3 cent MCU

[nctnico]

That argument is a bit far fetched because you'll need additional logic to communicate as well. At 1k volume the solution costs over $30 in parts alone because you'll need bypass capacitors as well (and not taking the cost of pick&place time into account). A larger microcontroller which can control 100 individual LEDs (using 20 I/O pins for a 10x10 matrix) is likely to be cheaper.

[b_force]

I don't get the confusion.
A decent known brand uC with similar (or better!) specs is around 20-25 cents a piece.
So on your whole design you save about 15-20 cents.
That isn't even a significant number when only looking at the PCB parts/connectors.

Plus for those extra 20 cents you know you have at least a decent customer service, readable datasheets and you don't have to learn Mandarin when having questions about certain functionalities.

It depends on how you want to safe every penny, but for those quantities my customers can get those 20 cents for free for having a more reliable product.

[Koen]

There's a whole bunch of people speaking Mandarin, they might be the main market.

[Siwastaja]

That argument is a bit far fetched because you'll need additional logic to communicate as well. At 1k volume the solution costs over $30 in parts alone because you'll need bypass capacitors as well (and not taking the cost of pick&place time into account). A larger microcontroller which can control 100 individual LEDs (using 20 I/O pins for a 10x10 matrix) is likely to be cheaper.

Your problem right here is that if you don't see how you'd be utilizing these parts, you assume no one else could work it out, either; it's beyond your imagination.

This is a fallacy, and it's not "far-fetched" at all; just engineering, sometimes a specific solution happens to work great even if it's unthinkable to many or even most engineers such as yourself. It may still not be far-fetched for others, with different ways to do their design work.

First, you don't necessarily need "additional logic to communicate". These microcontrollers have - gasp! - IO in them. Communicating is well possible with zero extra components; you'd bit-bang anything you need, most likely something simple such as UART, but doing I2C could allow you interfacing with some other existing system. Or maybe you need to go through a simple resistor; or even AC coupled / level shifted through a $0.001 (add another $0.001 for P&P and yet another $0.001 for PCB real estate) capacitor with Manchester encoding. There are countless options if you need them and have your mind open.

Remember I mentioned a certain BMS system I designed and tested, and sold in smallish quantities? In this, I used capacitor level shifted, custom bit-banged Manchester encoded communication. It was simple & reliable as hell. The ATTiny25 dominated the BOM cost, even over the board cost, P&P cost, and test/calibration. And because a "typical" system would use between 20 to 100 products, there is this effect that building 100 products produces 5000 such subassemblies!

"A larger microcontroller" driving 100 LEDs as a multiplexed matrix might be unable to provide, say, 10-bit PWM dimming. Also, there could be wiring reasons to go with "local logic" instead of spaghetti. Even if doable, the additional wires and connectors (or more complex PCB routing, and using a larger board) would be an even additional cost in final assembly stage.

This is engineering; you can't know beforehand which solution is optimal, without knowing the exact problem. The more you have tools in your (mental) toolbox, the more capable you are. This is the cliché of "thinking outside the box".

P&P time is a fair point, but it's somewhere around $0.01 per chip, so even if you fix my numbers to account for that omission, there is no significant change in the result.

[Siwastaja]

Plus for those extra 20 cents you know you have at least a decent customer service, readable datasheets and you don't have to learn Mandarin when having questions about certain functionalities.

This is the specific kind of false FUD argument that pisses me off. Not because I think a Padauk controller would be a nice&easy experience to work with, but because this gives you an impression that the "classic", more expensive alternatives would.

Yes, I use STM32, a reputable Western brand; I have used them extensively for years. No, I don't have a readable datasheet (reference manual to be correct), I need to reverse engineer the shit out of their broken and undocumented, or unintuitive designs, and yes, I have contacted the customer service several times, either to get help (clarification on manual omissions), or to report a silicon bug (with a minimum reproducible test code that I have tested with several chips to rule out unit variance, possibly spending a full day just to make a proper error report) to be included in the errata sheet, but they don't bother replying at all. There is no customer service. Or if there is, it's much more than the price you paid for the chip. It probably requires some kind of multi $10k support agreement, I don't know? Or maybe you need to be a massive customer (speaking of volumes).

I have learned to live with this and actually don't want to complain too much about it; I understand they (ST) are making cheap hi-tech shit quickly and that means limiting the resources in documentation and customer service. I do the same sin. I don't have time to document and support everything properly either.

So, I feel myself fairly safe with any kind of a microcontroller, because I can handle it, I'm used to it, and I'm used to slowdowns and can account for them beforehand. I'm positive I'd be just fine with a 3 cent Padauk, if the need ever arises, even if they had STM32-level customer service (this is, no customer service). OTOH, there is a possibility that they do have English customer service - look at their English documentation for example, they have clearly spent some resources doing it -, how do you know without trying it out?

The power of this chip is in simplicity. Given some very basic documentation, you can work out the missing details by the same process needed with an STM32 - through revense-engineering and guesstimation. But because it's simple, it's unlikely to be a completely dead end. And if it is, you'll see it very quickly.

[amyk]

Interesting device, Its built in a 180nm process, so the core is running way below the 5.5V maximum specified Vcc. That means it must have an internal regulator (which is a die hog), and its managing to operate that with no external pin for a cap.

Thick gate oxide makes 180nm 5V capable. I wouldn't be surprised if the Far East has some unique process variants that aren't found elsewhere.

[wraper]

Interesting device, Its built in a 180nm process, so the core is running way below the 5.5V maximum specified Vcc. That means it must have an internal regulator (which is a die hog), and its managing to operate that with no external pin for a cap.

Thick gate oxide makes 180nm 5V capable. I wouldn't be surprised if the Far East has some unique process variants that aren't found elsewhere.

IIRC Silicon Labs EFM8 MCUs are made with 180nm Globalfoundries process. Core is running at 1.8V with internal vreg not accessible from the outside. Those can be powered by up to 3.6V. USB capable chips have another up to 5.5V -> 3.3V vreg, up to 100mA. Need external decoupling and can power external load. Those also have 5V tolerant I/O.

[nctnico]

That argument is a bit far fetched because you'll need additional logic to communicate as well. At 1k volume the solution costs over $30 in parts alone because you'll need bypass capacitors as well (and not taking the cost of pick&place time into account). A larger microcontroller which can control 100 individual LEDs (using 20 I/O pins for a 10x10 matrix) is likely to be cheaper.

Your problem right here is that if you don't see how you'd be utilizing these parts, you assume no one else could work it out, either; it's beyond your imagination.

This is a fallacy, and it's not "far-fetched" at all; just engineering, sometimes a specific solution happens to work great even if it's unthinkable to many or even most engineers such as yourself. It may still not be far-fetched for others, with different ways to do their design work.

First, you don't necessarily need "additional logic to communicate". These microcontrollers have - gasp! - IO in them. Communicating is well possible with zero extra components; you'd bit-bang anything you need, most likely something simple such as UART, but doing I2C could allow you interfacing with some other existing system. Or maybe you need to go through a simple resistor; or even AC coupled / level shifted through a $0.001 (add another $0.001 for P&P and yet another $0.001 for PCB real estate) capacitor with Manchester encoding. There are countless options if you need them and have your mind open.

Now you are throwing oodles of engineering time at a problem which doesn't have to exist. A big problem with having microcontrollers communicating which eachother is that it will go wrong every now and then. More microcontrollers means more errors which in turns means you might need additional components for filtering in order to make the solution to work reliable enough so the customer doesn't notice it.

[coppice]

Interesting device, Its built in a 180nm process, so the core is running way below the 5.5V maximum specified Vcc. That means it must have an internal regulator (which is a die hog), and its managing to operate that with no external pin for a cap.

Thick gate oxide makes 180nm 5V capable. I wouldn't be surprised if the Far East has some unique process variants that aren't found elsewhere.

High voltage transistors are a lot bigger than the standard transistors for a process. Using them for the whole device, and running everything at up to 5V, would really bulk up the die area. A regulator to run the core at something like 1.8V, while the periphery of the chip runs at at to 5V, would be the normal way to approach such a device.

[coppice]

I don't get the confusion.
A decent known brand uC with similar (or better!) specs is around 20-25 cents a piece.
So on your whole design you save about 15-20 cents.
That isn't even a significant number when only looking at the PCB parts/connectors.

Plus for those extra 20 cents you know you have at least a decent customer service, readable datasheets and you don't have to learn Mandarin when having questions about certain functionalities.

