Driving 7 Segment LED Display with High Efficiency

[Kane]

Hi,

I am making a battery powered digital readout which uses 2 AA's and 4 7-segment red LED displays. I would like some help on how power the segments efficiently.
The LED's will be drawing minimal current, around 100 uA per segment (400 uA if multiplexing each digit) Enough to be visible indoors. Currently I have common anode 7-segment digits but I can get either if a solution works with one better than the other.

My main power supply from the batteries is a boost converter to 3.3V which powers the microcontroller and another sensor.
The Red LED's in the display have a forward voltage drop of around 1.8V. If I simply used a current limiting resistor nearly half the power is wasted in the resistor.
I thought about adding a buck converter to drop closer to the LED's forward voltage and then using a smaller current limiting resistor but the current regulation is poor when the supply voltage changes a small amount, I'm not sure if this would be an issue since the buck converter should keep it fairly constant. I've also thought about using simple transistor current sinks and I tried a few circuits up on a breadboard but they also seemed to change a lot with temperature such as when I touch the transistor with my fingers.

I would be happy with anything 80% or higher efficiency. I'm not really sure which direction to go as buck converter's aren't 100% efficient so by the time it goes through say an 80% efficient buck conversion and then if whatever I use to drive the LED's is 80% efficient, that whole system would be only 64% efficient which is almost as bad as just using a resistor.

I would greatly appreciate any help,
Thanks,
Kane

[Benta]

With this kind of power constraint, I'd personally use a 4-digit LCD instead. Perhaps combined with a "backlight-on-demand" if needed.

[Kane]

Hi, I did consider a 4 Digit LCD I wasn't happy with the visibility, both contrast and viewing angles can be quite poor depending on the lighting. Having a backlight on all the time which would be a requirement for visibility would put it in the same realm as an LED 7 segment display I would think so I opted not to use one.

[floobydust]

Another approach is to add a boost-converter to say 5V so you can drive higher voltage >3V (green, blue, white) LED's which are really much brighter, much more efficient than red.
The payoff is some energy losses in the boost-converter but more than recovered by gains in LED brightness and the new lower current needed for them.

[DavidAlfa]

Yes, if the driver/mcu is fine with ~2Vdd, you could take that approach.
Normally converters won't be very efficient at very low loads.

Still, 100uA will have very low brightness, have you considered a small oled screen? They will last a lot if not driven at full brightness 24/7.

[Benta]

Hi, I did consider a 4 Digit LCD I wasn't happy with the visibility, both contrast and viewing angles can be quite poor depending on the lighting. Having a backlight on all the time which would be a requirement for visibility would put it in the same realm as an LED 7 segment display I would think so I opted not to use one.

7-segment LCDs have excellent contrast and viewing angle. Just look at your multimeter (if you have one). Also at 3.3 V. You just need to select the right supplier instead of the cheapest Sumdingwong from PRC.
I never said anything about constant backlight. I said "on-demand". For example a button or a light sensor.

[Benta]

so you can drive higher voltage >3V (green, blue, white) LED's which are really much brighter, much more efficient than red.

I think you need to re-analyze that statement (and read a few data sheets). I don't completely disagree about the green LEDs, but for the rest...

[selcuk]

7-segments and 2 AA batteries are not a good idea. What is the expected operation time? You want it indoors but for example outdoor daylight performance of reflective or transreflective LCDs are generally better than LEDs.

I think you should control the LEDs with current mode control instead of voltage and eliminate the resistors completely to increase efficiency. Think about 1W or 3W lighting LEDs used nowadays. They all have current controller and no resistors.

[floobydust]

so you can drive higher voltage >3V (green, blue, white) LED's which are really much brighter, much more efficient than red.

I think you need to re-analyze that statement (and read a few data sheets). I don't completely disagree about the green LEDs, but for the rest...

Can you elaborate? Low voltage red LED's (AlInGaP 2V) are still at the bottom for efficiency and brightness compared to the 3V club green, blue.
InGaN phosphor-converted amber, red LED's have taken over when you need highest efficiency and brightness but they are basically blue >3V.
This is why I suggest using a boost-converter, as well as for getting more life out of the AA batteries.

[Kane]

Thanks for all the replies.

I have tried Kingbright red, yellow, green and Innolux red and blue 7-segment displays. I found for a given current that the reds/blues/greens were very similar brightness with red having much lower forward voltage which is why I picked red.

I considered OLED but was worried about burn in as it's displaying digits all the time and most datasheets specify many milliamps to operate.

