec requires

A USB transceiver is required to withstand a continuous short circuit of D+ and/or D- to VBUS, GND, other data line, or the cable shield at the connector, for a minimum of 24 hours without degradation. It is recommended that transceivers be designed so as to withstand such short circuits indefinitely.

But the way this protection works does not guarantee that data transfer would work at 5V. It may have a long recovery time, for example.

[PCB.Wiz]

It may be very well possible and it will be a pithy, such an powerful MCU operating fully at 5V AND with a working USB would be an exceptionally useful device. Even without USB is still very good, but USB will be perfect.

The MM32F5277 brief is a bit vague on if USB and 5V occurs, but the Nuvoton M4 NUC472 and M0 NUC029 parts mention a specific USB LDO.


Addit: Looking at how WCH do this, on their 5V USB parts, I see this in the CH32F103 data
VDD USB operating voltage
Disable USB5VSEL control bit 3.0~3. 6V
Enable USB5VSEL control bit  4.0~5.5V

Suggests they have either a simple LDO/level shifter to drop the USB pin drive a bit on 5V vcc.

[PCB.Wiz]

Rev1.13 (which is searchable by google, but is not on the official site for some reason) of the datasheet states:

– Standard operating frequency is up to 96MHZ
– Maximum operating frequency is up to 168MHZ

  They don't clarify much, but it looks like the core and the closest buses can run much faster.

A bit confusing, and quite a jump.
Sometimes vendors have higher CLKs for PWM blocks, for finer PWM steps, maybe it's related to that ?

[Kleinstein]

The 10 cent risc-V µC that Dave just showed in his new video
https://www.eevblog.com/forum/blog/eevblog-1524-the-10-cent-risc-v-processor-ch32v003/
seems to be running also up to 5 V, though no USB at the price level.
So 5 V operation seems to be not totally a thing of the past, though it gets less and less important.

[ataradov]

The MM32F5277 brief is a bit vague on if USB and 5V occurs,

The full datasheet is linked above. There is no voltage regulators, USB just does not work officially at > 3.6 V.

but the Nuvoton M4 NUC472 and M0 NUC029 parts mention a specific USB LDO.

That's because Nuvoton actually does R&D and not just shuffle the same ST peripherals around.

[ataradov]

A bit confusing, and quite a jump.
Sometimes vendors have higher CLKs for PWM blocks, for finer PWM steps, maybe it's related to that ?

No, they specifically mention "The maximum frequency of the AHB and the APB domains is 168MHZz." ("MHZz" is in the datasheet ). The core is in the AHB domain, so presumably can run at this clock.

This looks like some extreme overclocking, most likely at the expense of reliability.

[koendv]

It looks like a generic STM32, until you look closer.
- Clock generation is completely different, allowing for higher speeds.
- STM32 spi can do spi frames from 4 to 16 bits. The MM32 SPI has an extra register, where you can set frame sizes from 1 to 32 bit

If I compare the Star-MC1 MM32F5270 with the Cortex-M3 MM32F3270, I get the following impression.
- the F3270 has 128 kbyte ram at 0x20000000, and 512 kbyte flash.
- they ripped out the Cortex-M3 core and replaced it with the Star-MC1
- but the Star-MC1 has 32kbyte tightly coupled ram at 0x2000000 e.g. for stack
- so they moved the 128 kbyte ram to 0x3000000
- space got a bit tight, so they lowered the amount of flash from 512kbyte to 256kbyte.
But this is just the impression I get.