It depends on how you want to safe every penny, but for those quantities my customers can get those 20 cents for free for having a more reliable product.

A well known brand of MCU which is functionally similar to the Padauk device is not 20 cents in volume, although you won't normally get small quantities for a low price. That is what makes Padauk's marketing approach distinctive.

The kind of products these Padauk MCUs target don't have 20 cents slack in the price. The budget for the entire electronics in something like a musical greeting card or a small toy is only 10 to 20 cents, and the profit margin is only a cent or two.

[KL27x]

With micro this cheap, the chip could be the product, itself.

In days of old, most of the IC you buy might be completely analog.  In modern times, with extreme low quiescent current sleep micros, did you ever wonder if some of these chips are a very cheap micro that are flashed to perform as X/Y/Z? Buck converter, for instance. You need a decent ADC and maybe ~1Mhz execution to make a decent buck converter. Li ion management IC? Well, that would need an external transistor. But what about a voltage detector IC? There are probably better examples, but you get the idea.

I hope Dave manages to do a project video with this device. Just blink a LED in assembly, and it will be a great video.

[coppice]

In days of old, most of the IC you buy might be completely analog.  In modern times, with extreme low quiescent current sleep micros, did you ever wonder if some of these chips are a very cheap micro that are flashed to perform as X/Y/Z? Buck converter, for instance. You need a decent ADC and maybe ~1Mhz execution to make a decent buck converter. Li ion management IC? Well, that would need an external transistor. But what about a voltage detector IC? There are probably better examples, but you get the idea.  LED in assembly, and it will be a great video.

There are MCU cores embedded in all sorts of devices these days. Li-ion management chips are a good example. They may use a standard die, or a customised one, with special purpose things like pass transistors on the die.

[glarsson]

With micro this cheap, the chip could be the product, itself.

One well known product that is just a programmed microcontroller is the OBD II protocol decoder ELM327.

[b_force]

I don't get the confusion.
A decent known brand uC with similar (or better!) specs is around 20-25 cents a piece.
So on your whole design you save about 15-20 cents.
That isn't even a significant number when only looking at the PCB parts/connectors.

Plus for those extra 20 cents you know you have at least a decent customer service, readable datasheets and you don't have to learn Mandarin when having questions about certain functionalities.

It depends on how you want to safe every penny, but for those quantities my customers can get those 20 cents for free for having a more reliable product.

A well known brand of MCU which is functionally similar to the Padauk device is not 20 cents in volume, although you won't normally get small quantities for a low price. That is what makes Padauk's marketing approach distinctive.

The kind of products these Padauk MCUs target don't have 20 cents slack in the price. The budget for the entire electronics in something like a musical greeting card or a small toy is only 10 to 20 cents, and the profit margin is only a cent or two.

What part of saving every penny is to difficult to understand?
I literally just told this.

[PCB.Wiz]

With micro this cheap, the chip could be the product, itself.

In days of old, most of the IC you buy might be completely analog.  In modern times, with extreme low quiescent current sleep micros, did you ever wonder if some of these chips are a very cheap micro that are flashed to perform as X/Y/Z? Buck converter, for instance. You need a decent ADC and maybe ~1Mhz execution to make a decent buck converter. Li ion management IC? Well, those examples would need an external transistor. But what about a voltage detector IC?

That's part of what drives the volumes of MCUs.  They are now cheaper than many dedicated parts.

Look at Voltage Monitors, Temperature Sensors, and even the price of i2c ADC and DAC, and you can find a MCU can make solid sense replacing even a dedicated part.

eg SiLabs MCUs with 12b DACs are cheaper than any 12b DACs listed on Digikey.
That's a single peripheral function, that is cost competitive, so the ADC comes for free, as does the UART/SPI and of course the 50 MIPS MCU too !

[b_force]

With micro this cheap, the chip could be the product, itself.

In days of old, most of the IC you buy might be completely analog.  In modern times, with extreme low quiescent current sleep micros, did you ever wonder if some of these chips are a very cheap micro that are flashed to perform as X/Y/Z? Buck converter, for instance. You need a decent ADC and maybe ~1Mhz execution to make a decent buck converter. Li ion management IC? Well, those examples would need an external transistor. But what about a voltage detector IC?

