Tachometer black box for magnetos, Attn byacey, lark2004 (DONE?!?)

[Paul Moir]

Well, I never really got very far with this project. All I have is a pickup stage that works off one of the magneto lines: http://tinypic.com/dfwnd4.gif This gives me a measured signal like this: http://tinypic.com/dfwn4h.gif The red line is 0v, the top of the waves is 12v. The period is about 37ms (1600 engine RPM). Pulse length is about 2ms.An outboard tachometer usually has a little switch on the back to select the number of pulses per revolution of the engine. Most J/E engines produce 5 or 6 +12v/-12v pulses per revolution. So OMC tachs can be set for 5, 6, or 10 pulses pre rev. Teleflex tachs can be set down to 2. Also, all inboard tachs (including OMC) can be set down to 2 to accomadate four cylinder engines. Naturally, their signal is a little different.I can get two pulses per revolution easily enough by grabbing the signal from the other magneto line and adding it. That's simple as it only requires a pair of diodes.Anyway, to make a long story short, Iboats member gaugeguy assures me that any tach on the market will respond properly if fed with a 6v square wave or better. He also told me that duty cycle wasn't very important, but that 50% is ideal.But as you can see, I've got a 5% duty cycle there which I think is too short. I divide it by two, but then it wouldn't work with any of the tachs on the market since at best I could only get one pulse per revolution. Another idea I had was to run it through a PLL multiplier, but that will make the circuit a little complex.Ideally, I would like to simply "streach" the pulses out, but I have no idea how to do that. Especially since they should be streached less as the RPMS go up.Could anybody point me in the right direction here? Perhaps the PLL route would be easiest, or is there something really obvious I'm overlooking?

[byacey]

Most analog tach circuits rely on the varying duty cycle to operate the meter - more pulses per second= higher on time= more current to drive the meter. Essentially what we want to do is take the pulses each set of points, condition the inputs, add the two pulses to give us 2 pulses per revolution. This should work with a 4 cylinder 4 stroke tach. the only problem is the dwell time is different. Perhaps we could use an integrator to give a longer duty closer to what a 4 cylinder 4 stroke would produce. I'll have to do a little experimenting and see what I come up with.

[byacey]

I saw some tachs down at Princess Auto for around $30.00. I'm going to go pick one up and see what I need to do to get it reading correctly on a 2 cylinder 2 cycle. I'll let you know what I find.

[rodbolt]

asa dran it wont work at all. the magneto signal is grounded . try a redraw and repost. most outboard tachs use an AC drag cup type movement as most use an AC signal from the rectifirer or tach signal generator in the ECM/CDI. I think the FICT motors still use a squre wave fequency output that converts hertz to meter movement. but with the magneto signaal hooked directly to ground as dran the mag will see it as a shut off signal and wont fire. and you will find ,if you use a peak reading meter you will have a lot more than 12 volts on the coil primary circuits. but most outboard tachs require an AC signal. as the number of magnetic poles rotate faster around a stator the frequency and voltage increase thus making the meter movement move in a linear motion. man I gotta get my books back out

[byacey]

what is dran?

[lark2004]

Thanks for that Paul, I'll throw that together and do some experimenting. One of the girls at work made an analogue tach for a asighnment at university, We may be able to get this work with that. I don't know about using most comercially available tachs for the obvious reasons already pointed out. Personally, I want a digital tach. Way more accurate than a needle.

[rodbolt]

dran is how you spell drawn with a cat in your lap while typing one handed good thing they pay me to fix it not spell it

[byacey]

I think Paul just has the labeling mixed up. Block ground and tach ground would be the same. The points signal is what would be showing up at the base of the transistor. The waveform shown indicates the signal off the collector of the transistor. The most you will ever see here is 12V, because that is the maximum collector voltage when the transistor is cut-off. The base voltage is limited to about 2.5 V because of the clamp diodes on the base, to avoid destroying the transistor by exceeding the maximum base emitter voltage of the 2N2222. This circuit is basically an inverting buffer / signal conditioning circuit that would then in turn drive a tach.

