Interesting microcontrollers from China no one heard about - how to use them?

[soFPG]

While surfing the internet I found out about some very interesting asian tech-companies. Two of them especially drew my attention:

1. Micon Design Technology
2. Padauk

The reason why they fascinate me are their tiniest, most low-performance microcontroller products. Micon Design Technology produces (or produced? I don't know if this one is EOL) the MDT90P01, a 4-bit EPROM µC (http://www.yspringtech.com/Spec/MDT-OTHERS/90P01.pdf).
Some sellers on alibaba offer them for quite a lot of money for small, hobbyist-sized quantities (one offer I received was 10$ / Piece  ). Sad enough, the documentation for programming this device is only available in Chinese (http://www.mdtmcu.com/upfiles/file/MDT/90P01%20application.pdf), I found out a few things using google translator for documents. Now I know that the hardware-programmers are called "WMD011" or "WMD021(A)". I did not find any available offers, only a discontinued one for a whopping 250$. And they are still using a serial console to connect to a PC
This product must be quite old then.

On the other hand, Padauks controller are available. The PMS150C-U06 is the cheapest microcontroller I found (https://lcsc.com/product-detail/PADAUK_PMS150C-U06_C168658.html), it is only 3 cents! This controller has a better documentation than the one from MDT (https://datasheet.lcsc.com/szlcsc/PMS150C-U06_C168658.pdf), it looks like the device programmer is called "PDK3S-P-002". It is available for about 60$ on some asian websites, their so called "IDE" is quite useable, assembler works fine (http://www.padauk.com.tw/products.php?item=14).

Does anyone here have experience with these devices? The MDT90P01 seems to be quite unusable due to its high price on alibaba and no available hardware-programmer.
I wroter an email to both of their customer services, asking for a small batch of samples but I don't think I will receive an answer ever.

If there is a chance to receive these hardware programmers I would think about reverse engineering the interface and port it as a bit-banging software to an AVR or similiar µC.

(I know that there is the PIC10F200 - I already did a few things with it, I also know both devices described here are OTP, I know that it does not make sense at all to use these devices - but I am excited thinking about using them for a small project).

Edit: Just received a new offer from alibaba for 0.2$ / piece for the MDT90P01 - that's reasonable.

Edit_2: The Assembler for the 4-bit MDTs can be found here: http://www.enroo.com/info/1/263/index.aspx (direct download link: http://www.enroo.com/uploaddir/top/file/MdtAsm4bita.rar)

Edit_3: The Writer software for the WMD011 / WMD021 / WMD021A can be found here (direct download link): http://www.enroo.com/uploaddir/download/software/mdt_new_program/MDTWriterV263.rar

[soFPG]

As I said - I am doing this as a hobby with no commercial interest (at the moment). Its just fun for me (maybe this sounds crazy).

I know that there is a much easier way to do things with µC from more well known companies (Microchip, STM etc.) - but I don't care. In this case, not the project is the focus for me but the µC and its instruction set itself.

I am sorry but I don't want to discuss whether it does make sense to use these or not - at least not in this thread.

There are a lot of small companies in Taiwan, and recently, mainland China, making those ultra cheap 4-bitters and 8-bitters.

Oh really? Its hard to find any information about 4-bitters, the only other company I know producing them is "EM Microelectronics" in switzerland. Do you know more than the ones already named?

there's no reason for me to use them

It is just fun for me - that's the best reason, isn't it? 

[soFPG]

I made another search on some chinese-websites (e.g 1688.com) but they require AliPay as a payment method. Do you recommend to register an AliPay account? Is it safe to use?
They sell a MDT90P01 as cheap as 0.5 Yuan.

Do you have any experience with those tiny microcontrollers or help me getting one of those programmers for the MDT-µC?

[T3sl4co1l]

If you're doing something in such quantities that you are shopping for 4-bit processors, you'll be talking with OEM reps directly.  And also negotiating on a price, and probably getting that price down even further with mask ROM, and fun stuff like that.

