4-bit MCU Availability - Again...

[Ed Lee]

On Sunday, June 12, 2022 at 5:56:13 PM UTC-7, Ricky wrote:

On Sunday, June 12, 2022 at 6:58:57 PM UTC-4, bitrex wrote:
On 6/12/2022 6:33 PM, Three Jeeps wrote:
On Saturday, June 11, 2022 at 2:12:10 PM UTC-4, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209
I personally don\'t see the need for a 4 bit cpu. I see them as costing approximately the same as 8 bit cpus because they are a \'specialty\' device. My past designs were with the TI MSP430 or the microchip. you may want to check out stmicroelectronics - STM32g031 @ $1.80 USD in 100\'s is pretty cheap. Suppose you find a 4 bit @ <$1.00 USD, I suspect the additional cost of tools and time to program will easily offset the cost.
I don\'t know all your requirements so pls take these suggestions as just that. You may have already checked them out.
J

I don\'t see much power-saving advantage to a 4 bit CPU, either, modern 8
bit microcontrollers can be put into sub-uA sleep waiting on a pin
change interrupt.
Can they be put to sleep waiting for any key on the keypad to be pressed? I\'ve just never seen a device that had that flexibility.

Yes, that\'s the point of pin change interrupt. As long as you have the trigger pins (if not all) on the rows or columns of a matrix, it will wake up on any key. This is a very common feature on almost all micros: msp430, avr, pic, etc.

> I was actually hoping to hear from some people who design products that sell millions and only have a need for a very limited processor, like a 4 bit device. There was one guy in particular who used to post here and designed toys. He talked about design reviews removing a resistor they felt was not important enough to retain because of the cost.

As you said before, the price difference between 4 bits and 8 bits will be overwhelmed by setup, programming, testing and packaging costs. Unless you are ready to commit to 1M OTP parts now, you have to worry about them still available when you are ready for volume. When customer says they will buy hundreds of thousand, we usually go with 10k batch to start.

 

[John S]

On 6/12/2022 9:26 AM, Ed Lee wrote:

On Sunday, June 12, 2022 at 6:35:24 AM UTC-7, Ricky wrote:
On Sunday, June 12, 2022 at 12:45:53 AM UTC-4, Dave Platt wrote:
In article <e71be537-a2eb-4eed...@googlegroups.com>,
Ricky <gnuarm.del...@gmail.com> wrote:

The MCU has to run all the time to \"see\" that you pressed a button. It
doesn\'t have to run 100% of the time to do that, but it\'s a far cry from
only when you press the button. It\'s not like the MCU is just turned
off and you press a button to turn on the remote, then press the button
you want to send.
I suppose they could diode OR the buttons to the MCU power connection,
so the MCU is off until a button is pressed. Do they do that? I guess
they want to save the cost of the diodes, so instead they use a 4-bit
MCU that draws very, very little current when running.
Some deep-low-power MCUs these days have specialized logic for this
purpose. There\'s a static (non-clocked) pin-state-change detection
circuit for one or more of the pins, which will operate even if the
core of the MCU is clock-stopped. In a typical implementation,
this circuit can trigger an interrupt which brings the MCU out of
a clock-stopped idle state (with some amount of latency as the
clock oscillator starts up again).

This gets the power usage down to \"static CMOS\" levels of leakage.

Fancier chips will have multiple power wells on the die, with the
ability to turn off power to non-essential peripherals and even
to the core itself. In those designs, only the pin-change
circuitry may actually be powered up when the MCU is standing
by - the core may be powered down.
On many remote controls, there are enough buttons they need to be multiplexed. I suppose it is possible to connect every I/O pin to a change detector.

You can still multiplex on pin change interrupt. The Holtek chip has 18 to 26 pins to multiplex.

\"Fancier chips\" is not what this is about. It is about the absolute lowest cost product. Even the pin state change detector might price the device above the competition. Not only is there die area costs, but tiny chips like these are dominated by the time spent on the tester.

So, not much difference between 4 bits or 8 bits. You are dealing with much smaller segment in 4 bit manufacturers, and they usually want more profit.

This is the sort of situation where you can devise all sorts of fancy fasteners that have many advantages, but a nail is going to win the contest every time.