[rodbolt]

look at the magneto signal line. its third point is a ground. this would also ground the magneto coil primary and create a no spark condition on that set of points. also with Diode D1-4 tied to ground and the base of Q1 tied to ground and the collector of Q1 tied to 12V you would have a thermal runaway. you would need a base bias circuit or at least a Zener circuit to control voltage and current through Q1. I just dont see that circuit as being able to work. it also does not take into account the amount of "ringing" a magneto generates. i think that diodes 1-4 are trying to depict a full wave rectifier but its not the correct symbol. I have enough trouble trying to remember that in the "civilian world current flows from positive to negative and in the electronics world of the military its negative to positive, so excuse me if I get confused sometimes. but the way its drawn R2 is useless as the fwd biased resistance of any of the diodes is considerably less than 100K ohms. and the total resistance in a parralel circuit is always less than the lowest resistance in the circuit. we also need to knpw the input signaal voltage on the magneto signal line to prevent a reverse current breakdown on the diodes. but it is fun to try to remember basic circuits. dont get a chance to do it much anymore. but if D1 ever fwd biased the current flor from the collector to base on Q1 wilkl be rather large. a 10K resistor tied to 12v positive wont drop much current.

[byacey]

The transistor is not forward biased, so it will not conduct until it receives a positive going pulse from the points - ie. when they open. R1 creates a fairly high isolation resistance so the clamping action of the diodes does not load the magneto coil when the points are open. The diodes are not a bridge, D5 prevents the transistor from being reverse biased and breaking down the junction. D1 to D4 limit the voltage between the base emitter junction to about 2.4V. The Resistor R2 further limits the signal by forming a voltage divider with R1, and prevents the base from floating - this creates the bias, or lack of to keep the transistor from conducting except when a positive pulse appears from the input (points). C1 creates a low pass filter to remove any high frequency noise and ringing that inevitably will come off the magneto coil. What you see in the scope shot is what you will see on the collector of the transistor. If you don't think it will work, build it and see for yourself. By the way, if this uses an isolated 12 V supply (no common ground with the engine) the labeling as shown will work too.

[Paul Moir]

Yeah, byacey's got it exactly. There's a heck of a lot of protection just in case of arc-over, etc. You can ground out through the 220Kohm resistor and that won't affect the magneto's output.I have to redesign that so it doesn't use an isolated 12v supply (which it does, and my "ground" symbol is erroneous because that is not the engine ground, but mearly the negative on the 12v supply). The pricipal pulse from the magneto is negative. I guess that means going to a PNP transistor right?The magneto rings at about 15KHz. C1 seems to take that out nicely. Due to my ignorance and inexperience, most those values were determined experimentally. My capturing 'scope' is pretty primitive too.I'm after the needle tach for looks alone. Course, I'm striving for simplicity here to make it easily reproducable, so using an IO tach rather than an outboard tach helps a lot, since we don't need a PLL or other form of frequency multiplication. I haven't had much time to play with this lately. And I need an I/O tach to play with too. I might have some time sunday though. Glad to see you're all interested though.

[rodbolt]

it wont the points will be grounded by the junction on the magneto signal line, unless a decoupling device is used. if the points are grounded you wond have a broken circuit when they open and no collapsing field on the coil primaries to create a secondary voltage. an when d1-d4 are biased R2 is negated. you may get some spurious signals but nothing really usable. so would it drive a digital frequency counter? maybe, but its awful complex for a tach signal on a small outboard boat. but in the drawing above the 12v power supply is grounded with the engine ground again at the point on the magneto signal line.and why bother with the complexity involved when several lawn and tractors use a decent tach without the complexity of isolated power supplies and decoupling devices. someplace between R1 and the 12v positive you have to have a commom ground. otherwise there will be no current low. we all know a transitor is a voltage controled current controling device. I still dont think it will work as drawn. not for anything usable.

[byacey]

Rodbolt, I don't think you are getting it. The most load the magnetos will see is 220K, just about a quarter of a million ohms. This isn't going to affect the magneto operation.Paul, I think an NPN is fine, just the duty cycle will be inverted. I'm going to look into this some more.

[Paul Moir]

I don't intend to use decoupling devices nor isolated supplies in the final design. What I've pictured is simply what I've got at this point. I do intend, with the help of all you here, to go further.The schematic pictured was constructed, and the signal pictured was the result. Forget about will not/will work - that is the output. byacey - Ok, I'll work on this a little bit more tomorrow. I think I get it, just forward bias the transistor so the negative pulse shuts it off?

