What I want to do is have 1 12vdc square wave input, with a "splitter" that will alternate between 2 outputs, I want to do this electronically, and not with a Hall type sensor going high or low to reset

I thought I knew the D-jet system pretty well, what would this be for? Edumacate me.

You trying to simulate trigger points? Sounds like you're messing with that new RUNNING car already.

I don't believe the D-Jet really cares which pair of injectors fires on what cycle, just so long as they alternate. If that's so, it makes the problem a lot easier. If the D-Jet runs like crap with the signals 360d off, however, you'd need more than just a plain square wave in to make this work.

Assuming you don't have to worry about position, just alternating, you'd need something like a flip-flop to provide the alternating, and I'd use a pair of transistors (one for each output). There are some unanswered questions like what's the dwell on the trigger points, and does it matter? Do we need a variable-width pulse based on engine speed, or is just a fixed-width pulse good enough? No idea what the D-Jet ECU does with the trigger points pulsewidth, if anything.

http://www.fairchildsemi.com/pf/MM/MM74C374.html

Maybe this one would work? I think one flip flop would just turn one output on if the input gets a pulse, and off if it gets another pulse, and so on.. But this puppy has 8 of them, perhaps some combination of them along with an inverter might do the trick? Sorry, I'm not an EE, nor do I play one on TV.

This thread is useless without more data.

A square wave can be easily generated, then add an inverter as the second output.
The next question is: what frequency?
Then there are a whole bunch of other questions.

Can't be for injectors. 12V is too much for them. Also pulse length is critical. The computer takes a basic pulse train and modifies the hell out of it according to sensor inputs.

I *THINK* he is trying to modify a basic tach signal to feed to the djet computer so he can run a non-stock distributor, or something like that. That would mean the frequency would vary from about 15hz to about 100hz. The issue of dwell is probably important, the ECU may not like it if one signal was always on, and just alternating. Pulling it off would require an intimate knowledge of the djet ECU. I don't even know what voltage the trigger points are switching. If dwell is important, then it could get complicated really fast. The on time would have to be adjusted based on frequency, which can change. I agree, more info is needed

yes to a bunch of the questions

trying to eliminate the trigger points in the base of the distributor for the D-Jet

with the L-Jet, all 4 injectors fire when the coil collapses, with the D-Jet, there is a cam lobe on the dizzy shaft that in normally open, right before the intake opens, the contacts touch, sending a small 12vdc signal to the ECU, the ECU then grounds the injectors firing one pair (every ignition cycle one or the other contact closes completing the loop)....very small amperage amount from what I have read, less than the contact points driving the coil.

not sure if the ECU cares about dwell or not, it could be as simple as on/off, same as the L-Jet, as long as there is a voltage differantial, the trigger works

I think there is a small amount of time when both triggers are being switched.

Interesting project.

Maybe, maybe not. The whole D-Jet system is basically too clever by half, and obvious designed by someone who knew analog voodoo much better than I do (which ain't hard).

Without a good scope and lots of time, there's no real way to know. I'd think that you'd be right, and simple pulses would work. Only the frequency really matters. However, I'd also not expect the MPS to work the way it does, either.

The square wave generator and inverter idea might work. Feed the input square wave into a divide-by-two frequency divider (crib circuit from one of many Radio Shack handbooks), and pass that signal straight to injector pair A, then invert the signal and send that to injector pair B. This assumes the square wave in has a pulse every time a spark fires. Where are you getting the square wave from? If it's the coil trigger, you might have a problem with the flyback voltage, just like the MS tach input.

You might want to look at my analysis of the D-Jetronic ECU:

http://members.rennlist.com/pbanders/ecu.htm

I'm not exactly sure of what Mueller wants to do or why. The trigger contact points are probably one of the more reliable parts of the system, you only need to change them every 100K miles or so. The failure mechanism is usually not the contacts, but the phenolic blocks that ride the cam wearing down so much that disrupt the timing of the two switches and the overlap period.

I think he wants to run a distributor other than the D-Jet unit, like a Mallory Unilite.

I'd looked through your site quite a few times, but don't remember seeing so much data on the ECU before. Looks like only the opening (or is that closing?) of the trigger points is used, so dwell isn't accounted for. Or I completely misread that (which would be easy). Lots of data there.

The trigger points create a fairly short pulse. Shorter and closer together as the RPM rises, and of course timed to the crankshaft rotation. I'm not sure how to create a signal EXACTLY matched to the RPM without a direct link to the crankshaft. Trigger points, ignition rotor, crank position sensors all are linked to the crank. Think how crappy the engine runs when the distributor is not timed correctly, or the dwell is not correct. The factory way is simple and reliable.

Now that I think about it, the two trigger points have one lobe, the ignition points have 4. So to youse a tach signal, you not only need to flip flop between two outputs, but you aslo need to divide by two. Also, aren't the trigger points set up such that the injector opens when its intake valve is open? I think that would be impossible to get out of a tach signal.. if you could get that, it'd almost certainly involve a dead cat.

