556 timer triggering problem

[philba]

Funny, Otto Schmitt was born in the USA of parents also born in the USA. Invented the trigger in England just before the start of WWII.<p>For more than you would want to know: http://www.thebakken.org/research/Schmitt/Otto.htm

[Chris Smith]

Actually you can replace Q2 with a balanced pair of IRF 511 FET’s, and drive the led with a one ohm resistor at 5 to 12 volts, if your clock speed is fast and your duty cycle is short. <p>The led will out preform a ½ amp 6 volt flash light bulb for brightness and distance, and still not even threaten the life of the led because of the short duration of on time. <p>And all of this with out the use of a super bright LED, which will be only brighter. <p>If you really want super bright, use a dual flip flop chip instead of the 555/556, at 2% to 10% on duty cycle, 50k clock speed or higher, and 50 volts or more. <p>Just don’t stare at the LED, it will blind you.

[philba]

By the way, I don't believe the driver transistor is necessary. The 555 can directly drive the LED. It has the capability to drive quite a bit more than the LED needs. The spec shows 100mA on output as possible and the dip max power dissipation is 1600 mW (556) or 1100 mw (555). LEDs will be well within those limits.<p>On the differentiator, this is needed only if the interruptor will be moving slowly. i.e. such that it interrupts the LED/PT gap for longer than the 555 pulse duration (555 output high can not be shorter than the trigger pulse). I was curious about this and SPICED it. The capacitor needs to be higher or the pull up resistor needs to be larger. I got good results with 10K and 10n but 100K/1n also worked ok. The values from the circuit diagram did not work. I still think its not needed - strobing slowly doesn't make a lot of sense.<p>Phil

[Will]

Philba,
For my educational improvement only - How did you get the 110mS pulse time ? I figured (Wrongly probably !) that the time constant was 10 micro secs so thaT, given a clean square wave front input of infinitesimal rise/fall time the 5 volts at the trigger would drop to 0 to 1.0 volt but then, within 10 microsecs would have risen again to back to > 3.16 volts,> 4.32 volts with 20 microsecs etc. so, dependent upon the trigger cut-off level, the period would be somewhere between 10 and 50 microsecs ??
This is a good thread !

[jimandy]

I don't know how philba got 110ms, but I get the same as philba (approx) using Don Lancaster's handy dandy graph on page 183 of the TTL Cookbook. Much easier (for me) than doing the math.<p>[ July 06, 2005: Message edited by: jimandy ]</p>

[philba]

Well, its not exactly rocket science. per the datasheet, the pulse length is 1.1*R*C for monostable. 1.1*(10,000)*10*(1/1000000) = 1.1 *.1S You have to watch your units and convert to Farads (10 uF is 1/100000 of a farad, for example).<p>Interestingly, the chart from the datasheet shows something like 80 mS. I usually just do the math but the chart makes me wonder. The SPICE simulation I did gave about 110 mS for those values of R and C. But then, if they used the datasheet to build the model, it would be.<p>Its always a good idea to read the datasheet. They are usually only wrong 10-20% of the time
National LM555 datasheet<p>[ July 06, 2005: Message edited by: philba ]</p>

[Robert Reed]

I think you two guys aren't talking about the same thing, One is on the 10 microsec. differentiater of Q1--the other is on the 100Msec timing network of the 555 IC.

[jimandy]

Robert, we're talking about the 110ms that Will was asking about and that was referred to in a much earlier philba post re: the 555. I think philba believes it a bit too short - I think it's a bit too long. But we really don't know because we don't know how fast flattop will be spinning his wheel (or whatever).<p>Of course, what would really be neat is if that pulse width could be varied according to the repitition rate of the triggering signal.

[philba]

I'm fully prepared to believe we are all talking about something different.<p>On the flash duration. The problem is that under a certain duration, the flash will not be visible or at least provide enough illumination (it is an LED, after all). I dunno, the best length needs to be tested. I've always had trouble seeing an LED flash under 100 mS but that's my problem...<p>The differentiator is not really a square pulse but rather a sharp "trailing edge" with an RC rise. Here's what spice says it looks like:

Assuming we are using a Vcc of 5V and 1/3 of that for the trigger level (1.67V) then it looks like the trigger pulse is about 50 uS wide. I think 555 trigger pulses can be as short as 200 nS so that's not an issue. With the 1n capacitor, the pulse didn't get low enough to meet the 1/3 Vcc rule.<p>Frankly, I'd try getting rid of the 555 and just using the output of PT as input into a driver transistor and turn the LED on when the PT/LED is interrupted. I bet that would work quite nicely as a strobe. The faster it spins, the shorter the flashes.<p>[ July 06, 2005: Message edited by: philba ]</p>

[Robert Reed]

Heres the thing--we dont have a clue as to what the input wave shape looks like based on info that was given.
The differentiating network will pass only the higher frequency components of the wave form in question, according to its time constant. The initial negative spike can drop no faster than the waveform producing it and in fact will not even show up at all if its rise/fall times are too slow for the RC values employed. Up to now, we have no idea what that is. From a design concept, the whole circuit development is based on this parameter,(the actual out put of the PT).If the wheel were spinning slow eough, the output characteristics could be such as to not produce any 556 triggering at all. And I agree with you Philba that the cicuit could probably be made simpler. One thing that gripes me about some posts, is that there is never enough info to give complete answers.