If you're doing things in near-zero quantity... seriously?  I mean, you can literally ask a dozen western companies, and they will send you their parts, at their expense!  They want you to use their stuff.  TI has complete dev boards they sell at cost, like $10 to the end user, they probably cost at least as much to make let alone distribute!

So, do you just want 4-bit because it's quirky and weird?  I mean, you can play with that in an FPGA (there's probably an open core for that?), or emulated on any other CPU.  Semantically speaking, it's no different.  So I'd guess you're not after semantics, but then, what?  Just the end-to-end experience of trying to do something that a rude forum member with far more posts than politeness complains about? ... 

Tim

[wraper]

Semantically speaking, it's no different.  So I'd guess you're not after semantics, but then, what?

masochism

[T3sl4co1l]

Semantically speaking, it's no different.  So I'd guess you're not after semantics, but then, what?

masochism

Good point.

Not a bad description of many of my projects...

Tim

[mikeselectricstuff]

I could totally think of uses for a 4 cent MCU. I do wonder if that  price might be a  clear-out reduction rather than the real cost. If the competition is 20-30 cents there's no reason to make one at 4 cents

[soFPG]

I mean, you can play with that in an FPGA (there's probably an open core for that?), or emulated on any other CPU.

You are right, but a FPGA is much bigger, more expensive and it would only contains the CPU core, no memory or RAM (even though RAM can be synthesized with LUTs). Some FPGAs feature built-in flash storage, but most of them don't. So there would be the need of external components.

If the competition is 20-30 cents there's no reason to make one at 4 cents

But why are they doing it? Someone is buying their µC, otherwise the would be bankrupt by now.

Why not?

I don't know - my trust is not particularly high if it comes to chinese payment methods.

masochism

Maybe - but my point of interest is to design a circuit as small as possible which is able to do things no one thought it would be capable of.

So, do you just want 4-bit because it's quirky and weird?

Yeah! And I would like to reverse engineer their (OTP)-programming algorithm used by their hardware device-programmers. So in the future more people would be able to use those tiniest µC without the struggle to get a programmer (which is the real problem I think).

[wraper]

So, do you just want 4-bit because it's quirky and weird?

Yeah! And I would like to reverse engineer their (OTP)-programming algorithm used by their hardware device-programmers. So in the future more people would be able to use those tiniest µC without the struggle to get a programmer (which is the real problem I think).

Programmer is one of the smallest problems associated with use of such MCUs

[soFPG]

How can the programmer be a minor problem if it is not even available?
Now I am interested in your opinion on what would be a major problem to get these devices running?
I have some experiences handling devices with very limited ressources from working with a PIC10F200 (if that is what you are concerned about)

[mikeselectricstuff]

If you're in a situation where very low parts cost is important, then the extra effort for coding in assembler, writing an assembler, building a useable production programmer, reverse-engineering programming protocols etc. is probably not a huge deal

[wraper]

How can the programmer be a minor problem if it is not even available?
Now I am interested in your opinion on what would be a major problem to get these devices running?
I have some experiences handling devices with very limited ressources from working with a PIC10F200 (if that is what you are concerned about)

You have no documentation, no compiler, no nothing, cannot buy them through normal sources. There are so many showstoppers before you even think about programming, it does not even manner can you program them or not. And when they are OTP, you cannot even play with them without discarding MCU at each try.

[EEVblog]

I could totally think of uses for a 4 cent MCU.

Especially if you can order entire reels programmed for near that cost.

[soFPG]

There is indeed some documentation, at least for the Padauk uCs. The user manual describes how to connect the chip to the programmer and what software to use. I already played with the IDE. It is pretty straight forward to use even though its UI looks like from Windows 95.
There is less documentation for the MDT ones. But a document describes which tools are needed and how the device programmer is called. So it should be possible to figure that out as well.
Alibaba is a reasonable source - like Aliexpress where I ordered several times already.
You are right regarding the OTP thing. But you can buy 10 Padauk chips for one PIC10F200 - they even have timer and overflow interrupts!

