timed igniter for metal halide lamps

Are you supplying AC or DC to the lamp?

----== Posted via Newsfeeds.Com - Unlimited-Unrestricted-Secure Usenet News==----

The #1 Newsgroup Service in the World! 120,000+ Newsgroups

----= East and West-Coast Server Farms - Total Privacy via Encryption =----

240Volts AC/50Hz

Hi, Kfel. I'd guess you're running into one of the limitations of the

555 -- its timing is dependent on the R-C components, and any noise on the lines or in the power supply may cause premature timeout.

There can be several causes for this. If you want to stick with the

555, start by making sure you've got good bypassing on the power supply. Put a small (.01uF) cap between pin 5 (CTL) and GND. Keep the R and C as close as possible to the IC (preferrably tying one end of the R directly to pin 8, and one end of C directly to pin 1. Put your series resistor from pin 3 (OUT) to the opto as close as possible to pin 3.

But even those things may not always work well when dealing with inductive loads and arcs. And, without the test equipment needed to actually measure what's going on, you're basically guessing anyway.

A better solution is to use a CD4060 CMOS counter plus built-in oscillator for the 10 minute timing block. These came into general use as timing ICs for time delay relays in the 1970s, because they give more accurate timing in these situations.

The CD4060 is a divide-by-2^14 binary counter. As mentioned, it has a built-in oscillator. If there's electrical noise present, it may mess up one timing period, just like the 555. However, there are 2^13 (8192) timing periods, so messing up one or even several counts won't affect basic timing accuracy.

Here's a fairly simple way to drive your opto which should be more immune to timing glitches caused by EMI from the igniter (view in fixed font or M$ Notepad):

| | _/ |VCC-o/ o-o----o-------------------------------------------------. | | | | | | | .-. | .---)----o R | | | +| | | | | | --- | | .----. '-' | --- | .-o--------------------------. | | | | 10uF| | | 16 | | '---o | === | | Q14 | ___ |/ | | GND | | 4060 3o-o-|___|-o-| 2N3904 | | | | RS | | 12K | |> V ~ | | | 8 12 11 10 9 | | .-. | - ~ | | '-o--o----o-----o------o-----' | 12K| | === Opto-| | | | | | | | | | | GND Triac | | | === | .-. .-. | | '-' === | .001uF --- GND | R2| | | |R1 | | | GND | --- | | | | | | | === | | | '-' '-' C | | GND | o-------o | | || | | | | o-----o---||-' | | .-. | || | | 33K | | | | | | | | | | '-' '-----|

Thanks chris. I have a 60MHz digital oscilloscope (Tek make) and also a

40KV Tek probe to view the pulses from the igniter. There are certain other points that were not mentioned in the first post.

1. I did realise that noise caused due to arcing of the discharge tube was causing the timer to switch off prematurely.
2. To alleviate this i added a common mode rejection filter across the line and neutral.
3. Added a MOV in parallel to the line and neutral
4. Added a 0.047uF X2 capacitor in parallel after the filter.

None of these helped and the 555 was still prematurely timing out. As per your suggestion will try out the circuit with the 4060.

You'd probably need a storage scope to see what you want to look for on this problem. EMI is typically picked up by the traces, and fed into the 555 at the trigger pin. Also, even if you've got an isolated power supply, the transformer interwinding capacitance can feed a glitch into your power source.

The 4060, as noted above, was used for TDRs in the 70s, achieving quite a bit of success with very long time delays like this. It's possible you will have better luck with the new IC.

By the way, assuming you have a well-regulated supply, timing repeatability will be primarily dependent on the cap. Make sure to use one with tighter tolerance if the time delay length needs to be consistent to within a couple of percent.

Cheers Chris

Chris,

The oscillations do not stop using 1N4148 diode from the output to the oscillator capacitor. Is it ok if i use 1N4148 instead of 1N914?

For reference:

| | _/ |VCC-o/ o-o----o-------------------------------------------------. | | | | | | | .-. | .---)----o R | | | +| | | | | | --- | | .----. '-' | --- | .-o--------------------------. | | | | 10uF| | | 16 | | '---o | === | | Q14 | ___ |/ | | GND | | 4060 3o-o-|___|-o-| 2N3904 | | | | RS | | 12K | |> V ~ | | | 8 12 11 10 9 | | .-. | - ~ | | '-o--o----o-----o------o-----' | 12K| | === Opto-| | | | | | | | | | | GND Triac | | | === | .-. .-. | | '-' === | .001uF --- GND | R2| | | |R1 | | | GND | --- | | | | | | | === | | | '-' '-' C | | GND | o-------o | | || | | | | o-----o---||-' | | .-. | || | | 33K | | | | | | | | | | '-' '-----|

Dear Chris,

In the circuit diagram you have connected the 1N914 to the common point of capacitor and resistors. This way the oscillations did not stop after timeout. I then connected the diode between the output and pin 9 (with the cathode connected to pin 9) and this effectively stops the oscillations upon timeout.

I am using a regulated voltage supply. The EMI problem still persists when the lamps starts to arc. Is there no way to control this EMI with appropriate filters etc?

Alright -- we've got a good timer, anyway.

One thing which might help is pulsing the optotriac on and off at several hundred Hz. You can do this by having the output of your 4060 counter/timer going to pin 4 (RST) of a 555 set to oscillate at, say,

800Hz. The 555 output can drive the opto directly (through a resistor, of course).

You might want to try this and see.

Good luck Chris

Here's the circuit with the 555 pulsing the opto:

| From | >----------------------. | CD4060 | | VCC | | + | | VCC | | | + .--o------o---. | | | 8 4 | | .-. | | | | | | | | | | | | | '-' | | | | | | ___ | o-----o7 3o-|___|---. .----- | | | | | _|_ | .-. | 555 | V ~ V_A | | | | | - ~ / | | | | .--o2 | | | | '-' | | | .---' '----- | | | | | | | o--o--o6 | | | | | | === | --- | | GND | --- | 1 5 | | | '--o------o---' | === | | | GND === --- | GND --- | | | === | GND (created by AACircuit v1.28.6 beta 04/19/05

This works better sometimes with certain types of inductive or lighter triac loads. You might want to set the 555 with something less than a

1KHz frequency, R1 = R2 for a 66% duty cycle.

By the way, you haven't described the other side of the optotriac. Make sure you're using a good 4-quadrant triac that doesn't require a ton of gate current to turn on. I like the Teccor isolated tab TO-220 triacs for most loads (e.g. Q6010) -- they're guaranteed to turn on with only 40mA of gate current or so for all 4 quadrants, and IMHO, they tend to work more predictably with oddball loads than many other triacs.

Cheers Chris

Hi Chris,

Thanks for the idea regarding the 555. I did not try it out. Instead i used one of the outputs of the counter itself to cycle the triac on and off. This helped solve the problem of condcted EMI to a some extent. I have also used a transient voltage suppression diode to guard against conducted EMI. This combination seems to be working since i have been testing it for the past one week.

Great, Kfel. Glad you've got it working.

Remember that the 4024 output probably is going to be stressed driving an opto (10-20mA). You might want to use a standard NPN transistor to boos the current:

| | VCC | + | | | | _|_ | V ~ V_A | - ~ / | | | | ___ | | .--|___|--' | | R | From ___ |/ | 4024o-|___|-o-| 2N3904 | 6.8K | |>

| .-. | | | | | | 6.8K| | | | '-' | | | | | === === | GND GND | (created by AACircuit v1.28.6 beta 04/19/05

Cheers Chris