Driving 74LS from 3.3V uP

I realized that seems to be the problem. The admittedly rather wimpy boost and battery pack can handle everything if everything is sequenced slowly, but it can't handle all the sudden startup currents of the display, Numitron, and itself without falling over.

I brought the boost voltage down from 5 volts to around 4.5. The LS still seems to work fine, and the segments are still plenty bright, but the current consumption is significantly less.

Numitron and 74LS IC I should say

Seven cold filaments could pull a lot of current. You could do software tricks to soften the turn-ons, but the display might look silly.

I'm going to give preheating the filaments via PWMing the "shutdown" pin of the boost module a try, though I'll have to check the datasheet to see what the delay from a high level applied to that pin to output power good is.

If that's not plausible I can try PWMing the "ripple blanking" input to the '47.

rd)

volts

an

pulls

n the

d up

an say

at 5

from

hat's

being

.4W which at 85% or less efficiency (Chinese special) = 1.65W or more fro m the LiPo or 1.65/3.5= 0.5A *average* with possible 1A peaks for the boo st. Ckt is not the only thing hopelessly underpowered around here.

and

It could only look funny for a fraction of a second. You could prewarm the filaments by turning each on for its own time slice. Then turn on 2 at a ti me etc. It can all be done so quick the eye barely notices.

NT

Could be easier just to add resistors across LS47 outputs to preheat, depending on ease of changing hardware vs software I suppose.

piglet

I'm hoping to get a few PCBs of the circuit I'm working on made if it turns out alright, so I'd like to keep everything as compact as possible and do it in software if I can, rather than add a resistor network to the board.

In the final design I'd like to ditch the LS chip entirely and simply use a uP of the same architecture with enough pins to handle all the segments directly. Many AVR chips are rated to sink 200mA to ground total through all pins, which surprised me a bit...though this circuit would push that limit and perhaps make me feel a little uncomfortable.

I'd also like to dynamically adjust the voltage to the segments depending on how many are lit to increase battery life via PWM, but that may not be possible if the boost's enable pin can't handle being PWMed, so I may have to use a driver of some type. The cost/space advantage of using one 20 pin SOIC uC vs an 8 pin SOIC uC and a 16 pin SOIC is not gigantic.

You can use tristate outputs to adjust the boost's voltage by switching feedback resistors in and out. I've sometimes used PLDs as analogue elements to switch between photodiode segments.

Cheers

Phil Hobbs

Remember the old T-bird sequential turn signals? Original Ford design was a motor driving a cam-shaft that operated switches that sequenced the tail lights.

I converted it to a chip driving power transistors, but power transistors couldn't cope with the 10X current surge, without pre-warm, which was complex and power wasting.

Ultimately used SCR's. ...Jim Thompson

Well, you could fast-forward 40 years or so and use LED displays.

Why not just use transistor arrays?

I suspect the OP's problem is collapsing the "eBay special"... persistently ignoring the fact that filament-type bulbs have ~10X current when cold. ...Jim Thompson

Doesn't look as cool at Burning Man, no fun at all.

Hey! I recognized that fact this time yesterday.

That 10x could be reduced with a little resistance. And the lines could be turned on one at a time.

NT

in

er

If PWM isn't an option there are other ways to reduce display current - tho ugh not greatly of course. Eg 2 output pins driving one filament, one via a resistor, or even a slightly adjustable voltage supply. Or an LDR/photodio de could adjust brightness according to ambient light etc.

NT

pin

wer

to

If his ebay special is one of those Canton Power jobs then it's not half ba d. It should be using an all-in-one IC with UVLO, OVP, OCP, OTP, a bunch of other P's, soft-start, and external /EN. Goes to show you really can't mak e stuff idiot proof. My reading of the problem is adverse interaction with the LiPO or bench PS and the turn-on surge. It should be fixable with 1000u F/ millisecond turn-on required at the input of the boost- to keep it above the 2.5V minimum. Power supplies just don't "fall over" for no reason.

I haven't seen a spec for the "eBay special". I wonder if the OP bothered to observe the "eBay special" output during power-up? ...Jim Thompson