IOT Power Supply Design

[Kean]

and be clever

"Clever" programmers drive me nuts!  I'm sure we all have a story...   

So if someone is asking/discussing in general terms how to do something, then it is usually best to give a simple option.  At least until more details come out... like "I do not have any micro on the board."
If they have enough experience to implement a more complex solution then they're unlikely to be asking.  In this case, handling I2C in timer interrupts won't help with I2C's limited bandwidth.

[JxR]

Using a slave micro is a good idea. May be I can have the master - slave communication over I2C and the slave MCU->Display over SPI.
Thanks for the input. That should work. All I need to implement is a command parser for the slave MCU.

No it is not. Using two MCUs is a recipy for dissaster. Use one MCU and be clever about how the tasks are divided. A simple way to make sure a process gets enough time is to run it from a timer interrupt.

Better not open any test equipment if you don't want to see disaster in production.  Having a separate micro run the display section is a fairly common.

[nctnico]

I'm not saying it can't be done but it adds more complexity like needing a protocol and dealing with the situation the communication is lost (or impaired).

A more simpler solution would be to have two I2C busses and use interrupts and/or DMA to transfer the data so the CPU doesn't have to wait for the transfers to finish. Yes, this can also be done using bit-banging and timer interrupts.

[Kean]

I'm not saying it can't be done but it adds more complexity (like needing a protocol).

Defining that protocol layer also allows a clean point to split the product development time between people/teams - much like modern application development splits UI vs application logic.

This can* also make a better product - e.g. more flexibility, easier repair/upgrades, common UX across product range, longer product life.
*can, but doesn't always...

[mariush]

What's your budget?

10$ 40mm x 35mm 8x2 parallel
 https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-FL-GBW/NHD-0208AZ-FL-GBW-ND/1701134
https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-FSW-GBW-33V3/NHD-0208AZ-FSW-GBW-33V3-ND/2773587
https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-RN-YBW/NHD-0208AZ-RN-YBW-ND/1701132
https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-FL-YBW/NHD-0208AZ-FL-YBW-ND/1701133
https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-RN-GBW/NHD-0208AZ-RN-GBW-ND/2165845
https://www.digikey.com/product-detail/en/newhaven-display-intl/NHD-0208AZ-RN-YBW-33V/NHD-0208AZ-RN-YBW-33V-ND/2773586

If you use these in 4bit mode, you need 6..7 wires to "talk" to one. IF you make it one way only (if you don't have to poll to see if it's busy or you don't need to read data from it), a simple shift register would solve your problem. Shift 8 bits, send them to display (4 data, rs, rw, enable), repeat.
You can solder the lcd contrast resistor(s)/potentiometer directly on the lcd display board.
 

If you can stretch that to ~ 53 mm width , you could use 8x1 14 segment LCDs :

~4.3$ each on Digikey, 2.5$ if you get 100, but you could probably source them from lcdc or other places, it's probably standard format.

reflective, 3.3v : https://www.digikey.com/product-detail/en/varitronix/VIM-878-DP-RC-S-LV/153-1115-ND/1118605
transflective 3.3v .. 4.6v : https://www.digikey.com/product-detail/en/varitronix/VIM-878-DP-FC-S-LV/153-1113-ND/1118603

reflective, hv version (up to 7.7v) : https://www.digikey.com/product-detail/en/varitronix/VIM-878-DP-RC-S-HV/153-1114-ND/1118604

Cheap in volume, super low power, if you absolutely need backlight a simple led can be added on the sides.
Use a microcontroller with built in lcd controller, or just use an actual lcd controller, there are some which have i2c or spi

Yes, it's only 8 digits, but you can print voltage (and being 14segment you can also print a v), wait a couple seconds, print current, blink a short error message ex err. 100 , or statuses like "uSb Conn" etc

The above are 36 segments, 4 commons...

Example of lcd driver that I think would work  just fine:

36segment / 4common, ~ 1$ each at Digikey.

i2c PCF8551ATT: https://www.digikey.com/product-detail/en/nxp-usa-inc/PCF8551ATT-AJ/568-13181-1-ND/6576096
spi PCF8551BTT-AJ https://www.digikey.com/product-detail/en/nxp-usa-inc/PCF8551BTT-AJ/568-13772-1-ND/8041379

40segment / 4 common (can just leave 4 unused)

i2c/spi PCF8553DTT-AJ https://www.digikey.com/product-detail/en/nxp-usa-inc/PCF8553DTT-AJ/568-14373-1-ND/9449822
i2c PCF8576DT-2118 https://www.digikey.com/product-detail/en/nxp-usa-inc/PCF8576DT-2118/568-3558-1-ND/1144586

You can order some custom PCBs from jlpcb or whatever, place the chip and the 4..6 pin header on one side, and leave the other side matte (so you can glue an aluminum foil or some reflective sticker.
Optionally, add some right angle surface mount leds under the lcd digits for backlight purposes

You can probably find lcd drivers at lcsc for half a dollar or less, that would work just well.

[EEEnthusiast]

Thanks for all the valuable inputs.

More than the budget, I have a size limitation with the current case. My display size is restricted to 40x25mm or smaller. Budget can be flexible, but $10 for a display is overkill. I would like anything < 5$ .
But Digikey prices are overrated. I may be able to get those at 1/2 the price from some other distributors.

I am wondering if they make LCD screens like this with I2C interface.
https://jingshimei.en.alibaba.com/product/62019597651-810314584/Custom_LCD_screen_minimum_size_segment_code_Positive_FSTN_LCD_screen.html?spm=a2700.icbuShop.41413.23.6dc164f0Cm72P1

Those should be very easy to use and run off really low power.

[mariush]

4.8$ @ 1 , 3.5$ @ 25 : 30mm x 11.5m  128 x 32 px OLED white text: https://www.tme.eu/en/details/rex012832dwpp3n0/graphical-oled-displays/raystar-optronics/rex012832dwpp3n00000/
same series in
yellow : https://www.tme.eu/en/details/rex012832dypp3n0/graphical-oled-displays/raystar-optronics/rex012832dypp3n00000/
sky blue : https://www.tme.eu/en/details/rex012832dspp3n0/graphical-oled-displays/raystar-optronics/rex012832dspp3n00000/


A variation of the above series , (F instead of D), and the ribbon cable is slightly different,  but several colors available each in lower stock amount:
 
yellow https://www.tme.eu/en/details/rex012832fypp3n0/graphical-oled-displays/raystar-optronics/rex012832fypp3n00000/
sky blue https://www.tme.eu/en/details/rex012832fspp3n0/graphical-oled-displays/raystar-optronics/rex012832fspp3n00000/
white https://www.tme.eu/en/details/rex012832fwpp3n0/graphical-oled-displays/raystar-optronics/rex012832fwpp3n00000/


Both use same controller, and serial communication

lcd (no controller, plain glass)

3.1$ @ 1 , 2.6$ @ 10,  2.12@42: https://www.tme.eu/en/details/de124-rs-20_7.5-3/lcd-digital-displays/display-elektronik/de-124-rs-20-7-5-3-volt/
seven segment 8 digits
Dimensions    37.99x20.29x1.1mm    
Window dimensions (H x W)    34.1x11.63mm

 $ 3.60 @ 10 ,  $ 2.80 @ 10 : https://www.tme.eu/en/details/de188-ru-30_7.5-3/lcd-digital-displays/display-elektronik/de-188-ru-30-7-5-v-3-volt/
seven segment 8 digits
Dimensions    34.9x13x1.1mm    
Window dimensions (H x W)    31.9x6mm

[EEEnthusiast]

Thanks for all the inputs. I was able to get the I2C based OLED display to work with the FTDI with the whole refresh of the screen taking well below 100ms.  That's good enough to refresh once every 1 second.

Some more updates on the design. Got my first prototype built and the performance looks within expectations. Will post some results soon. If anyone volunteers to test an initial build, I would be happy to send it across at shipping costs. Please message me. Limited to just 2 units now.

To summarize what it is.
It is a USB based power supply, power profiler and energy meter to test IOT products (For developers). Similar to Otii Arc but with enhanced performance.

> It can provide a programmable output voltage between 0 to 6V to the product under test. Programmable in 5mV steps.
> Output current limited to 2A. Output short circuit protection.
> The current measurement range extends from 100nA to 1A in both directions. The expected accuracy is better than 1% +/- 100nA across the range.
> Sampling rate for the current is 1Msps with a rise time of < 2uS (10-90%).
> Recording length of few hours. Tested for at least 24 hours without any errors.
> OLED display to indicate output voltage, average current and other status information.
> Digital logic capture in sync with the current.(within 1us)
> USB 2.0 B type jack.
> Powered by external adapter 15V, 3A.

[SWR]

Hi EEEnthusiast

This is propably a bit late reply, but have you considered using a two dollar ESP32 Wroom module?

It has built in WiFi and BLE including the antenna.
It's got two 12 bit ADC's with 18 analog input channels.
Regarding processing power I don't know your algorithm, but it's got three 32bit cores:
- One is a slow low power core for simple house keeping including ADC sampling and simple I/O.
- The two others are 240MHz cores with individual floating point units and zero overhead looping & subroutine preserve/restore (rotating register window).
      One of them could be used for WiFi and other communication while the other could do the compression?

I don't know if it has the power to do your compression algorithm?, but it does pack a serious punch compared to other two dollar micros and it would provide isolation.
I'm interested in purchasing a product like this.

Best regards
Soren

[EEEnthusiast]

Soren
Thanks for the inputs..
I did not consider a Wireless link as it does not guarantee a sustained throughput.
Writing embedded code is much more difficult than writing it on a desktop, with its large computing power. So I just stick with the signal processing on the PC for now.

The hardware will just do the data capture and stream it in raw format to the PC.

I am planning to launch it on CrowdSupply
https://www.crowdsupply.com/zscircuits/zs1100a-power-meter

The campaign is not yet launched. Should be live in a couple of days...

[EEEnthusiast]

Fellow Geeks,
The CrowdSupply campaign is live now.
https://www.crowdsupply.com/zscircuits/zs1100a-power-meter

[EEEnthusiast]

Just an update....
The IOT Power meter is in production now and meets most of the design goals.
I thank you for all your inputs and feedback.

The hardware is open source and the design files are available in the link below
https://github.com/zscircuits/zs1100a

Any review or criticism is most welcome..

[Kean]

meets most of the design goals

Can you clarify this statement a little?

I'm not too concerned by it not being perfect, as it would never get finished - but it is important to know of any major shortcomings compared to the published specs.
I already have a Joulescope, so I expect there will be pros/cons to each.

Looking forward to getting my unit once Crowd Supply get their act together.  Thanks for publishing the docs etc on github, I will take a detailed look later.

[EEEnthusiast]

The deviations are from the specs I had posted on EEVBlog. The ones on the CrowdSupply page are being met

The few deviations are
1. Output voltage is programmable in 10mV steps and not 5mV steps.
2. Current measurement accuracy is now 1% +/- 0.2uA as against my original target of 1% +/- 0.1uA


Some enhancements added are
1. Programmable current limiting
2. Programmable output resistance
3. Over voltage protection and reverse current limiting