Edit: I do not speak korean or chinese. Would be kinda sad if there would not be a place for this discussion on this forum.

[BravoV]

I made another search on some chinese-websites (e.g 1688.com) but they require AliPay as a payment method. Do you recommend to register an AliPay account? Is it safe to use?

Why not? China has a no cash society (at least for young people), and AliPay and WeChat payment are used for everything low value (food, taxi, super market, etc.).

Just saying ...

"... online payments is the clearest example where China has leapfrogged the United States, with mobile phones replacing credit cards almost entirely as a form of payment in major Chinese cities, while “many Americans still use cheques.”

Source -> https://www.reuters.com/article/us-usa-trade-china-intellectualproperty/trade-war-or-not-china-is-closing-the-gap-on-u-s-in-technology-ip-race-idUSKBN1HK187

[coppice]

I don't know how on earth did they get to the 4 cent price point, but even in HQB, a simplest chip (74hc14) is more expensive than that.

The 74HC14's price is held higher by its 14 pin package. Packaging costs tend to dominate with these simple die parts. The 3.82 cent MCU is in a 6 pin pack. In volume that can probably be sold for 4 cents. What looks strange is they seem to be offering them in small volumes for 4 cents.

[amyk]

I don't know how on earth did they get to the 4 cent price point, but even in HQB, a simplest chip (74hc14) is more expensive than that.

The 74HC14's price is held higher by its 14 pin package. Packaging costs tend to dominate with these simple die parts. The 3.82 cent MCU is in a 6 pin pack. In volume that can probably be sold for 4 cents. What looks strange is they seem to be offering them in small volumes for 4 cents.

Indeed,and if you want bare die (large volumes only --- they won't sell you a single dice, nor even 1000), you'll find the price drops through the floor... I've heard <$0.01USD for 4-bit mask ROM ones (the mask cost is not cheap, however, so these are for ultra-high-volume applications.)

[Kjelt]

Back in 2006 I was talking to a uC manufacturer in China about their 4 bit uC then costing $0,11 and lower with high volume we were discussing. It could replace $0,40 of logic on our boards so quite interesting with massproduction.
There were some bizar quirks in those things. They were manufactured for one single goal: cheap toys.
The manufacturing process was not as you expect from fabs these days with a result that in your firmware at boottime you had to test the ram since unknown bits would (not could) be dead and you had to make a map so not to use that byte in ram. Also upto 2% of the devices would have problems in the alu or other critical parts. I asked how I could know that the first code would run ok if the ram was not to be trusted. The answer was you could tell from the end product if it worked or not.
I couldnt believe this but had it b&w in an email.
They probably also sell 100% tested uCs that had no flaws but that was pricewise much less interesting.

[soFPG]

Call me crazy but I ordered an ICE for the Padauk-uCs yesterday 
I hope I will receive it - you never know how trustworthy those chinese shops are.

I also added links to some executables needed to work with the MDTs - their programmer costs about 100USD.

you had to test the ram since unknown bits would (not could) be dead and you had to make a map so not to use that byte in ram

Does not sound like smth I would look forward to work with. Let's see...

Taobao has a lot of those controllers for as low as $0.02- that's crazy.

[janekm]

Chips like these do have their place. For example, I have a project that is using an I2C bus for extension modules. If I want to use a chip on one of those that doesn't offer I2C interface I'll have to create an adapter; there are specialised chips for that but they are expensive.

Similarly, you might like to have a microcontroller in a battery pack to keep track of charge cycles.

OTP is not such a big hurdle at the cost of these chips they're literally cheaper than potato chips...