I have a fluke 87V and a brymen 786 multimeter and I find myself reaching for the backlight button often, it's just the nature of LCD's without a backlight in poor lighting conditions which can often be the case indoors. I haven't ruled it out but for now I will try make the LED 7-segment display work.

I don't have a hard target for operation time, currently it would probably last around 6 months being used 8 hours a day 5 days a weeks. I would be okay with that length of time as a minimum but I know it can be much better. Software optimisations such only lighting up non 0 digits and an efficient drive system for the LED's I can see being able to hit 12 months or 6 months with double the current for better visibility.

Thanks again

[floobydust]

You can also add a tinted display window for more contrast in bright ambient light.

I see Kingbright offers red 7-seg displays that are 10x brighter than the other reds. Hyper Red AlGaInP on GaAs substrate 1.85-2.35V at 10mA.

But then I see you have to consider the human eye's sensitivity also factors into it. A bright LED (high luminous intensity rating) does not mean it appears bright.
https://www.slideserve.com/ofira/phosphor-converted-white-led

[Benta]

Can you elaborate? Low voltage red LED's (AlInGaP 2V) are still at the bottom for efficiency and brightness compared to the 3V club green, blue.
InGaN phosphor-converted amber, red LED's have taken over when you need highest efficiency and brightness but they are basically blue >3V.

You're talking lighting LEDs, which is complete different story, running at 350 or 700 mA or even higher.
For low-power LEDs, red or green are the most efficient, due to low forward voltage. Red: ~1.8 V, green ~2.0 V, as well as current/light conversion ratio. Red leads here.
Example: https://www.vishay.com/docs/83180/tdcx10x0m.pdf

Green has an advantage in "eye efiiciency". The human eye is most sensitive to green light, followed by red (and at the bottom: blue).

Hope this clarifies my opinion.

[Peabody]

I've never used one, but it appears there are buck converters that would operate on the two AAs and output nominally 1.8V.  Of course there would be losses, but it might be significantly more efficient than using resistors with 3.3V.

Traditionally, with IR LEDs and such, the solution is to power two LEDs in series, which doubles the output power on the same current, with a tiny series resistor.  But I don't know of a way to do that with 7-segment displays where you never know which segments will be lit.  But I wonder if there might be 7-segment displays which have two red LEDs in each segment.  Then you could feed them 3.6V, but at half the current, and get the same brightness.  Actually, Adafruit sells a 7-segment display with two LEDs per segment, but it's 1.2 inch.

[Kane]

Yea it's typically the larger 7-segment displays that have 2 per segment Kingbright's smallest one with 2 LED's per segment is 1"

At the moment I have the single digit 0.56" SA56-11SURKWA with a typical rating of 27 mcd at 10 mA. Running those at 100 uA per segment is okay but a little dim but I also have some smaller 7-segment displays and those appear brighter,  I assume because the light is concentrated in a smaller area.

Inolux also have INND-TS56GXX which are green which have a rating of 218 mcd at 10 mA but I can't find them available anywhere to test them.

[Kane]

Now that I check, they do have whites in stock with a rating of 130 mcd at 10 mA and with a forward voltage of 3.2 V at 20 mA. I may have to get some to try

Edit: Upon further inspection, it seems that the mcd rating of these displays are done two different ways. Kingbrights has a higher and lower rating,  the lower value is specified as "Luminous intensity value is traceable to CIE127-2007 standards" while the higher value is specified as "Luminous intensity / luminous Flux: +/-15% "

The SA56-11SURKWA that I have is 97 mcd but 27 mcd when measured in accordance to CIE127-2007 standards.

I think Inolux is mixing these measurements in a single table as their amber, white and especially green seem absurdly high.

[Kane]

I have decided to try powering the 7 segment digits directly from the the battery with a constant current source using the op amp and dac built into the microcontroller. I'm also multiplexing per segment rather than per digit so I can use a single current sink.

My thought on powering the segments directly from the battery is that as the battery voltage decreases, the efficiency increases. Using red 7 segments with a forward voltage of 1.8V will ensure it will work across the entire battery range until they are dead. Initially it would be around 56% efficiency (1.8 Vf / 3.2V Vbatt) but would increase toward say 95% efficiency (1.8 Vf / 1.9 Vbatt). Assuming a fairly linear voltage decrease over time which from the discharge curves appears mostly to be the case then it should have an average efficiency of 75% or so I would think?

I've attached a schematic that I tried on the breadboard and seemed to work really well.