That's part of what drives the volumes of MCUs.  They are now cheaper than many dedicated parts.

Look at Voltage Monitors, Temperature Sensors, and even the price of i2c ADC and DAC, and you can find a MCU can make solid sense replacing even a dedicated part.

eg SiLabs MCUs with 12b DACs are cheaper than any 12b DACs listed on Digikey.
That's a single peripheral function, that is cost competitive, so the ADC comes for free, as does the UART/SPI and of course the 50 MIPS MCU too !

Well, to be honest, I still use as much "analog" electronics if I can.
The real estate on your board isn't that much different, but it saves the time in developing and programming the microcontroller as well as programming it in production.
(with all errors and issues that come with it).
In the end that actually saves quite some money as well as less issues in production and a more reliable product.

I am not saying this is always the case btw.

[ali_asadzadeh]

There are 10 kinds of people, those who does not want to bother this padauk bullshit MCU and those who wants to touch the clean with their dirty hands(me included), so if you don't want to do it, do not play it at home!

[nctnico]

Look at Voltage Monitors, Temperature Sensors, and even the price of i2c ADC and DAC, and you can find a MCU can make solid sense replacing even a dedicated part.

eg SiLabs MCUs with 12b DACs are cheaper than any 12b DACs listed on Digikey.

Only if you don't care about analog performance. '12 bit' stamped on an ADC or DAC doesn't mean it really is 12 bit. On the cheaper parts it usually isn't.

[b_force]

Look at Voltage Monitors, Temperature Sensors, and even the price of i2c ADC and DAC, and you can find a MCU can make solid sense replacing even a dedicated part.

eg SiLabs MCUs with 12b DACs are cheaper than any 12b DACs listed on Digikey.

Only if you don't care about analog performance. '12 bit' stamped on an ADC or DAC doesn't mean it really is 12 bit. On the cheaper parts it usually isn't.

Correct, a lot of times the error in bits is pretty bad.
I have seen up to +/- 5-10 LSB. Even for external ones.
Not to talk about the drift.
You can get better performance with it to program self calibrating routines, but not many people think about it.

[mikeselectricstuff]

I got a few of the 14-pin ones (PM154C) as the 8 pin had sold out - x-ray shows die size approx 0.7x0.5mm

[wraper]

Program Memory Size... 1KW... right.

I did this project in under 1K

the 2nd time misunderstood so let me clarify, i wasnt complaining about the size, i was complaining the SI unit used

FYI KW is not SI unit, SI unit would be kW

[Rasz]

With micro this cheap, the chip could be the product, itself.

One well known product that is just a programmed microcontroller is the OBD II protocol decoder ELM327.

this is  a business model of Nanjing Qin Heng Electronics http://www.wch.cn/ . They make cheap microcontrollers, PCI interface chips, USB interface chips(universal, HID, USB Storage etc), serial interface... you get the picture most of their devices are the same die microcontroller with custom tailored firmware.

[bson]

The CH551 looks really interesting.  There is a dev board:
https://www.electrodragon.com/product/ch551-mini-dev-board-ch55x-series/

With a lot of demo code on github:
https://github.com/Edragon/WCH

Maybe run the demo code through google translate to decipher the comments.

This is really cool... I ordered one of those boards.  ETA 12-50 days.   China post.
Going to see if I can get something going with sdcc; they seem to use Keil.  My guess is it'll take ten minutes or less to make it compile with sdcc and gmake.

They claim to sell the IC for $.25, but I couldn't find it on their site or I would have ordered some at the same time.

Edit: $.21 qty 10 from LCSC.  Ordered 25.  Awesome! 

[brucehoult]

This is strange. Looking at the page now, they seem to have slipped another decimal place!  Now 0.47c each for 10, down to 0.3c each for 5000.

Shirley shome mishtake?

https://lcsc.com/product-detail/PADAUK_PADAUK-Tech-PMS150C_C129127.html

[mikeselectricstuff]

This is strange. Looking at the page now, they seem to have slipped another decimal place!  Now 0.47c each for 10, down to 0.3c each for 5000.

Shirley shome mishtake?

https://lcsc.com/product-detail/PADAUK_PADAUK-Tech-PMS150C_C129127.html

Seems to be fixed now