Keeping the entire chip in zero power state other than a low rate RC oscillator to wake up the device is going to win this contest every time. For the remote control, it has to wake up maybe 5 times a second to prevent perceptible lag. The CPU probable runs at a similarly slow clock rate, then shuts off. This is going on 24/7, so needs to be adequately low.

The Holtek chip can run on timer interrupt at 10uA for 10 years. Spec-ing battery for more than 10 years is just meaningless.

Dunno about that. 10 years at 10uA is about 876mA-hours.

 

[Ed Lee]

On Monday, June 13, 2022 at 1:31:23 PM UTC-7, John S wrote:

On 6/12/2022 9:26 AM, Ed Lee wrote:
On Sunday, June 12, 2022 at 6:35:24 AM UTC-7, Ricky wrote:
On Sunday, June 12, 2022 at 12:45:53 AM UTC-4, Dave Platt wrote:
In article <e71be537-a2eb-4eed...@googlegroups.com>,
Ricky <gnuarm.del...@gmail.com> wrote:

The MCU has to run all the time to \"see\" that you pressed a button. It
doesn\'t have to run 100% of the time to do that, but it\'s a far cry from
only when you press the button. It\'s not like the MCU is just turned
off and you press a button to turn on the remote, then press the button
you want to send.
I suppose they could diode OR the buttons to the MCU power connection,
so the MCU is off until a button is pressed. Do they do that? I guess
they want to save the cost of the diodes, so instead they use a 4-bit
MCU that draws very, very little current when running.
Some deep-low-power MCUs these days have specialized logic for this
purpose. There\'s a static (non-clocked) pin-state-change detection
circuit for one or more of the pins, which will operate even if the
core of the MCU is clock-stopped. In a typical implementation,
this circuit can trigger an interrupt which brings the MCU out of
a clock-stopped idle state (with some amount of latency as the
clock oscillator starts up again).

This gets the power usage down to \"static CMOS\" levels of leakage.

Fancier chips will have multiple power wells on the die, with the
ability to turn off power to non-essential peripherals and even
to the core itself. In those designs, only the pin-change
circuitry may actually be powered up when the MCU is standing
by - the core may be powered down.
On many remote controls, there are enough buttons they need to be multiplexed. I suppose it is possible to connect every I/O pin to a change detector.

You can still multiplex on pin change interrupt. The Holtek chip has 18 to 26 pins to multiplex.

\"Fancier chips\" is not what this is about. It is about the absolute lowest cost product. Even the pin state change detector might price the device above the competition. Not only is there die area costs, but tiny chips like these are dominated by the time spent on the tester.

So, not much difference between 4 bits or 8 bits. You are dealing with much smaller segment in 4 bit manufacturers, and they usually want more profit.

This is the sort of situation where you can devise all sorts of fancy fasteners that have many advantages, but a nail is going to win the contest every time.

Keeping the entire chip in zero power state other than a low rate RC oscillator to wake up the device is going to win this contest every time. For the remote control, it has to wake up maybe 5 times a second to prevent perceptible lag. The CPU probable runs at a similarly slow clock rate, then shuts off. This is going on 24/7, so needs to be adequately low.

The Holtek chip can run on timer interrupt at 10uA for 10 years. Spec-ing battery for more than 10 years is just meaningless.
Dunno about that. 10 years at 10uA is about 876mA-hours.

Typical alkaline AAA battery is 860 to 1,200mAhr, but you might need two or three for the voltage. Rechargeable 18650 is typically 2200mAhr at 4V, but it might be too thick for remote.

 

[Martin Brown]

On 13/06/2022 21:31, John S wrote:

On 6/12/2022 9:26 AM, Ed Lee wrote:
On Sunday, June 12, 2022 at 6:35:24 AM UTC-7, Ricky wrote:

Keeping the entire chip in zero power state other than a low rate RC
oscillator to wake up the device is going to win this contest every
time. For the remote control, it has to wake up maybe 5 times a
second to prevent perceptible lag. The CPU probable runs at a
similarly slow clock rate, then shuts off. This is going on 24/7, so
needs to be adequately low.

This is nonsense - most low power MCUs today have a sleep mode
(sometimes a deep sleep mode too) and wake up on input pin level change
interrupt. Even on a cheap low power watch xtal at 32kHz the time lag
for a button press is imperceptible to a human.

The Holtek chip can run on timer interrupt at 10uA for 10 years.
Spec-ing battery for more than 10 years is just meaningless.

Dunno about that. 10 years at 10uA is about 876mA-hours.

10 years is broadly comparable with the shelf life of most commonly
available decent quality batteries. Working life may sometimes be
slightly extended by a very small current flowing in an external load.
The only snag is that some brands (Duracell I\'m looking at you) corrode
on a timescale that is shorter than their working life at very low
currents. My favoured batteries these days are Panasonic and Eveready.

You can get various exotic chemistry batteries for low currents with a
~40+ year working life for specialist applications (at a price).

https://embeddedcomputing.com/technology/analog-and-power/low-battery-self-discharge-the-key-to-long-life-remote-wireless-sensors

To go longer than that you are into thermal electric generators and
radioactive decay of eg Pu238 as still powers the Voyager probes nearly
50 years on since they were launched.

https://www.acs.org/content/acs/en/education/whatischemistry/landmarks/plutonium-238-production.html


Zamboni piles are another technology that can almost last forever as
long as you don\'t want them to do much work (well almost any). Mine is a
just post WWII era night vision PSU and is still going strong. The limit
of the work it can do is attract a piece of aluminium foil from one
terminal to the other (until the metal work hardens and snaps off). A
paupers version of this amazing physics demo running since 1840

https://en.wikipedia.org/wiki/Oxford_Electric_Bell

--
Regards,
Martin Brown

 

[Ricky]

On Sunday, June 12, 2022 at 9:50:57 PM UTC-4, Ed Lee wrote:

On Sunday, June 12, 2022 at 5:56:13 PM UTC-7, Ricky wrote:
On Sunday, June 12, 2022 at 6:58:57 PM UTC-4, bitrex wrote:
On 6/12/2022 6:33 PM, Three Jeeps wrote:
On Saturday, June 11, 2022 at 2:12:10 PM UTC-4, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209
I personally don\'t see the need for a 4 bit cpu. I see them as costing approximately the same as 8 bit cpus because they are a \'specialty\' device. My past designs were with the TI MSP430 or the microchip. you may want to check out stmicroelectronics - STM32g031 @ $1.80 USD in 100\'s is pretty cheap. Suppose you find a 4 bit @ <$1.00 USD, I suspect the additional cost of tools and time to program will easily offset the cost.
I don\'t know all your requirements so pls take these suggestions as just that. You may have already checked them out.
J

I don\'t see much power-saving advantage to a 4 bit CPU, either, modern 8
bit microcontrollers can be put into sub-uA sleep waiting on a pin
change interrupt.
Can they be put to sleep waiting for any key on the keypad to be pressed? I\'ve just never seen a device that had that flexibility.
Yes, that\'s the point of pin change interrupt. As long as you have the trigger pins (if not all) on the rows or columns of a matrix, it will wake up on any key. This is a very common feature on almost all micros: msp430, avr, pic, etc.

Oh, you mean the higher priced parts...

I was actually hoping to hear from some people who design products that sell millions and only have a need for a very limited processor, like a 4 bit device. There was one guy in particular who used to post here and designed toys. He talked about design reviews removing a resistor they felt was not important enough to retain because of the cost.
As you said before, the price difference between 4 bits and 8 bits will be overwhelmed by setup, programming, testing and packaging costs. Unless you are ready to commit to 1M OTP parts now, you have to worry about them still available when you are ready for volume. When customer says they will buy hundreds of thousand, we usually go with 10k batch to start.

Packaging? What packaging? At this quantity, they use chip on board. There\'s no programming. That\'s done at the mask level. This is why I was looking for people with specific experience with high volume, low cost devices..

--

Rick C.

--+ Get 1,000 miles of free Supercharging
--+ Tesla referral code - https://ts.la/richard11209

 

[David Eather]

On 12/06/2022 4:12 am, Ricky wrote:

So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

Dude, just use a low power 8 bit MCU. No appreciable difference in power
consumption.