Any thoughts?

[Peabody]

At low battery voltages, I question whether the NFETs will turn on.  In effect, you have two N-channels in series, and need to get above the threshold voltage of both of them, and that's over and above the drops across the series resistor and the LEDs.  Do you have specific part numbers in mind for those mosfets?

[floobydust]

The strategy can work but the brightness will change depending on how many segments are on, because you are controlling total display current. So a "1" is brighter than an "8". Also the extra (voltage) energy is wasted as heat in Q10, R12 operating in a linear mode. You could use Q1 to do the work for fewer parts.

I am not sure if you are multiplexing the displays or not. I would try using narrower pulses to drive the LED display dimmer, to save energy. The human eye will integrate the pulsed LED light and it appears to be steady on. So a narrower pulse is less energy that a wide pulse, for the same voltage. Basically PWM.

[Kane]

The N channel FETs were just some 2N7000's I had laying around to test the idea. I am multiplexing per segment, I didn't draw the other 3 digits on the example schematic but I would be turning on a specific digit with the PMOS at the top of the schematic and selecting which segment with the NMOS's. The brightness doesn't vary as only one segment is on at any given time. I have tested it all the way down to 1.8V where the current then reduces as it goes under the forward voltage of the LED.

Are there any constant current sink IC's out there that have many channels that also consume microamps of quiescent current not counting the sink current flowing through it?

To clarify regarding the threshold voltages, a microcontroller is driving the MOSFETS at 3.3V from a boost converter. I just didn't want to constantly run the LED's from 3.3V due to the poor efficiency. The current setting resistor R12 will be set to have a maximum 0.1V drop across it so the N-FETS should have no problem turning on. Q10 will always be trying to sink 100 uA or whatever I end up settling on so the sources of the N-FET's connected to the segments should be at 0V if all segments are switched off or 0.1V when the 100 uA is flowing through the 1k resistor.

Edit: Would it actually be 0.1V + the voltage dropped by Q10? This would be about 1.4V dropped with fresh batteries and approach 0V as the battery voltage meets the forward voltage of the LED's. This would mean 1.8V Gate to source on fresh batteries. It seemed to work fine though.

[BennoG]

have you considered dumping the series resistor and use short pulses e.g. 5% PWM
Then you don't have the power loss in the resistor.

Benno

[mikeselectricstuff]

You may get better efficiency (visible with lower current) with a white LED display, or deep-green if they make them.
Maybe consider using individual LEDs to make up the segments, as optical efficiency may be better

[schmitt trigger]

What I once did to reduce overall power consumption on a display was to use a PIR sensor to enable the display only when there was someone in the vicinity.

You may want to implement a light sensor to vary the display’s brightness depending on the actual illumination level. At night, you can reduce the current to 1/5 or less and still have a pretty readable display.

[Buriedcode]

That running off 2 AA's requirement has been a stiucking point in a few projects I've worked on.  These days, some phosphor coated (blue) LED's can be remarkably efficient, being very visiable in low light running off <0.2mA.  But... as has been stated, they need > 3.2V to run, 3,3V is pushing it.  The 2-3V from AA's is fine for micros and even analogue, but it does making powering indicators annoying.

Boost converters have their place, and can get to >90% even for mA level output, if they use pulse skipping, but they generate noise (especially when hicupping to save power) which is pretty bad for RF or sensitive analogue.  By the time you've tested and designed a reasonably efficient converter to power the blue/white/phosophor LED's, any efficiency gain those LED's have has gone.  As you already have one there, I'd try white segment displays.

Re: LCDs.  The old STN types do indeed have crappy viewing angles - or rather, they're fine for a narrow range, at the right position - but terrible from the opposite side.  But modern VA displays are good up to about 160.  They also have a much better contrast, with on being pretty black and let a lot more light through when "off" making the backlight brighter for a given current.
Segment LCD displays, even the older types (TN) - the kind used on most multimeters can easily run below 100uA.

It's hard to get VATN digit displays, but the old HD44780 character displays are all over Aliexpress and the like.  They're about 1mA Icc, with 0.5mA on the backlight being more than good enough to see in low light.  However... AFAIK they are all 4.5V min.  Sure many character displays can run on 3V, but they have a negative voltage generator on board for the glass that will increase current consuption.

Like all engineering, sounds like you gotta compromise if you want to reduce power consumption further.

