Author Topic: EEVblog #1242 - Memory LCD+Supercaps+Low Power Design  (Read 2320 times)

0 Members and 1 Guest are viewing this topic.

Online nctnico

  • Super Contributor
  • ***
  • Posts: 17890
  • Country: nl
    • NCT Developments
Re: EEVblog #1242 - Memory LCD+Supercaps+Low Power Design
« Reply #50 on: September 08, 2019, 07:52:48 pm »
The current consumption likely goes up with temperature as well so that is another factor.
Sure, but not by an order of magnitude.

I imagine the datasheet was written for the first production samples and they've improved the process since then without changing the datasheet (in case there's any batches of old chips in a warehouse somewhere).
Total nonsense. You can't ignore the numbers in the datasheet. Actually; these may even be too optimistic. In a large batch you may find devices which barely meet their specifications. Remember that chips are made using a lithographic process which has inherent errors. The datasheet reflects the margins of these errors. It is foolish to believe manufacturers are putting wider margins in their datasheets than absolutely necessary.
« Last Edit: September 08, 2019, 07:54:37 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Fungus

  • Super Contributor
  • ***
  • Posts: 10122
  • Country: 00
Re: EEVblog #1242 - Memory LCD+Supercaps+Low Power Design
« Reply #51 on: September 09, 2019, 01:58:57 am »
I imagine the datasheet was written for the first production samples and they've improved the process since then without changing the datasheet (in case there's any batches of old chips in a warehouse somewhere).
Total nonsense. You can't ignore the numbers in the datasheet.

I never said you could.

(At least: Not unless it's a small production run and you hand-select and test the individual chips  :) )
 

Online Howardlong

  • Super Contributor
  • ***
  • Posts: 4756
  • Country: gb
Re: EEVblog #1242 - Memory LCD+Supercaps+Low Power Design
« Reply #52 on: September 09, 2019, 12:59:15 pm »
As usual, you should never trust the datasheet when it comes to low power current draw specs: sometimes it's better, sometimes it's worse, and often the difference is dramatic.

The typical sleep current for the PIC16LF1904 is listed as 160nA @ 3V, I was measuring only 17nA. The same applied to the PIC16LF1934 where the datasheet quotes 80nA @ 3V, and I achieved 16nA right off the bat.
Well your unit used 17nA but one from the next batch may use what the datasheet says, so you still need to expect and design for the worst case in datasheet figures to still result in a product working to the expected spec. You'll maybe (or rather likely) get lucky, but it would be annoying to expect to get lucky and get bitten manufacturing 1000 units and that reel of chips happens to be on the datasheet's max values...
I agree. The numbers from the datasheet should be used for anything that needs to go in production. The current consumption likely goes up with temperature as well so that is another factor.

In general, I agree, but I'll caveat that with a couple of things if I may.

Firstly, this is making the assumption that the figures you require for your use case are there in the data sheet. For example, LCD static mode consumption just doesn't appear in most PIC data sheets that have an LCD controller, but they do sometimes include some figures for the multiplexed modes that require power hungry charge pumps and bias networks. Worse still, even from the same manufacturer, it's often difficult to compare like with like, because consumption figures aren't specified under the same criteria. Sometimes you will just have to characterise those figures yourself.

Secondly, in most circumstances I'd agree that the datasheet (and any errata) should be the go-to place for definitive figures. Doing things like ignoring timing limitations and sticking 5V on a non 5V-tolerant pin is asking for trouble. Specifications such as current consumption are often specified as "typical" and it's very rare, if ever, to see a distribution. Sometimes there is a maximum quoted for a given temperature and Vdd: whether you choose to use this or not is more subjective, and depends on the context and project requirements. Irrespective, you'd be nuts not to measure and characterise yourself. Yes, temperature is a big delta on static current and needs to be taken into account based on the project requirements. It's not even uncommon to see disclaimers on datasheets on current consumpmtion figures like "These parameters are for design guidance only and are not tested". In such circumstances, your only option is to characterise these figures as best you can yourself.

In particular, in my experience of Microchip parts, which is where by far most of my experience lies, I find quoted current consumption specifications at the low end are sometimes significantly different to those specified. Sometimes it's worse, sometimes it's better. Sometimes they're spot on. Sometimes the figures are just so out of kilter, you know something's not right. It's not uncommon to spend hours trying to work out what tweak I need to make to get the quoted figures, and sometimes it just doesn't work out.

FWIW, the sub 20nA@3V for PIC16LF1xxx devices in "all-off" sleep mode is quite common (although note, it's not universal) across the range IME (e.g., see https://www.eevblog.com/forum/microcontrollers/pic-power-consumption/msg2113987/#msg2113987). Keep in mind that ultra low power on PICs is one of my "skills" shall we say, I have a fair bit of experience across the range, but equally it's a big subject with plenty of detail and nuance, and I accept you can never have never enough experience.

What I can say is that I encountered similar measurements on PIC16LF1906 and '1907 devices, the same family, but likely from different batches and wafers. The same applied to the '193x family. Not a great sample, I agree, but enough for me to be confident that I didn't just get lucky. For the '1917x series, where claimed sleep figures at 3V were miles out, I only tried two devices, almost certainly from the same batch, but also something with the way there was a dramatic increase in current as I turned up the volume from 1.8V to 3.0V. The jury's still out on that one, it could easily be an operator error!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf