Timing light inductive delay?

[Tuner]

I had to look this up because I didn't know wether altitude made a clock run faster or slower. I thought slower because it would be moving faster, but gravity is more important it seems so it runs faster. These clock are now way more accurate than the crappy old regular atomic clock from the 70's. If you have 2 of these on your lab bench and put one on a stack of books, it will measure that just fine. They are trying to get their next model to be good enough to measure a centimeter of altitude!

Keeping time is an old Navy tradition and this stuff is funded by the US navy at NIST (National Institute of Standards and Technology). I was in Washington DC recently and happened to drive by and they display the time to many decimals on a clock outside the base.

http://www.nist.gov/public_affairs/rele ... 092310.cfm

So does anyone have numbers for timing light delay?

I don't remember the exact value, it was 25 years ago and that notebook is being evasive. I used a Tektronix 2221 to measure several lights. A scope confirms what you see if you use a distributor machine to compare the lights to each other to see which has the least retard.

At my last job we had a converted LED flashlight to act as a strobe for medical certification on a device. Let's whip up a circuit to convert Harbor Freight LED flashlights to accurate timing lights by replacing the switch with a trigger. Extra points if we can fit it in the flashlight or make a little replacement cap for the flashlight to provide this function.

Great idea, I like it. However, what is the slew rate of the LED to reach illumination? A brief search didn't produce much but finds a value mentioned for some of approximately 40us, which is about 2 degrees at 8000 RPM. 48,000 degrees per second at 8000 (eng) RPM, each degree is 1/48,000th = .00002083 sec. = 21us, so 40us is nearly 2 degrees.

[upinthehills]

I had the idea because I thought it looked like the accurate timing light you showed a photo of used LED's for it's light.

If there is an issue with the time it takes to get the LED to turn on, we can keep it turned on at a minimum value and then greatly increase the current when it is supposed to make a flash. LEDs are not very ideal diodes, but I think that would work. I'll look for some part numbers tonight.

There may be a problem with using white LEDs because the white comes from phospher. Do folks think that green or orange would work? That might prevent the cheap Harbour Freight flashlight idea though.

Edit:
I took a quick look and it seems we should be able to turn on and off times down to a fraction of a microsecond pretty easily. Let's aim for 100 nSec to start. It also looks like the average transistor to turn the LED off and on would leak enough to help speed up the turn on time.

[MadBill]

if a near-zero delay light is created, existing SA data banks (which of course incorporate typical delays) will need to be adjusted to suit...

[stokerboats]

Does anybody ever use the search function?

No dial back, it's just a simple strobe, the Kowa Seiki KE-50-7 “Synchro Beam” timing light is by far the best I have ever used. Small, pocket sized, two AA batteries, brightest ever, brighter than 12V powered lights. They are the most accurate light I know of because they have the least lag between the spark pulse trigger and light flash.

Unobtainium

[Rick Finsta]

upinthehills...

check this out - oscilloscope using sound card inputs for a few bucks in Radio Shack parts...

http://www.msextra.com/forums/viewtopic ... 94&t=38560

[Zmechanic]

LEDs should be able to driven well into the double digit nanosecond delay times. It's largely dependent on the effective capacitance of the junction. I have personally driven optocouplers in the uSec region, which is turning on a LED, having the light hit the junction of another transistor then undergo photoelectric emission.

1. Never turn the LED completely off, keep the junction biased with a lower current, just not enough to be bright enough to matter.
2. Use a peak and hold turn on circuit, just like with some injectors.

The corollary to this, though, is creating something to verify this to prove just how low the trigger to light emission time is.

[upinthehills]

Thanks Rick. I don't think my laptop has those anymore, but I'll go read that - I'm sure I'll learn something.

I did go look at some spec sheets and this seems doable still. For about $1 you can buy a chip with two timers on it. The spec sheet says 0.1 micro second propagation delay and 0.025 uSec rise time.

The thing about using a flash tube is they are very bright. So if we light an LED for one degree of rotation, it follows that it is off for the next 720 degrees, which sounds like a recipe for a dim flash. So the first thing I'm doing is trying to make a pulse that is wider when the engine is idling and narrows up as it ramps up RPM. I suppose you would like it to appear the same brightness whatever speed you are using it at.

The other thing I need to do is measure how quickly you can flash the white LEDs which use phosphor. Technically I should be working on my ECU...

[turbobaldur]

Thanks Rick. I don't think my laptop has those anymore, but I'll go read that - I'm sure I'll learn something.

I did go look at some spec sheets and this seems doable still. For about $1 you can buy a chip with two timers on it. The spec sheet says 0.1 micro second propagation delay and 0.025 uSec rise time.

