Shop OBEX P1 Docs P2 Docs Learn Events
Looking for a recommendation on a low power display: Updated with results — Parallax Forums

Looking for a recommendation on a low power display: Updated with results

JohnR2010JohnR2010 Posts: 431
edited 2015-03-31 06:07 in General Discussion
To accompany my recent low power (1 year run time on 3 AAs) project I would like to add a display. It doesn't have to be color but I would like the ability to do graphics and I would like to keep the external components and required pins to a minimum. I have been looking at the Sharp Microelectronics PNLC Memory LCDs that operate on 3v @ 25uW http://www.mouser.com/new/Sharpsma/sharppnlclcds/. I know Ken was having a driver written for some ePaper displays but from what I can tell there is a lot of supporting hardware required to work with ePaper. I'm trying to keep the size of the project as small as I can.

Before I order one and spend the time creating a driver I thought I would ask for forum for input.

* * * * UPDATE TO POST FOLLOWS THIS LINE * * * * :

This week I was able to create a PCB and test the above SHARP Memory LCD.

Test Circuit: Test was with my test PCB (No Xbee installed) measuring current (at ~3.8v battery voltage (3AA rechargeable batteries)) into the LDO voltage regulator on the PCB. All unused Prop pins were set to output and high, only two cogs running main and a cog that toggled the LCD's VCOM. When measurement was made Main Cog was at WAITPNE and VCOM cog was sending SPI data to MLCD to toggle the mLCD bias. Board's LED was off.

Test Results:
RCSLOW = 45.14uA ambient lights On, 7.9uA ambient lights Off
RCFAST = 1.03mA (Didn't test with ambient lights off)

Observation; with my bench's fluorescent light on the power usage is around 45uA with the light off it jumps between 8uA and 28uA DC. This happens when running the circuit from USB power or battery power. With the room's overhead lights off the power usage dropped even more. It got as low as 7.8uA. I used the LED light on my iPhone and when it was shining on the LCD the power usage jumped back up so it doesn’t seem to be RF interference. It appears the power usage varies with ambient light.

attachment.php?attachmentid=113182
«1

Comments

  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-30 18:04
    JohnR2010 wrote: »
    To accompany my recent low power (1 year run time on 3 AAs) project I would like to add a display. It doesn't have to be color but I would like the ability to do graphics and I would like to keep the external components and required pins to a minimum. I have been looking at the Sharp Microelectronics PNLC Memory LCDs that operate on 3v @ 25uW http://www.mouser.com/new/Sharpsma/sharppnlclcds/. I know Ken was having a driver written for some ePaper displays but from what I can tell there is a lot of supporting hardware required to work with ePaper. I'm trying to keep the size of the project as small as I can.

    Before I order one and spend the time creating a driver I thought I would ask for forum for input.

    You have left out a lot of information you could have provided at first that would help shape a response. First off you say "3 AAs" but you don't say what type and capacity as they could be anything from Alkaline, Lithium, NiMh etc and all those are of varying quality too. secondly you don't state what the rest of the system is consuming and how often it wakes up etc. Also the other thing is the operating temperature range which will affect leakage and consumption.

    As for memory LCDs if they require some extra components then that's what they require if that is what it takes to get low power operation. However looking at all the documentation I can't see anything special required for it as it is basically an SPI interface and standard supply voltage etc although there is the low frequency VCOM clock that would probably need to be sent to it which adds to the power consumption. I don't know why they didn't allow for an internal oscillator as they have left this up to software or external hardware and certainly makes the power consumption figures look better for the display itself.

    The fact that you would "like" a graphics display and low-power just complicates things, work out what you do need and work from there.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-12-31 00:32
    in general, the back-lighting of an LCD consumes far more power than the actual liquid crystal manipulation. So that is the first feature I would do without.

    There are small LCD displays that are graphic and others that are character based. But the character based ones are simpler to interface via ASCII. Once you go over to graphics, you are going to have to create your own set of characters and provide data for each and every pixel on a refresh cycle that is faster than the eye can see (over 30 times per second).