 

[Ricky]

On Tuesday, June 14, 2022 at 8:39:42 PM UTC-4, David Eather wrote:

On 12/06/2022 4:12 am, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

Dude, just use a low power 8 bit MCU. No appreciable difference in power
consumption.

Dude, that isn\'t the question.

--

Rick C.

-+- Get 1,000 miles of free Supercharging
-+- Tesla referral code - https://ts.la/richard11209

 

[Martin Brown]

On 15/06/2022 05:21, Ricky wrote:

On Tuesday, June 14, 2022 at 8:39:42 PM UTC-4, David Eather wrote:
On 12/06/2022 4:12 am, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

Why do you expect that? They will probably tell you what you want to
know although they might well quote a price that indicates rather
clearly that they don\'t actually want your business.

Dude, just use a low power 8 bit MCU. No appreciable difference in power
consumption.

Dude, that isn\'t the question.

If you were likely to be one of their customers at the volumes they
operate then you would already know how to get in touch with them and
have done it. Equally they would likely already know about you if you
were operating at mass market volumes they consider sufficient.

You either have to knock on their door and convince them that you are
looking at serious mass market volume and so worth them talking to you
or simply do it some other way (8 bit MCU would be an obvious way).

--
Regards,
Martin Brown

 

[Ed Lee]

On Wednesday, June 15, 2022 at 12:18:34 AM UTC-7, Martin Brown wrote:

On 15/06/2022 05:21, Ricky wrote:
On Tuesday, June 14, 2022 at 8:39:42 PM UTC-4, David Eather wrote:
On 12/06/2022 4:12 am, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.
Why do you expect that? They will probably tell you what you want to
know although they might well quote a price that indicates rather
clearly that they don\'t actually want your business.
Dude, just use a low power 8 bit MCU. No appreciable difference in power
consumption.

Dude, that isn\'t the question.
If you were likely to be one of their customers at the volumes they
operate then you would already know how to get in touch with them and
have done it. Equally they would likely already know about you if you
were operating at mass market volumes they consider sufficient.

You either have to knock on their door and convince them that you are
looking at serious mass market volume and so worth them talking to you
or simply do it some other way (8 bit MCU would be an obvious way).

--
Regards,
Martin Brown

OP is obviuously new to this market/design segment. His customer is buying 1M masked-rom chip for unknown 4-bit mcu from unknown supplier. They probably already have the market data, but none of the design data, and hiring a new designer for the job. There are two possibilities:

1. Previous designer died with covid and took everything to his grave.

2. New hot shot manager trying to upend existing department without touching any of the existing supply chain.

 

[three_jeeps]

On Saturday, June 11, 2022 at 2:12:10 PM UTC-4, Ricky wrote:

So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

FWIW, TI recently announced a BT LE Arm Cortex-M0+ 32 bit CPU with price as low as $0.79 USD, CC2340. Standby current of < 700nA, 48MHz, and quite a number of peripherals and I/O. I\'d imagine with an Arm architecture, it would go a long way to future-proofing the product.
J

 

[Ricky]

On Wednesday, June 22, 2022 at 10:23:08 AM UTC-4, three_jeeps wrote:

On Saturday, June 11, 2022 at 2:12:10 PM UTC-4, Ricky wrote:
So anyone have knowledge of truly low cost/low power MCU devices? I found this company.

http://upt-ic.com/en/index.aspx

They have a line of 4-bit MCUs. Can\'t find a way to get prices unless I call them. I expect that would not be very effective.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209
FWIW, TI recently announced a BT LE Arm Cortex-M0+ 32 bit CPU with price as low as $0.79 USD, CC2340. Standby current of < 700nA, 48MHz, and quite a number of peripherals and I/O. I\'d imagine with an Arm architecture, it would go a long way to future-proofing the product.
J

The details of the CPU part of an MCU is actually the least important part of a chip when considering \"future proofing\". It\'s much harder to deal with the peripherals and other I/O when changing chips, as they are very brand dependent.

--

Rick C.

-++ Get 1,000 miles of free Supercharging
-++ Tesla referral code - https://ts.la/richard11209