[westfw]

Even if I've never fallen into the whole "RAM is free" trap that us embedded folks decry in desktop programmers, I'm afraid that my workflow patterns are no longer compatible with OTP microcontrollers.  Or even "UV-erase and program in a separate device", though I could probably manage.

Which is a shame, because I once bought a bunch of Toshiba OTP 4bit chips (TMP47P443 I think?) (alas, no tools exist.  I don't think free tools ever did.  There was an assembler you could purchase, and programming is done with some sort of adapter board that makes it look "almost" like one of the EPROMs of the day.)

These days, if I'm looking for 'really cheap' in the "sorta big quantities, for a hobbyist" range, I browse through Newark's "overstock" list.  (currently for example, NXP/Freescale/Motorola MC9RS08KA1CSCR (1k pgm, soic8) for $0.14.  A little while ago, MC9S08QD4 (4k PGM, better CPU) for $0.08 and MC9RS08KA (8k, SOIC16) for $0.06) (I bought 100 each of those.  I almost entered the "Black Widow" contest that was sponsored around the QD4, some years ago.   I think I have the evalBoard/programmer around.  Somewhere.))

[soFPG]

What are these https://item.taobao.com/item.htm?spm=a230r.1.14.21.58ea4577rsSxvo&id=544473902221 ?
The only kind of resource I find about them online are chinese shops - no manufacturer. They say they are PIC10F200 compatible but some of them in a similiar price range are OTP - how can the programming interface (PicKit) be compatible in this case?

[soFPG]

Thanks alot - your knowledge about hardware and what to find where is surprising to me 

Are they just exact clones of PIC-µCs? As I mentioned already - I don't find any documents about them. Do you also have a hint for me this time?

[Kjelt]

Programmer: https://m.intl.taobao.com/detail/detail.html?id=544300886130&spm=a21wu.9600033.recommend.1&main_itemid=544473902221&go_item_id=544300886130&scm=1007.10766.85773.1002003000000001&pvid=7a34b622-f0c6-4682-8f4d-b86eb237a3f1&track_params=%7B"pre_seller_id"%3A"107061321"%2C"match_type"%3A"3_544473902221_107_667_361_460_150_"%7D

Below they sell pic micrcontroller for 0,5 to 2 CNY that would be €0,06 to €0,25 
Too bad I can't read chinese, on the other hand I think my room is already full enough 

[janekm]

What are these https://item.taobao.com/item.htm?spm=a230r.1.14.21.58ea4577rsSxvo&id=544473902221 ?
The only kind of resource I find about them online are chinese shops - no manufacturer. They say they are PIC10F200 compatible but some of them in a similiar price range are OTP - how can the programming interface (PicKit) be compatible in this case?

Sometimes what they mean is just that the are pin-compatible. If you are making 1000000 of a product, you may not mind rewriting the firmware.

[Elasia]

Interesting company

https://item.taobao.com/item.htm?spm=2013.1.w4004-16634479443.12.22113245XUQCwS&id=36803356892

scroll through that and they more or less say they authorized by Microchip

[coppice]

Interesting company

https://item.taobao.com/item.htm?spm=2013.1.w4004-16634479443.12.22113245XUQCwS&id=36803356892

scroll through that and they more or less say they authorized by Microchip

Where? I only saw they carry genuine pic, not saying they are authorized dealer.

I think he was referring to the picture showing they are (or were - its from 2007) a Microchip Design Partner. There are a lot of those. Its fairly easy to become one.

[^_^]

There's GD (GigaDevice). They're not that small (e.g. Keil supports some), but being embedded engineer in Europe for few years, I've never ever heard of this...

These guys basically did a ST clone (ST's are apparently very popular in China).
They also license ARM, they also license USB ipcore. Peripherals are the same.
The best (funniest) thing is that they use same memory map, so most code will run without modifying it. One can also use ST libraries 

Why would one use them? I guess quantity price. Commercial purposes

[soFPG]

I also read about them at the olimex blog (https://olimex.wordpress.com/2015/11/09/chinese-clones-attack-stm32-microcontrollers/).