[byacey]

That's basically it Paul. I would bias it for maybe 5 to 10 ma collector current as a starting point, and increase maybe to 50ma if there is false triggering from noise getting through. Or you could go with the PNP transistor and have the negative pulses switch the transistor on like you mentioned. One method will give a low duty cycle "on" time, the other a low duty cycle "off" time.- The tach will probably favor one of these over the other.

[Paul Moir]

I finally got this one sorted out (I think!). Here's the schematic:Click here.I got rid of the safety diodes on the front end since I'm raising the voltage at C1 to about 3 volts. I don't think they were really needed anyway. Perhaps I should drop in a zener though. I also got rid of the transistor since I had an extra op-amp left over and so that saves a part. Let me explain a bit. The voltage at C1 is about 3 volts and it drops slightly when a negative pulse from the magneto arrives. This pulse is detected and amplified by U2d, which is set up as a schmitt trigger firing around 2.2v with about a .3v window (depending on the battery).You would think the output would be fast enough to satisfy the rise/fall times for the CLK on the 4013, but it's not (at least on mine!) so I'm using half of the 4013 as a buffer to square up the signal more. The next half of the 4013 flip-flop divides the signal by two, producing a 50% duty cycle square wave at 1/2 the magneto fire rate (ie, once per rev). The square wave is then converted into a trianglish wave with R6/C2, with the centre of the wave at 1/2 supply voltage (averaged) and the high point of the triangle occuring just as the square wave drops from high to low.Now, the triangle wave is "compared" with 1/2 supply voltage (R7 and R8, with U2a doing the comparing) producing a square wave that is 90° out of phase with the square wave coming from the 4013.U2b and c are set up like comparator gates to XOR the two square waves, producing one with 50% duty cycle that is twice the frequency of either - which is to say at 2xRPM, exactly what we need for a 4 cylinder I/O tach.Up in the top-right corner is a power supply for the thing that will protect it incase of overvoltage. I guess since that can never happen if you don't have a generator, it's really unnecessary. The circuit shouldn't be dependant on battery voltage (ie, needs no regulator), but the 4013 will fry if it sees more than 16v.R6*C2 is a bit of a compromise - the higher the number, the more triangular the wave. But the higher the number, the smaller the signal as the frequency increases (RPMS come up), and I didn't want to be amplifying too weak a signal. It gets pretty distorted around sub-idle speeds causing the duty cycle of the output to vary a bit, but never less than about 33% at 300rpm, which should be fine for the tach.Right now I haven't tried it with a tach - I'll order one tomorrow. All I really have is some nice blinking LEDs. The output 'stage' will probably need a little work and I'm waiting until I have the tach in hand before I fiddle with that.I've been trying to keep the parts count to a minimum, which is why I eliminated the transistor. I had a spare OPAMP, and half a flip-flop as a buffer. Also I'm trying to use only easily found components. Although that's made for a pile of resistors, I think it's still pretty reasonable & buildable.

[byacey]

Hi Paul, you've been busy. So far I picked up a tach (which didn't have any instructions or diagrams in the box) and fed it with varying duty cycles from a function generator to see if the tach reading changes much with duty cycle. So far it doesn't appear to be to picky about what it sees. One thing I would do is consider building another parallel buffer stage using the other unused LM324 portion to isolate the magneto signals from each other. With the amount of spikes and ringing off the coils, it could short or zener the diodes on the front end and leave you out in the middle of the water. I would also put a couple of back to back 3.6V zeners across C1 to protect the input of the opamp.

[Paul Moir]

OK, will do. I was thinking about the destruction of those two input diodes as well. What I thought would be a simpler solution is simply to run two 220K input resistors into C1, one for each diode. That way if one or both diodes failed it wouldn't affect the other (or anything, since 220K doesn't significantly load the magneto). The only reason I didn't implement it is because I've been testing with only one magneto line and adding the resistor would mean I would have to recompute R22,23,24. Just lazy really.

[byacey]

Yep, you're right, no sense adding more stages than what are necessary.If I get a chance I'll whip up a PC board and do some testing with this tach that I have, A Princess Auto special.Do you have Princess down in Nova Scotia? This is a good resource for surplus materials and parts.