    One year operation on 3 AA cells might only be feasible if your project is able to spend the majority of its time in a 'sleep mode'.

    Sharp does NOT explain the power consumption of the unit you mention. You would need to test it to determine that.
  • JohnR2010JohnR2010 Posts: 431
    edited 2014-12-31 04:53
    Back lighting will be left off not required. Please don't worry about my battery chemistry as I have all that worked out I was trying to say I currently have a propeller circuit and program that can achieve over a year run time on batteries and I would like to add a display to it. The prop does spend most of its time sleeping and is woke up by other very low power devices. Today it drives a 50ua panel meter to display a simple value. I would like to have the option of replacing the 50ua meter with a low power black and white display.

    The display I linked at the start of this tread will work just fine for what I'm wanting to do I just wanted to check and see if anyone has worked with newer even more efficient technology? Thanks!
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-12-31 05:08
    JohnR2010 wrote: »
    Back lighting will be left off not required. Please don't worry about my battery chemistry as I have all that worked out I was trying to say I currently have a propeller circuit and program that can achieve over a year run time on batteries and I would like to add a display to it. The prop does spend most of its time sleeping and is woke up by other very low power devices. Today it drives a 50ua panel meter to display a simple value. I would like to have the option of replacing the 50ua meter with a low power black and white display.

    The display I linked at the start of this tread will work just fine for what I'm wanting to do I just wanted to check and see if anyone has worked with newer even more efficient technology? Thanks!

    Chemistry? What's important here is to understand how the display's consumption relates to the rest of the system because there's no point going for a sub-microamp display if the rest of the system is drawing 50ua, is there? How often does the display need to be updated? Is epaper an option? Can the display just be turned on manually or automatically?

    Basically if you ask for some advice then be prepared to provide some detail and context, even if you "have it all worked out".
  • JohnR2010JohnR2010 Posts: 431
    edited 2014-12-31 06:15
    Thanks Peter for the help I do appreciate it. I'm simply looking for a recommendation on a black and white display that is low power for use in a battery based solution. You have to trust I have all the other details worked out. It is okay if you don't have a recommendation based on the information I supply.

    I'm not trying to offend or be disrespectful I have got a lot of information from your previous post so please understand this is just a quick post to ask for high level recommendations on a low power display if anyone has one.

    Thanks
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2014-12-31 21:33
    Well, you posted a link to a 3VDC display device and mention 3 AA batteries (that can be anywhere from 3.6VDC to over 4.5VDC).

    At some point, the inconvenience of using an LCD is that it requires its own regulated power source unless it falls in line with what the Propeller is using and/or can work within the whole range of the battery output voltage.

    Low power with a goal of battery operation for a year or more tends to demand a comprehensive design solution. To get the most out of batteries, no regulation would be optimal.

    A 3.0VDC primary Lithium cell might work better in this context, but you might desire the cheaper AA cells. One just can't limit their choices in that way.

    Just selection of the wrong linear voltage regulator, like an LM2940-3.3v might cause excessive battery drain from the quiescent current drainage.... 10ma typical. Ideally, avoiding all regulation eliminates a lot of wasted battery power.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2015-01-01 10:22
    John,
    I like COG (chip on glass) displays e.g. these from Newhaven. The power consumption is low by display standards, about 1mA in operation excluding the backlight. They are readable without the backlight in good ambient. Character displays with i2c interface are easy to use and could serve your purpose if you can get by with a bar-graph instead of a dial. I haven't tried the COG graphics displays. Those come with parallel or SPI serial interface, and the operating current is still less than 2mA.

    For long battery life, it still comes down to a mechanism to activate the display (and backlight) only when someone is there wanting to see it.
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-01-03 05:11
    Tracy, at the beginning of this thread I posted a link to a Sharp display and on page 14 of its spec-sheet states that it only uses 6uW to display a static frame and 12uW to update it. They are targeting this display to compete with ePaper. I like it because it seems to have much less supporting hardware than ePaper. Do you have any experience or thoughts about it? It sounds almost too good to be true. I feel like I'm missing something in the spec sheet.