I don't know if this can be real, but a Cortex M3 is just 0.32$: https://item.taobao.com/item.htm?spm=a230r.1.14.17.176a27ebJhOfTt&id=566046633333&ns=1&abbucket=2#detail

http://www.gigadevice.com/products/microcontrollers/gd32/arm-cortex-m3/value-line/gd32f130-series/gd32f130c8t6/

If they are really so close to the STM-ones I bet a ST-Link would do the programming job?

[janoc]

They are real - here is a series of articles comparing STM32F103 vs GD32F103

https://smdprutser.nl/blog/stm32f103-vs-gd32f103/
https://smdprutser.nl/blog/stm32f103-vs-gd32f103-round-1-solderability/
https://smdprutser.nl/blog/stm32f103-vs-gd32f103-round-2-blink-a-led/
https://smdprutser.nl/blog/stm32f103-vs-gd32f103-round-3-uart/

It is using regular ARM SWD and JTAG, so STLink (or any other SWD-capable) dongle should work with it. It is not a 100% clone, though - there are some differences and gotchas.

[^_^]

Yup. For development you can go for JLink as well (fast memory readout for debug).
ST loader works perfectly even for the option bytes

[Elasia]

anyone here buy their chips in a large run? if so how did they pan out?

[poorchava]

Actually GigaDevice are not that small. They're a part of consortium that have bought SSI, beating Cypress' bid.

Their STM clones actually have a separate flash die in the package and there is extra SRAM to cache the flash so that they can run at high speed.

[JanJansen]

Maybe STM should copy them back.

[^_^]

Maybe STM should copy them back.

First GD should make a proper datasheet. The information density is a joke. One has to refer to ST's one, which is so funny!

[Elasia]

Maybe STM should copy them back.

First GD should make a proper datasheet. The information density is a joke. One has to refer to ST's one, which is so funny!

It actually would make sense that they were a former factory for ST that ST pulled out of hence the near identical clones and not much of their own doc work.. happens all the time on scale.. best factory wins and all the others get scraps, rossmann just did a recent video on ftc vs ice and mentions it as well.

[JanJansen]

First GD should make a proper datasheet. The information density is a joke. One has to refer to ST's one, which is so funny!

The datasheet cost more efford to develop then the chip i am sure.
Those Chinese are very clever, only not so hard workers, more of the lazy kind.

It actually would make sense that they were a former factory for ST that ST pulled out of hence the near identical clones and not much of their own doc work.. happens all the time on scale.. best factory wins and all the others get scraps, rossmann just did a recent video on ftc vs ice and mentions it as well.

There you have it, they already have 1 quarter of the market to start with, and are taking over many more if we keep buying.

[coppice]

Maybe STM should copy them back.

First GD should make a proper datasheet. The information density is a joke. One has to refer to ST's one, which is so funny!

Use the Chinese documentation. Can't read Chinese? Then you're not really their target market.

[slugrustle]

soFPG: any updates on programming your PMS150C-U06 MCU?

I just ordered dev tools for the Holtek HT66F/HT68F and HT48R. I'm looking to use an HT48R003 in a personal project, but buying parts on lcsc got me looking at others. The Padauk PMS150C looks most interesting to me.

I'm probably going to stick with Holtek's 8 bit parts for a few reasons: I'm familiar with PIC10/12/16, and the architecture is similar. I like the PICs, but it always bugged me that the free compiler is crippled and they charge to much for the full version; I've programmed them in assembler in protest, and for the fun of it. It's so cool that Holtek just gives you the IDE + compiler. A few products at my work use Holtek MCUs, so understanding them might come in handy that way.

It is a lot of fun to program these little barebones parts where it's not too much work to fully understand everything that they can possibly do. There's something satisfying about that.