[jimandy]

Well now, Robert, I'll reiterate my suggestion for a Schmitt Trigger following the PT. Not only would we have nice squared up pulses, but they would be roughly proportional in duration to the speed of the rotating wheel (I should say inversely proportional). One gate in a 7414 could do the job and the remaining five gates could be paralled to give a pretty healthy drive signal to the LED. But I agree with philba that if the pulses were very short (wheel spinning really fast) the effective illumination would be quite low. This scheme would eliminate all transistors except the photo sensor.

[Will]

Of course you were right Philba, as was Robert Reed - I was looking at the wrong time constant. The effective time constant of the device (Once it has been triggered) is controlled by the 10 mic capacitor - I was looking at the wrong on i.e. the pulse input with the 1.0 nS capacitor. As I figured, with the 1.0 nS on the trigger input the trigger volts would recover to 63.2% (somewhere greater than 3.3 volts) in 10 microsecs with a perfectly square input pulse but, with other than a perfect square front, as you said, it would never trigger. I did learn something and now I have more math to play with to determine the slope of the best input pulse which would NOT allow the trigger to reach the 2/3 Vcc level. Thank y'all for the interesting discussion.

[philba]

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by ROBERT REED:
Heres the thing--we dont have a clue as to what the input wave shape looks like based on info that was given.
The differentiating network will pass only the higher frequency components of the wave form in question, according to its time constant. The initial negative spike can drop no faster than the waveform producing it and in fact will not even show up at all if its rise/fall times are too slow for the RC values employed. Up to now, we have no idea what that is. From a design concept, the whole circuit development is based on this parameter,(the actual out put of the PT).If the wheel were spinning slow eough, the output characteristics could be such as to not produce any 556 triggering at all. And I agree with you Philba that the cicuit could probably be made simpler. One thing that gripes me about some posts, is that there is never enough info to give complete answers.<hr></blockquote><p>Yes, this is correct. I played with a much less crisp trigger in my model (10 mS slope from 5 to 0) and it takes a 1 uF cap to trigger the 555. If it were me, I'd just eliminate the cap altogether as it really doesn't buy you much and seems to add a lot of uncertainty. The larger cap could stretch the output pulse by nature of its slower RC rise. A 1 uF cap would keep the trigger below 1/3 Vcc for a touch under 40 mS. 10uF would be 400mS. I am a big fan of keeping circuits simple. Yes, at low speeds the 555 will be on for longer (because the trigger is still low) but this seems a special case and who needs a strobe for a very slow moving object. You can calculate the rotational speed below which it will happen but that needs the width of the interruptor blade(s). <p>A schmitt trigger would help but i still think its not necessary - just a PT/pull-up, as I showed earlier, directly coupled to the trigger of the 555. As long as the output gets below 1/3 Vcc and stays there for 200 nS, the 555 will trigger on.<p>To be fair to the OP, he did include a schematic which is vastly more info than most requests for help have. I doubt we will ever see a posting that has every last piece of info needed.<p>Phil

[Robert Reed]

Jimandy
I too like the idea of a Shmidt trigger. The 74HC132 "nand" gate has Scmidt triggers built into it's inputs. Just one of these driven by the PT output and then followed by a very sharp differentiating network will drive a one shot very cleanly. Since this is a quad device, one of the stages can easily be configured to a one shot. Pick the time you want and drive a 2N3904 with it and you have a nice simple cicuit for powering the LED indicator.
Since the output of the first nand gate can be differentiated to an extremely short trigger pulse (100 nano seconds or less), this will no longer be of a concern for any operations on down the line.Just pick your desired RC time constant for the one shot and your done.If you Need help with the one shot flattop --E-Mail me.

[jimandy]

Ah, Robert...Thanks. I'm glad someone is agreeing with my idea of using a Schmitt trigger. And you are also coming around to my idea posted on Jul 2 of using the 132 part. I don't think there is any need to differentiate the output of the 132 (or optionally a 7414). By definition of the Schmitt trigger action, the rise (fall) time is clean and fast .