    I should know the answer to this but I don't. My wife has an LCD wall clock that displays the indoor and outdoor temperature (has a remote sensor) and runs for a year on two AA batteries. Its not back lit but stays on all the time. This display has to be pulling much less than 1 to 2ma to run for a year. It looks like it is a custom display in that it can only show numbers in a 7 segment display format. Is this custom LCD display allowing them to achieve super low power usage?

    Thanks for the recommendation on the Newhaven display. I use one of their COGs (black and white with the SPI interface) in one of my current products. I have even sent them an eMail to check if they have anything in the sub mA range. They don't, just came back with the same COG recommendation.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2015-01-03 06:40
    You are indeed right that there are Liquid Crystal Displays in products that run for years on a tiny battery. LCD watches are everywhere, and have been around a long time.

    But if you desire micropower and a 24/7 display, you may find that these devices are not sold in signal units as a componet. The device you provided a link to, looks pretty good. But you just may have to buy one to completely investigate what it does.

    The Newhaven product that is similar seems to demand 500uA without backlight at 3VDC.

    The fact that the device you mention lists wristwatches as a use is promising. But custom LCD devices may really use a fraction the their whole surface for display as a way to further reduce power demand.

    If you really want to get 1 year out of a couple of AA cells, you may have to clock the Propeller far lower than 5Mhz. Watch crystals at 32760Hz are often used to minimize power drain.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2015-01-03 10:27
    That's an interesting possibility, John.
    Mouser has those here at $18 each.
    Adafruit sells a breakout for $40.
    More PR from Sharp

    I too am familiar with small watch faces and relatively small displays that have a fixed number of special icons along the edge and run on meager power. I've stayed away from their demanding parallel interface and the lack of versatility. I believe the reason for thier low power requirement is that they are not multiplexed. The voltage on each pixel can be constant until it the pixel needs to be changed. The slow AC clock is applied to all pixels at once to avoid degradation and permanent polarization. It is the constant updating of of rows and columns that requires power in larger LCDs. Sharp must have figured out a way to make it work.
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-01-04 04:34
    Tracy, I was looking at those exact links. I think I will give the larger $39 unit from Adafruit a try. If it works as I hope I will share the results back to the forum.

    Loopy, today I drop down to the 20kHz internal clock to save power and stay there until woke up by an external circuit and then jump out to the 5MHz external crystal. At the 20khz speed I have found that I can drive a 50uA analog panel meter with a counter (PWM) to display a value and keep total current under 200uA. The plan is to replace this panel meter with the sharp display. I was thinking I could send data to the sharp display when running on the external crystal and then drop down to the internal clock to drive the refresh signal to this sharp display. If I cant I will give your 32kHz watch crystal ideal a try. I wonder if that crystal will still allow me to use Tracy's 4 port serial object? Maybe at a 9600 baud rate??

    Thanks Tracy and Loopy for the help!
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2015-01-04 04:58
    The Sharp LCD spec sheet is very misleading as it strangely does not include the VCOM oscillator (even as an option) which in itself will draw far more power and also in the necessary clocking of the LCD itself. The trouble with a graphics display is that there are a lot of segment/pixels to multiplex (power power) whereas the wall clock only needs one segment for an icon or even the hours digit, and even then it can be driven directly without multiplexing.

    If you used a 32.768kHz crystal then you might do an exact 1/8 of that for serial transmission and you could dedicate a cog to receive for the same but a 4 port object has to operate at a much lower speed than the clock frequency.

    The problem I see here is that you are tackling a project with "professional" specs but you don't seem to appreciate these basics. That's just to say you need to be realistic with your requirements so that they do not outstrip your own capabilities. Either that or jump in and "burn n learn" as I say, and gain some experience.
  • tonyp12tonyp12 Posts: 1,951
    edited 2015-01-04 07:42
    TI uses the sharp lcd for a booster pack, and they say 10uW (~3uA) so that is very little.
    As it does not have back-light it could be that is the power consumption while actually displaying something.
    http://www.mouser.com/ds/2/405/slau553-489640.pdf

    Getting a system to only use 10-20uA with every IC and LDO IQ taken in to the account, requires skill.
  • RDL2004RDL2004 Posts: 2,554
    edited 2015-01-04 08:08
    Sharp indicates that the display operates in a very low power mode when the display is static and that the update rate can be as low as 1Hz. It seems like if it's used to display relatively slow changing data that the power consumption could actually be very low.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2015-01-04 11:26
    I don't think there is a VCOM oscillator in this display. It seems to have a one bit memory for each pixel, no multiplexing when the display is static, thus low current demand. The data sheet is quite specific about the power consumption. Typical 6µW at 3V when no display updates are in progress, and 12µW typical when the display is updated at 1Hz using a 0.5MHz SCLK. (maxima TBD) The data sheet implies that the update current is an average over time. The current required during the update itself is not stated, . At 0.5 MHz, it would take about 20ms to update all 9216 pixels. Guessing 25ms given the overhead in the SPI protocol. Working backwards, I = (12 - (6*0.975))/0.025 = 246 µA during the update.

    A screen buffer in the Prop would require 288 longs. The interface looks to be quite straightforward SPI mapping to the rows of pixels. I don't understand the effect of the EXTCOMIN, EXTMODE and the sirial (sic) input flag. Have to try it to discover the gotcha's, if any. There is a screen memory refresh command that evidently needs to be sent about once per second, but it is only 16 bits.

    John, you can't use the 4-port object to do anything useful at 20kHz.

    That has nothing to do with this display though. It is SPI, sychronous, not asynchronous serial. With SPI I believe you could do updates and certainly the refresh the screen memory when operating on RCslow.
  • tonyp12tonyp12 Posts: 1,951
    edited 2015-01-04 11:53
    Little unusual that ChipSelect is active high, though could be good if you share that pin with another spi device that is active low, you only need one CS pin to control both.
    data is sent LSB also unusual.
    Poorly written and translated data sheet, "sirial input" what is that?

    Here is better version
    http://www.sharpmemorylcd.com/resources/ls013b4dn04_application_info.pdf

    VCOM is some type of inversion safety to not build up a charge.

    Sharp recommends keeping maximum time between
    VCOM toggles to no more than one second, and
    refreshing data every two hours, to prevent stuck pixels.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2015-01-04 12:22
    Thanks Tony, that data sheet is far better than the older preliminary one. It makes clear that there is a VCOM to keep the screen from over-polarizing, but it needs to be toggled only about once a second either by flipping the EXTCOMIN pin or by sending a command or a frame update over the SPI interface.
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-01-05 04:00
    Yep I knew it was a SPI interface to communicate with the display, I use the 4-port object to talk to an xBee and I didn't think the slow crystal would let me use it.
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-01-05 05:25
    Thanks Tony! The PDF you found helps a bunch, I struggle with poorly translated spec sheets!
    I'm planning on driving everything at 3v and at first I thought I would be fine based on the power supply table on page 6 it says VDD and VDDA typical value = 3v Max = 3.3v. However, several places in the document refer to VDD/VDDA as being 5v. At the top of page 8 and again on page 15 they even give reference power supply circuits that convert VIN of 3v to 5v. Do you think part of this tech sheet was from another document and they just cut and pasted the last half of the document or am I completely missing something on the power requirements?

    They refer to this application note from Ken Green several times and on page 9 it gives more details on power requirements. One power scenario states the quiescent current for the panel is 22uA. A write to the entire panel will take 1.5mA and last about 50ms at a 2MHz SPI clock speed. This is all just fine for my needs I will only update the screen when I get a new value and I can control how often the updates are sent.
  • LawsonLawson Posts: 870
    edited 2015-01-05 15:30
    I have been playing with this display. While it specifies 3.3v operation, it still has useful contrast at 1.5v. Most often I've run the display off an IO pin so I can shut it down as needed. The SPI interface can be overclocked quite a bit. About 4MHz is the highest I've tried it so far. The SPI also works just fine with a 2KHz clock. (updating the display while running at RCSLOW) The display will remember it's contents through quick power cycles. You have to short it's power leads together for >10sec for a power cycle to clear the display.

    I've got a few candidate drivers working, but haven't gotten around to posting them on the forum or OBEX. I'll dig them out later tonight and post them.

    Marty
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-01-06 12:36
    Lawson, when i get my head back above water I will dig into your examples. Thanks very much for sharing!!
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-02-03 05:22
    Marty do you source the power for both VDD and VDDA (analog) from the prop? What is the difference between the two?
    Lawson wrote: »
    Most often I've run the display off an IO pin so I can shut it down as needed.

    Marty
  • LawsonLawson Posts: 870
    edited 2015-02-04 16:46
    Yes I source VDD, and VDDA from prop IO pins. If i remember right, the driver assumes they're linked. I'll just use one pin once I make a custom PCB for the screen. (initial testing was done on an ADAfruit breakout board) The data-sheet doesn't spell out the differences between VDD and VDDA. Based on other chips, VDDA powers all the noise sensitive sections of the chip, while VDD powers the logic. They bring them out on separate pins and join them on the PCB to reduce noise coupling between the two sections.

    Marty
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-02-14 09:58
    This week I was able to create a PCB and test the SHARP Memory LCD. I will be happy to share the Dip Trace files if it will help anyone.

    Test Circuit: Test was with my test PCB pictured below (Xbee not installed) current was measured as it feed the PCB's LDO 3.3v regulator. Supply voltage was ~4v (3AA rechargeable batteries).
    All of the Propeller's unused pins were set to output and high, only two cogs running main and a cog that toggled the LCD's VCOM.

    When measurements were made Main Cog was at WAITPNE and the VCOM cog was sending SPI data to MLCD to toggle the mLCD bias. PCB Board's LED was off.

    Test Results:
    RCSLOW = 45.14uA ambient lights On, 7.9uA ambient lights Off
    RCFAST = 1.03mA (Didn't test with ambient lights off)

    Observation; with my bench's fluorescent light on the power usage is around 45uA with the light off it jumps between 8uA and 28uA DC. This happens when running the circuit from USB power or battery power. With the room's overhead lights off the power usage dropped even more. It got as low as 7.8uA. I used the LED light on my iPhone and when it was shining on the LCD the power usage jumped back up so it doesn’t seem to be RF interference. It appears the power usage varies with ambient light.

    attachment.php?attachmentid=113182
  • yaddatranceyaddatrance Posts: 5
    edited 2015-03-28 00:18
    Wow, thats beautiful. I'd love to see your diptrace files. I just picked up one of these memory LCDs myself. Been banging my head against this display for the last two days. Also some tips on how you used this in RCFAST mode would be nice as well, I can get stuff going in PLL mode, but when I switch to RCFAST nothing seems to work. Were you using Lawson's code above?

    When doing the wiring did you choose EXTMODE as L or H?

    Also, I followed your youtube back to your laser etch of the PCB. What was your workflow getting the PCB data out without scaling? I've been using the PCB-in-a-box method, but I have a nice 100W CO2 table at the shop and I love how your board turned out.
  • JohnR2010JohnR2010 Posts: 431
    edited 2015-03-31 06:07
    Wow, thats beautiful. I'd love to see your diptrace files. I just picked up one of these memory LCDs myself. Been banging my head against this display for the last two days. Also some tips on how you used this in RCFAST mode would be nice as well, I can get stuff going in PLL mode, but when I switch to RCFAST nothing seems to work. Were you using Lawson's code above?

