Low current LED 7-segment

I optimize two-parameter searches by downloading a spreadsheet file of the search, adding a column with my criteria, then sorting on that.

Cheers, James Arthur

Not the case IME. The eye responds magnificently accurately to the average--not peak--brightness, which I tested quasi-exhaustively over extremes of duty cycle and frequency.

The only ones that benefited from pulsing were some old LEDs--green, IIRC-- that had poor efficiency at low current.

Cheers, James Arthur

But replacing the resistor with an inductor and adding a flyback diode would save resistive losses. Synch rectifier, if you're really adamant.

(Only makes sense if Vcc is >> Vf(led).)

Cheers, James Arthur

Oh, it's a totem-pole output, so the synch. rectifier feature's already built-in.

======= digit .-.-.-. .--------------------. drive >--' ' ' '----|----+----------. | | | | | | V~~> V~~> | | --- --- | | | ... | | '----|----------|----' to segment drivers

Mind your duty-cycles!

Cheers, James Arthur

For pure visibility, it saves power to flash an LED really bright, maybe once a second. But that's not too friendly for a numeric display.

The display will be used for two purposes. The first is to display a value continuously for about one second every time an operating parameter is changed.

The second function is just a blinking indicator that the device is working - once per second, using one of the display segments, probably the middle horizontal segment.

But to display numbers, I would have to cycle through the seven segments repeatedly, firing one at a time, fast enough that the segments appear to be lit continuously, if that's possible. Even if I don't save power doing that, as long as it's no less efficient than just turning on all the needed segments continuously, then at least I get the benefit of using less power on the short indicator flash.

An additional advantage of sequential pulsing is that I would only need a single resistor in the common cathode of the 7-segment display, instead of a resistor for each segment.

Well, I'll have to experiment to see what will actually work for the value display. Initially I would be looking at an interrupt every 4 msec, so for seven segments that would be about a 36 Hz refresh rate. But that might be a bit blinky.

I don't recall the exact numbers, but I think a 100 millisecond flash, or maybe a bit less, is a good bright blink. Below that, you can trade off less time * more current and get about the same visual brightness.

So one flash per second could save about 10:1 on power. Maybe 20:1?

I think that is a wash on power.

To multiplex a display you need to cycle through all the segments at least at 50 Hz. As someone else pointed out if the display is moving with respect to the eye, it may need to be closer to 1000 Hz. I "see" taillights on cars as many cars as I turn my head because of the flashing. There should be a regulation to flash fast enough to not cause that effect. It was very disturbing the first few times I saw it in traffic. I kept feeling there were cars I couldn't see.

The human vision system has a quite bad efficiency requiring several photons within a short period of time, before a nerve pulse is generated. At very low lighting levels (less than moonlight) it doesn't make sense to send a constant stream of photons, since no nerve impulses will be generated. Sending all the photons in a pulse will enable a nerve impulse to be generated.

Apparently the OP was not talking a bout that low lighting levels.

To minimize daily power consumption, an LCD with white LED backlight would be a better solution. An ambient light sensor could be used and adjust backlight power accordingly. For operating in darkness relying on scotopic vision, select a cool white LED to further reduce the power demand.

Yes. The flickering of street lighting can be detected by rapid movement of the eyes. Probably 120 pulses per second.

In our products we keep the refresh rate above 150Hz.

ect, and

oes

rent

I believe that threshold notion is not correct. There are, IIRC, reference s indicating human eyes respond to tens(?) of photons per second, and integra te them over long enough intervals that you don't have to cram them all into a pulse.

A quick search gives these points-of-reference:

I know I can see LEDs with sub microamps of current flowing through them. How low does anyone need to take the current?

Rick, I know your are right but you might want to qualify your statement with info on your ambient conditions. Just a suggestion.

Yup. When it starts getting that dark you can't even make out objects--your resolution goes to nil.

For ordinary-use power savings I like the synchronous buck switcher scheme I posted up thread. It wouldn't be hard to do with a uC. I might even use it some time.

Cheers, James Arthur

Good idea. However, all I could find was a reflective LCD single digit display, and no backlighting types. Also, 1" minimum size might be a bit large: For really low power backlighting, phosphorescent sheet or tritium glow panels draw zero power. I used phosphorescent paper and plastic sheets on marine radios that required extrememly low lighting to prevent blinding everyone in the cockpit when running at night. "Dim to extinction" is a standard requirement for marine navigation and communications equipment.

I don't recall the exact current. I had a pair of red/green LEDs in series with a 3.3 volt supply. I figured the total voltage was below their turn on threshold... but it isn't a threshold. I put a resistor in series and found the current was amazingly low but the LEDs were still visible in normal room lighting. I'm not saying they were anything like regular brightness. I'm simply saying it doesn't take much current to make an LED glow so you can see it. I can't see the advantage (pun intended) to pulsing an LED as they are pretty low power if you don't need full brightness.

Yup. That's what we are doing. Our product will be in direct sunlight and in the cab of a truck at night. Right now we are using a 50ms pulse as a health indicator. During the day it is either the green one or the blue one and is pulsed via 1k and 3V3 every 3 or so seconds. At night they are pulsed via about 6k at the same rate. One test has revealed 100 days from two AA batteries, but we have now cut that to 1/10 the average current. We could PWM them if necessary but I am now concerned that the intensity in daylight might be lacking. My partner is giving me another set of eyes on this and, ultimately, the client will need to approve.

The latest high efficiency LED seem to output light down to sub uA currents unlike the early ones which all leaked like sieves.

No it doesn't. Your resolution when fully dark adapted is limited by the geometric aberrations of your eye lens at full aperture ~6mm which is generally worse (and different to) your normal glasses prescription.

A good test of visual acuity and limiting magnitude is to count the number of stars that you can see in the Pleiades. Or see if you can spot the double star Alcor and Mizar in the Plough Handle/Ursa Major.

The Roman lookouts test for visual acuity was to spot the close visual double star in the constellation of Lyra. It takes young perfect eyes and near ideal conditions to do this reliably as it is right on the limit of the resolution of the dark adapted eye at 3' arc. A handful of people can see Galilean satellites at maximum elongation with their naked eye (but there it is glare preventing most from seeing them).

It is generally reckoned that 1-10 photons in a single rod will trigger it and that several adjacent ones must be stimulated within a short period to see it as light. When fully dark adapted you also see some statistical flicker noise at ultimate sensitivity after an hour or so (sensitivity incidentally increases if you breathe pure oxygen).

The most sensitive area for dark adapted vision is in the periphery of your eye and you have to learn to look away from the faint object to see it better. Nudging the scope also helps as moving edges stand out. In the past this had evolutionary advantages of not being eaten by a hungry night predator. Cats eyes have much better sensitivity.

If you really want to save power the way to do it is direct drive of an LCD display and have optional back lighting for when it is dark.

I got dark adapted (in Truckee, where it's really dark at night) and looked right at one of the green LEDs that you sent me, really close. The threshold current to barely distinguish light was about 800 pA.

We have LEDs here that are decently visible in office light at 1 uA.