The thing about using a flash tube is they are very bright. So if we light an LED for one degree of rotation, it follows that it is off for the next 720 degrees, which sounds like a recipe for a dim flash. So the first thing I'm doing is trying to make a pulse that is wider when the engine is idling and narrows up as it ramps up RPM. I suppose you would like it to appear the same brightness whatever speed you are using it at.

The other thing I need to do is measure how quickly you can flash the white LEDs which use phosphor. Technically I should be working on my ECU...

I pondered this about 10-11 years ago and did some experiments using an electric motor with a hall effect sensor to trigger a one shot timer which controlled a very bright blue LED. If I turned the one shot time short enough to freeze the motion acceptably, the light was so dim that it was of no use. A xenon strobe light can discharge a lot of joules in a very short time.

[user-23911]

Typically LEDs are flashed on an off very quickly anyway....that way they actually appear brighter for the same power consumption than they would if they were permanently on....for the same power consumption.
The way to tell is to look at an LED or better still, a modern display and turn your head quickly, you'll see them flash.

Another example of the flock of sheep trying to re invent the wheel.

The only way to eliminate delays in electronics is to eliminate all capacitance and that's just not going to happen.

[Zmechanic]

Typically LEDs are flashed on an off very quickly anyway....that way they actually appear brighter for the same power consumption than they would if they were permanently on....for the same power consumption.
The way to tell is to look at an LED or better still, a modern display and turn your head quickly, you'll see them flash.

Another example of the flock of sheep trying to re invent the wheel.

The only way to eliminate delays in electronics is to eliminate all capacitance and that's just not going to happen.

So what if someone or some people are curious just how low the latency on a timing light can be made? Praxis was having an issue, which sparked a discussion. Regardless, there is something to be learned. Something in the engine is retarding timing, the timing light's delay is causing a continual timing retard with frequency, etc. If trying to investigate a problem then finding a way to test it which may in turn spark an improved product makes me a sheep, then so be it.

And there are plenty of ways to combat capacitance to enhance speed.

[user-23911]

Every transistor has capacitance associated with the PN junction.........It's what you learn when you go to school.
It can't be eliminated.
It's always there and each switching stage adds more.


Within the microprocessor, there's the "clock" which sets the data rate, the faster the clock, the less delay between inputs and outputs........that's the second source of delays. The software it'sself adds delays.

It's something that you have to live with and understand how it works.



If you look up the PDF datasheet for any typical electronic component, it will usually tell you what the delay is....it's that delay which determines the maximum frequency that any device will run at.

For example a VB921 which is often associated with a megasquirt.

http://www.alldatasheet.com/datasheet-p ... VB921.html

Typically a 20 to 30 microsecond delay, not a lot but that's only in the output stage, it doesn't include anything in earlier stages.


An ignition coil is an inductor and delays are caused by inductors as well as capacitors.....so an ignition coil it'sself will always add to any delay, purely because it's an inductor.

[upinthehills]

If I turned the one shot time short enough to freeze the motion acceptably, the light was so dim that it was of no use. A xenon strobe light can discharge a lot of joules in a very short time.

Well, I think that's the issue. Still the ones on cars are pretty bright and clearly visible on a sunny day. I bought a 50 Watt LED for a track light the other day too. So they should also be available in 12V with a little parabolic reflector package.

I guess the next step is to measure how fast the phosphor is on those things.

[Zmechanic]

I picked up a cheapy timing light just to test out (and because I'm not pulling apart my good one). I hooked it up to my oscilloscope at my desk. The first thing I didn't expect was it fires on the falling edge of the signal coming from the spark plug pickup. AKA, when the plug current drops off. Seems like a bad idea to me..

I triggered my scope off my function generator I used to run the light, and monitored the SCR they use to fire bulb (through the HV step up transformer and trigger winding on the tube). It showed consistently about 20us of delay. Oddly enough, the advance dial did basically nothing. Although I'm not too surprised based on where I got this thing. And I don't have a way to measure light emission (yet). So there could potentially be some delay from when the arc over happens in the tube. My next order of business is to get a fast photodiode that has at least some sensitivity in a range other than iR so it can be used for LED based lights and Xenon.

So at
1000 rpm = 0.12 deg
2000 rpm = 0.24 deg
4000 rpm = 0.48 deg
8000 rpm = 0.96 deg

I suppose they figure that's probably fine to not cause anyone much issue. They are probably right for a majority of cases of average users.

[turbobaldur]

If I turned the one shot time short enough to freeze the motion acceptably, the light was so dim that it was of no use. A xenon strobe light can discharge a lot of joules in a very short time.

Well, I think that's the issue. Still the ones on cars are pretty bright and clearly visible on a sunny day. I bought a 50 Watt LED for a track light the other day too. So they should also be available in 12V with a little parabolic reflector package.

I guess the next step is to measure how fast the phosphor is on those things.

The phosphor is hopelessly slow actually.

[SchmidtMotorWorks]

Does anyone know of a timing light that can be delayed up to 720 degrees?