0
\$\begingroup\$

I am trying to build an automotive strobe light for a two-stroke engine.
The problem with strobe lights sold in the market is I have not been able to find one which works well at high RPMs.
My application is a strobe light that should work at 20,000 RPM.
I do not need a pickup as the ECU puts out a 12 Volt pulse on the tach signal which I plan to use.

My only concerns are:

  1. Should I use an LED or look for some other light source? (I am reading the old posts related to LED on and off latency).
  2. The light needs to be bright.

This is the kind of circuit we are thinking of:

R3 controls brightness

\$\endgroup\$
5
  • 2
    \$\begingroup\$ 20000 RPM is only 333.3 Hz frequency, or period of 3 milliseconds. The problem is, how short light pulse you want? \$\endgroup\$
    – Justme
    Commented Jan 6, 2022 at 12:38
  • \$\begingroup\$ how quickly does the light chosen operate? if the pulse is too short will it even flicker or even glow? \$\endgroup\$
    – Solar Mike
    Commented Jan 6, 2022 at 13:29
  • 3
    \$\begingroup\$ An LED can offer very short pulses in excess of 100kHz which will be necessitary for your application. To get a pulse that is on for 1° of engine rotation and the duty-cycle will be about 0.27% at 333Hz. For you to see the flash with that low duty cycle, you'll need a bright LED (1-watt or more and a lens that can focus that light to your working area. At those low duty cycles, you can overdrive the LED - try 3 to 5watt range during the short pulse. An LED will be fine but, your Tach control may be an issue. Where in the engine cycle does the tach pulse occur? 0°, 10°, 20° or what ever? \$\endgroup\$
    – GT Electronics
    Commented Jan 6, 2022 at 13:40
  • 1
    \$\begingroup\$ I don't believe that there isn't an off-the-shelf solution already for this problem. I'm all for building stuff, but only if the goal is to build stuff (like you enjoy it). If your goal is just to get a functional strobe, suggest you keep looking at what's on the market already. Not sure if it would work, but note there are strobes that don't require ANY connection.... You put a reflective mark on your shaft, and you aim the strobe at it... The strobe has a dial, you adjust it's flashing rate until the reflective mark is stationary and then read the value from the strobe directly. \$\endgroup\$
    – Kyle B
    Commented Jan 6, 2022 at 19:13
  • \$\begingroup\$ GT Electronics You nailed it. This is our problem. Trying to get 1 deg accuracy with conventional automotive strobe lights designed to work on TCI ignitions. At much lower RPMs. The spark duration on TCI based ignitions seem to be much longer vs the spark duration of a CDI ignition. (Most automotive timing lights are not designed keeping in mind the fact that it must work on a CDI also. This is our problem trying to verify 1 deg change in timing at high RPM. We mark the magnet with graph paper and verify the timing. The duration of the 12 Volt tacho output pulse is also very short. \$\endgroup\$
    – Chinoy
    Commented Jan 8, 2022 at 12:10

2 Answers 2

Reset to default
2
\$\begingroup\$

Almost certainly your problem is that you are attempting to use white LEDs. These LEDs use a blue LED and a phosphor which provides the other colors. The phosphor has a fairly fast turn-on time, but has a long persistence time, so latency is not the issue.

Persistence for white LED phosphors is typically in the milliseconds, so your target of 20,000 RPM (333 Hz, or 3 msec period) is going to be a problem.

The solution is to not use white LEDs. I'm assuming you're using high-power (> 1 watt) LEDs, so your choice is going to be somewhat limited. However, just as an example you could look at https://www.sparkfun.com/products/15200 for a decently powerful LED for cheap money.

You would provide a different current limit resistor (the equivalent of R3 in your schematic) for each LED, but still only need a single driver transistor. I would also be very cautious about trying to run this unit at 3 watts per LED, since if a sketchy Chinese-language data sheet says it gets "very hot" I'm inclined to believe it.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ Phosphorescent lifetimes on typical white LEDs are less than 1 microsecond, and you can generate 1-2 MHz square waves from normal white LEDs pretty easily, so I don't think that is the issue here. \$\endgroup\$
    – user1850479
    Commented Jan 6, 2022 at 14:55
  • \$\begingroup\$ @user1850479 - When you say "typical white LEDs", what power levels are you talking about? High-power LEDs are used for illumination, and use of a high-persistence phosphor is a GOOD thing in such applications, as it reduces flicker at line frequencies. \$\endgroup\$
    – WhatRoughBeast
    Commented Jan 6, 2022 at 19:27
  • \$\begingroup\$ When a phosphore enters its excited state it becomes unable to absorb little until it relaxes. For this reason short lifetimes are needed for high power lighting, otherwise you would saturate the phorphore and turn the light blue. Using a short lifetime enables high power density. \$\endgroup\$
    – user1850479
    Commented Jan 6, 2022 at 19:54
  • \$\begingroup\$ I've measured the lifetime on various lighting LEDs and typically found values on the order of 500ns. We use them for testing the step response of photodetectors. A phosphore that could buffer optically for 100 Hz flicker would need a lifetime ten thousand times longer and would thus have to be ten thousand times thicker. I don't think that is practical, but if you have measured such I would be interested to see your measurements. \$\endgroup\$
    – user1850479
    Commented Jan 6, 2022 at 19:58
  • 1
    \$\begingroup\$ "However, the phosphor-based white-light LED has a limited modulation bandwidth of about a few megahertz that is caused by the slow response time of phosphor", arxiv.org/ftp/arxiv/papers/1612/1612.08477.pdf <- This paper matches what I have measured. \$\endgroup\$
    – user1850479
    Commented Jan 6, 2022 at 20:00
1
\$\begingroup\$

Should I use an LED or look for some other light source? (I am reading the old posts related to LED on and off latency).

I've built strobe lights for a 20 kHz camera system. For simplicity I used an array of 1W Cree white LEDs, a 12 V power supply, a resistor, a power FET for switching, and some optics to focus as tightly as I could. With this setup I could easily turn the LEDs on and off in a few microseconds, limited by the parasitic capacitance. If you want to go really fast, you can pull the diode negative to more rapidly reverse bias it, in which can get down to the phosphor lifetime of a few hundred nanoseconds.

The bigger challenge is usually getting enough light. If you're aiming for sub-millisecond illumination, your light will be very, very dim and you will need a lot of LEDs to take usable picture, clearly see the object by eye, etc. Below a certain point you may need a flashlamp to get enough brightness.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.