    When doing the wiring did you choose EXTMODE as L or H?

    Also, I followed your youtube back to your laser etch of the PCB. What was your workflow getting the PCB data out without scaling? I've been using the PCB-in-a-box method, but I have a nice 100W CO2 table at the shop and I love how your board turned out.

    Here are a couple of snapshots from DipTrace.
    attachment.php?attachmentid=113713

    All the pins that are tied to the display bus are tied to a pin on the propeller. The EXTCOMIn line could be tied to VSS since I use a software to send the inversion signal.
    attachment.php?attachmentid=113714

    I did refer to Lawson's code and it got me pointed in the right direction but I ended up writing my own. I think he tied EXTMode to VDD and I have it tied to VSS. With it tied to VSS the software has to send the VCOM inversion command to the LCD once a second. If you don't the screen can be damaged. One night during my testing the COG sending the VCOM inversion command got shutdown and the display was on all night with no vCom Toggle. The next morning the display looked horrible. I just unplugged everything and after a few hours or so the display was fine.

    My test code is too ugly at this point to share. I ended up rewriting it as an object that is included in other projects. I need to go back and clean up the original test code before I share. If your still stuck let me know I will be more than happy to do it.

    This document that Tony posted helped a great deal in putting together the code. http://www.sharpmemorylcd.com/resources/ls013b4dn04_application_info.pdf.

    Here is a link to the FPC connector for the display http://www.mouser.com/Search/ProductDetail.aspx?R=FH28-10S-0.5SH%2898%29virtualkey64550000virtualkey798-FH2810S05SH98 I had to do a lot of hunting to find it.

    Tracy really helped me out in regards to changing the clock speed with this post http://forums.parallax.com/showthread.php/157627-Clock-Speed-change-from-external-xtal-to-internal-RC-and-back-again-to-external-xtal?p=1296092&viewfull=1#post1296092

    I added my board creation notes to my original Laser PCB etching process in this post: http://forums.parallax.com/showthread.php/160556-How-I-use-a-40W-Laser-to-make-PCB-Prototypes?p=1323669&viewfull=1#post1323669

    I have already included this display in a couple of projects and I'm sure more in the future. Good Luck!
    attachment.php?attachmentid=113716
    597 x 652 - 83K
    1024 x 690 - 94K
    1024 x 1020 - 466K
  • Hi John, Are you still using and liking this? I've just picked up a couple of 1.28" LS013B7DH03 to try out on a micropower product that could use an at-a-glance display. 128 x 128 pixel. The Sharp lineup of memory LCDs seems to have evolved over the past year.
  • e-ink displays might work.
  • Hi John, Are you still using and liking this? I've just picked up a couple of 1.28" LS013B7DH03 to try out on a micropower product that could use an at-a-glance display. 128 x 128 pixel. The Sharp lineup of memory LCDs seems to have evolved over the past year.
    That project is on the shelf right now but it ran fine for 6 plus months. The display worked flawlessly I had it displaying a graph updated hourly as shown. The Propeller had no problem driving it. The reason I went with it over ePaper is the interface is much easier to work with.

  • Tracy AllenTracy Allen Posts: 6,664
    edited 2016-06-08 17:50
    I don't see many options for small e-ink/e-paper displays. One from Pervasive Displays and a shield to go with it are offered by Seeedstudio. I too find both its hardware and firmware interface daunting. I want a smaller screen anyway, no larger than 40mmx40mm and the Sharp displays fit that bill nicely.

    Having code examples for the Sharp (Marty's) is great. John, did you polish up your object? I may ask you more about it when I get the hardware (the 10 pin FPC) hooked up. I'm making a really simple carrier board with just the bypass capacitors and means to connect it to free points on an existing Prop board. I see your nice display in the bezel, that says, "TDS" "hazardous". Is that water quality, total dissolved solids?
Sign In or Register to comment.