[brucehoult]

Wow. 990 bytes of (one time programmable!) program memory. A few bytes of RAM. A single 8 bit accumulator. Control flow is only compare-and-skip-next-instruction-if-equal or increment/decrement-and-skip-if-zero or test-bit-and-skip-if-hi/lo.

Cruder than even 6502 or older PIC or 8051 or 2650 or SC/MP.

[slugrustle]

Aside from the OTP vs. flash, this is on par with a PIC12F508 / 509. This architecture is cheap to implement and produce, but surprisingly useful.

[niladherbert]

Interesting company

https://item.taobao.com/item.htm?spm=2013.1.w4004-16634479443.12.22113245XUQCwS&id=36803356892

scroll through that and they more or less say they authorized by Microchip

They have a 'selection of authentic smoke'
https://yueyd.taobao.com/?spm=2013.1.1000126.d21.6ed04dafJEF0NU
scroll down to the ICD and evaluation boards
All this 4-bit stuff looks really interesting. if you have some success I might follow yous after I use up my AVRs. Remember that Hackaday 1kb contest? Gather up all the guys on here that are interested and run one for 4-bit stuff.

[funkathustra]

For what it's worth, I did a lot of research into low-cost MCUs from the Chinese marketplace for my $1 Microcontroller Review. I considered a lot of parts discussed in this thread. Before getting into the crazy 4-bit MCUs out there, anglophones may want to stick to the offerings from Holtek, Nuvoton, and STC. All three have good English documentation for their parts and wide availability on Taobao/AliExpress/etc.

Also, all three parts have decent development experiences (Nuvoton and STC have Keil; Holtek has HT-IDE3000), and their programming tools are low cost (STC can actually be programmed using a 3.3V / 5V TTL USB UART adapter).

Before ordering hundreds of dollars worth of stuff, dip your toes in the water with one of these three offerings and you'll find out if it's something worth pursuing further.

In my case, while I initially thought I'd be going back to these vendors more, I haven't found a killer application for any of these chips — and considering you can get fully-functional SiLabs EFM8s for $0.32 each in single quantities, it's tough to justify the added development pain.

Sure, that particular EFM8 only has 2 kB of flash, but it has a much nicer peripheral selection than a lot of ultra-cheap MCUs: 18 I/O with 15 channels of 12-bit ADC, plus 4 timers, two comparators, SPI, I2C, UART, and PWM modules.

Food for thought! Keep us updated on your adventures!

[Doctorandus_P]

A long time ago Dave did a teardown of a toothbrush on his blog, and it had some weird 4 bit uC.

When the battery of my (braun) toothbrush was worn out recently I opened it and it had an MSP430 in it, and so do other (braun) tooth brushes, but the model is not always the same.
This suggests these are very cheap if you negotiate properly with the manufacturer and buy enough of them.

In some of the installation guides on https://toothbrushbattery.com you can also read the type number on a number of toothbrushes.

But I do not care much for the lowest cost uC. I'm more interested in decent peripherals and availability of free high quality development tools such as GCC.

[coppice]

A long time ago Dave did a teardown of a toothbrush on his blog, and it had some weird 4 bit uC.

When the battery of my (braun) toothbrush was worn out recently I opened it and it had an MSP430 in it, and so do other (braun) tooth brushes, but the model is not always the same.
This suggests these are very cheap if you negotiate properly with the manufacturer and buy enough of them.

In some of the installation guides on https://toothbrushbattery.com you can also read the type number on a number of toothbrushes.

But I do not care much for the lowest cost uC. I'm more interested in decent peripherals and availability of free high quality development tools such as GCC.

The 4 bit MCUs go in the higher voltage Braun brushes. The MSP430s they use are the 0.9V parts. which work in brushes with a single NiMH cell. The lower battery cost more than pays for the use of a specialist MCU.

[Rasz]

[electronic_eel]

Any progress with the Padauk controller?

[ebclr]

Here you can buy with paypal

https://lcsc.com/brand-detail/1246.html

[PCB.Wiz]

On the other hand, Padauks controller are available. The PMS150C-U06 is the cheapest microcontroller I found (https://lcsc.com/product-detail/PADAUK_PMS150C-U06_C168658.html), it is only 3 cents!

Keep in mind the very cheapest MCUs are quite brain-dead, and worse, usually OTP.
ie if you use high volumes, they can make sense, but for general MCU use, in modest volumes, you are better to get FLASH parts.

You can get FLASH 8051 cored parts, with ADC/UART/SPI etc, for the 20c~30c region, and there is a resurgence of 8051 in Asia, so variants are now very common.

[xaxaxa]

the stm32f030f4p6 (probably a clone) can be had for $0.3 from taobao... it is cortex-m0 @ 48MHz and has 4KB ram and 16KB flash.

[brucehoult]

the stm32f030f4p6 (probably a clone) can be had for $0.3 from taobao... it is cortex-m0 @ 48MHz and has 4KB ram and 16KB flash.

Does that have any memory that can hold program variables while the power is off? Like the EEPROM on an AVR. Can a running program write to the flash?

[mikeselectricstuff]

the stm32f030f4p6 (probably a clone) can be had for $0.3 from taobao... it is cortex-m0 @ 48MHz and has 4KB ram and 16KB flash.

Does that have any memory that can hold program variables while the power is off? Like the EEPROM on an AVR. Can a running program write to the flash?

Are there any flash micros today that can't write their own flash?
Flash can be less than ideal for param storage due to page sizes & endurance, but usually manageable if minimum cost is your goal.
External eeproms aren't exactly expensive -several under 5 cents at qty 1k at Digikey

[brucehoult]

the stm32f030f4p6 (probably a clone) can be had for $0.3 from taobao... it is cortex-m0 @ 48MHz and has 4KB ram and 16KB flash.

Does that have any memory that can hold program variables while the power is off? Like the EEPROM on an AVR. Can a running program write to the flash?

Are there any flash micros today that can't write their own flash?

No idea, I only have experience of a tiny subset of all microcontrollers made, which is why I asked.

[coppice]

Are there any flash micros today that can't write their own flash?

Depends what you mean by "write their own flash". Quite a few MCUs require an external high voltage, as the die area needed for a charge pump has been saved. Many require a complex protocol on the JTAG port, with fairly precise timing, in order to program the flash. Many which can do their own programming just by being supplied with the necessary data require software loaded into them to be able to do this - e.g. a JTAG port is used to load the programming code into RAM, or an initial JTAG controlled programming sequence gets the programming code into one segment of flash, from which the other segments of flash can be programmed just by feeding the MCU the data.

You might expect savings to only be made on small devices, but some big complex automotive MCUs lack the internal charge pump. This might possibly be a safety issue - ensuring the code can never change accidentally in the field, if the MCU gets kicked into an unfortunate mode. This is a common cause of corrupt flash MCUs when complete internal control of the flash is provided. Quite a few projects need their code updated for this kind of problem, when returns from the field flood in. 

When you look at things like FRAM density versus flash density, the high voltage charge pump is a big factor. Its a large overhead when the flash is small, but a small overhead when the flash is large. This is why you see things like TI focusing its FRAM efforts on smaller MCUs.

[Mr. Scram]

Though I don't have any practical use for cheap and possibly dubious microcontrollers, I totally understand the fascination with slightly exotic microcontrollers that come are at the extreme of what the industrial world can produce.

[romhunter]

Call me crazy but I ordered an ICE for the Padauk-uCs yesterday 

I know this thread is quite old but did you received it yet?
I'm interested in those 4 bitters too.

[moeabode@netzero.com]

Well, maybe a long shot but here goes; anyone have firmware for    PMS131-S14 PADAUK ? prototype that was made in China used this micro controller . They said they don't have firmware which causes a big issue when using same chip and mass producing
Thanks in advance, regardles