I do agree that LED segmentt displays and "low power" don't really work.  I also agree with your numbers

Edit: Just noticed Mike mentioned the white displays already, sorry 

[mariush]

edit :
Alternative to boost regulators would be charge pump regulators that can double the input voltage with up to 97% efficiency ... so for example 2 x 1.2v-1.35v = 2.4v-2.7v for rechargeable -> 4.8v-5.4v which is fine.

The problem is when user adds two fresh alkaline batteries that may be up to 1.65v each but even if they're not fresh you double 2x1.5v = 3v to 6v and you have issues because most microcontrollers support maximum 5.5v 

end edit

Yeah, if you want lowest power consumption, a LCD screen would be lowest power and a lcd segment driver can be very cheap.

You could also use an actual led driver that would allow you to set the maximum current and then multiplex the digits and give you brightness control so you could reduce the current as much as you want.

For example, this 14 segment 8 digit lcd display is only 3$ at Digikey and it's transflective, meaning it will be quite bright with regular light : https://www.digikey.com/en/products/detail/varitronix/VIM-878-DP-FC-S-LV/1118603

The reflective version is also cheap at 4.6$ if you get only 1 : https://www.digikey.com/en/products/detail/varitronix/VIM-878-DP-RC-S-LV/1118605

They get cheaper at 10pcs, around 1$ off each price.

Here's an example of a reflective display, with 4 digits, which allows you to have a couple leds on the sides and provide backlight if needed : https://www.digikey.com/en/products/detail/lumex-opto-components-inc/LCD-S401C52TR/469783

Both of these work with minimum 3v but you produce 3.3v anyway with your boost regulator.

Some yellow/green/amber leds would run direct from battery (2.4v to 3v) or from 3.3v and you could simply have a button user could press to turn on light for 10-30s and then auto shut down to save power.

The 8 digit displays seem to be  32 anodes x 4 cathodes , the 4 digit one naturally less

Here's loads of other segment lcd displays:  https://www.digikey.com/short/qvtfj58q

example of lcd digit drivers

32 x 4 :

Holtek-Semicon-HT1621B https://www.lcsc.com/product-detail/LCD-Drivers_Holtek-Semicon-HT1621B_C7873.html  or
Titan-Micro-Elec-TM1621B https://www.lcsc.com/product-detail/LCD-Drivers_TM-Shenzhen-Titan-Micro-Elec-TM1621B_C2980111.html
GN-Semic-GN1621CB  https://www.lcsc.com/product-detail/LCD-Drivers_GN-Semic-GN1621CB_C265477.html

36 x 4 :

Titan-Micro-Elec-TM1726 https://www.lcsc.com/product-detail/LCD-Drivers_TM-Shenzhen-Titan-Micro-Elec-TM1726_C2937431.html

32 x 8 :

GN-Semic-GN1622A https://www.lcsc.com/product-detail/LCD-Drivers_GN-Semic-GN1622A_C2758336.html

and loads of others here:  https://www.lcsc.com/products/LCD-Drivers_584.html


example lcd drivers on digikey : https://www.digikey.com/short/p9mq7885



If you want LED based digits, you could use proper dot matrix drivers to limit the maximum current, and then have the driver adjust brightness per led.. the problem is the quiescent current of some of these drivers is quite high at 1-2mA

For example

Lumissil IS31FL3738 https://www.digikey.com/en/products/detail/lumissil-microsystems/IS31FL3738-ZLS4-TR/14308389

Easy to use TSSOP package , 8 sources, 6 sinks , so ideal for 8 digits with 6 segments or 4 digits with 8-12 segments,  2.2 mA quiescent current, works down to 2.7v

Lumissil  IS31FL3746B  - https://www.digikey.com/en/products/detail/lumissil-microsystems/IS31FL3746B-QFLS4-TR/12675543 -

It does 4 sources (anodes) x 18 sinks (up to 72 leds), so you could have 8 digits each with 9 leds or 4 digits with 14 leds or something in between. Quiescent current is average at ~ 2.3mA and can run down to 2.7v


Lumissil IS31FL3728 : https://www.digikey.com/en/products/detail/lumissil-microsystems/IS31FL3728-QFLS2-TR/5319755

It can do 8x8 , 7×9, 6×10 or 5×11 ... quiescent current is high at 4mA, but shutdown current is uA worth of current, and can run with as little as 2.7v

[PlainName]

How fast does the display have to change. If it's slow then perhaps eink might be a solution: basically zero power between updates and easily visible in normal lighting.

Check out your local Aldi to see the displays they use for price labels - if you didn't know you'd swear they were printed paper things.