[Paul Moir]

Yes, they finally arrived here a couple years ago. Couldn't believe the place when I first walked into it. I wasn't the only one either - about a year later they doubled the size of the store; it was always packed with customers. Only wish they had more marine & stainless stuff and less farm stuff. And a metal lathe...I was looking at the tach that was on sale, then I got looking at the new $45 laser tach. But I really needed a proper one for the boat so I gave it a miss and ordered a good looking Faria.

[Paul Moir]

Mini-update. Tach arrived and I got some time to play with it this evening. It doesn't seem to care much about duty cycle either. So far I've only tested it up to 1700RPM; the most I can get on my 4hp spun with a drill. The plugs out and the L/U is off. But I get the same reading on the tach if I connect at the output of the detector as I get if I connect to the 50% signal. Looks like I'm in for a dramatic simplification!

[byacey]

That's good! Most Auto hand held RPM / Dwell meters don't seem to be affected by the dwell between 4 / 6 / 8 cylinder engines, so I didn't think it was too critical.

[dafox99]

Hey guys .. while we're all doing our two cents .. Three thoughts come to mind:On the input, the LM324 can take high differential input voltage differences, but you can't drive the inputs below Vdd .. in this case ground. I think you are fine there (don't need a clamp). I don't know ANYTHING about a magneto output, but the 220K should isolate you just fine as the input bias current is about 45 nanoamps, so it's fairly impossible to hurt the input.The second thought on the input is I don't think I'd let it drift. Again, don't know anything about the nature of the input, but if is is some coil with a magnet going by, then it may be some sort of current pump with a fairly high output impedence. Have you scoped pin 13? It seems like we are relying the bias current from the op amp to bring the voltage back up after the mag brings it down? If it is sloshing around (with respect to the input) it might make sense to set the voltage with a ladder just like you did on pin 12. I guess my concern is is the filter cap is making pin 13 slew at a much different rate than the input. Which brings us to my third thought:Man, that is a pretty fast amp. 50v/uS by my spec. This means you should have had a pretty square wave with very sharp slews. less than a 1/4 uS to slew 12V. You shouldn't need the flip buffer. No big deal, but clearly something is amiss. You aren't driving a capacitive load, is Vcc current limited in any way?Just thinking out loud ...Dave

[Paul Moir]

Thanks for the input davef! I don't know what was going on with the CLK input of the 4013 - maybe the one I have on hand here is plain broken. I thought I had a few but when I went looking there was just one left. Vcc is current limited by the power supply protection parts, but I tried bypassing that (along with a bunch of other things) to get it to switch properly. The wave I captured at pin 14 appeared very square, but the capure equpitment I have is somewhat primative.The detector OP AMP is set up as a bit of a schmitt trigger to help cancel out any noise. The output of the magneto has a lot of ringing and needs (IMO) a little more than simple filtering. I'll try to get a pic of the output of the magneto up so you can see what I mean.But now I think I'm back to squre 1. If the tach doesn't care, most of that circuit is pointless. I think if I'm just amplifying and cleaning up the signal, then I may switch back to discrete transistors since that'll simplify the power supply. (wait a minute though - I just have to loose the 4013 for that. Hmmm...)Right now I have a perfectly good working circuit, tested up to 1700 RPM. It's just overcomplicated.

[dafox99]

Maybe just a emitter-follower or darlinton off the op amp? Also, again, I don't know anything about magnetos, but I would imagine this is a pretty noisy environment electronically (including ground). If the output is differential, you might try twisted pairs and possibly isolate the ground of your circuit. If it is single-ended, you might try some ferite beads. Noisy environments, especially when ground is noisy as well are sometimes more art than science. If ground is very stable (check case ground to different ground points in your circuit with a scope), then eliminating noise is a bit easier. In general, lower the input impedence of your ciruit as much as possible without swamping out the mag .. that's what you are doing at high frequencies with C1. In some cases, WHERE you put the components matters as much as their value if you have a noisy environment. Many many years ago I designed electronics for targeting systems in tanks. We'd spend hours hand-building a high gain amp (that was amplifying EVERYTHING we didn't want to amplify), and would move twisted pairs a couple of inches, and then relocate a cap on the board and tah dah! I found you can pretty much build one of anything .. it's building 100 a month by little old ladies that is the trick. In general, put filter caps on all your chips and inputs as close as possible to the chip. Well, enough rambling .. looks like you clearly know what you are doing. Nice work!