I know you wanted to avoid a hall type sensor, but what if you put an induction timing light sensor on the plug wires for cyls 1 and 3 (or 2 and 4, whichever is closer to what the trigger points do).. then you can make each bank trigger at the correct time. Eh, nevermind, my cheesyness alarm just went off. There's no way to make such an installation elegant

Nope. The injectors open two at time with the fuel ready and waiting for the valve to open.

The trigger points only tell the alternating pairs of injectors to turn ON. Thats all they do.

The ECU then calculates the pulse width based on the other sensor data to turn them off.

Leave the deer killer with all four corners on the ground Mike...

HEY YOU THERE...YES YOU...PUT DOWN THE TOOLS AND STAND AWAY FROM THAT 914....

Unless of course you are going to drive it....

We love you man....

Huh, I guess you'd have to do one at a time to inject while the valve is open. Why the extra complication of doing two at a time and not all 4 at a time? Too much fuel pressure drop?

It sure is fun to think about this stuff

Royce, I try not to think why. It just confuses me.

James is correct...Mallory Unilite distributor, has no "stock" method of triggering the trigger points

Jeff...thanks for the confirmation on the trigger points, I took them as just a simple signal for the ECU to fire the injectors

good news...my wife likes the new 914, wanted to go out to dinner in it until I mentioned "no heater"...I'll have to fix that very soon

why not just put a pick up on the fly wheel to trigger the signal, making a circuit to transfer a hall or VR sensor to make or break a connection is easy. the hardest part is getting it to only do it on every other rotation . or you could use a pick up in the dizzy, that you add in. i would assume that the points are 180 out of rotation to each other. it could be as simple as 1 very small reflector glued to either side of the shaft or rotor with two small optical pick ups ( basicly avalible as one led looking thing ) either drilled in to the dizzy body or cap, this puts the pick ups 180 out of phase if the pick up are 180 apart.

a little harder ( but not much ) to do would be as some one said install a pick up on one spark plug wire to trigger a circuit to average the signal and trigger two transistors per spark or two pick up and just build 2 seperate triggers. and it could be done with 3 small wires to the spark plug wires. no large clamp like a timing gun needed, just a loop or loops on the plug wire

any of the above systems could be built as an add on board that the injector points wire just plugs in to. and it could be sold once worked out. i think price to make this board would be 20 to 30 bucks

After thinking about this for awhile, I'll have to call it Far-To-Complicated-For-Not-Enough-Payback. I wouldn't be surprised if there was a performance hit because of the D-Jet being "tuned" to the advance curve of the stock distributor. I say, if you're gonna switch something out, get a KitCarlson and have fuel and ignition in one unit. I like you're thinking on this one, but it's time to start engineering something different. If you want some ideas of stuff to engineer, let me know. I've got 8-10 crazy ideas that I'd love to get some help on.

Mike

It looks to me like you need to find the right end of the mule. Drive the car with the D-jet and trigger points of change the FI and ignition.

Enjoy the car while you get your MS going. Then swap the to the Mallory and MS. If it takes a year you still have a car you can drive.

Dave

i dont think the trigger points are tied to the advance plate, i could be wrong because i dream in ljet

Hi Mike, This is were I say :finger2: :finger2: DO NOT MESS WITH THIS CAR! :finger2:
Just drive it.

Mike

I'm not saying the advance is tied to the trigger points. I'm no even talking about the trigger points.

To put it differently, what happens to D-Jet when you drastically change the advance curve? Do you then have to screw with the MPS to retune it? I would bet you would have to if you wanted to actually make more power.

1st off...my car is still running and I have not mechanically molested anything

Brad, do you have a BOM for the ECU schematics that are on your website??

D-Jet ECU schematic...information overload...I \"like\" it

another question for you EE types:

what happens if I supply a clean sqaure wave signal to a debounce circut? will it work or will the debounce circut be looking for a ripple?

You'll get a square wave with a delay. A debounce circuit will look for an "on" signal (pulled high) for some period of time before generating a square wave out. The idea is that a mechanical switch will have noise and the beginning and end of the signal (and maybe in the middle, too). Said noise may pull the signal all the way below whatever you're considering a threshold for "off" or "low" (often well above 0v), which can look like two pulses instead of one. By delaying across the likely time for this transition, it will only "see" one pulse, and only send one pulse out.

The length of time is the crucial bit. For a human-operated switch, this can easily be several milliseconds, since two deliberate up/down/up/down cycles will nearly always take 100ms or thereabouts. Humans are slow.

For mechanical points, the bounce time will depend on the max engine speed. For a four-cylinder running at 10K rpm, you'll get a pulse every 1.5ms, so the debouce delay needs to be short, probably around 0.5ms for something with loose timing requirements like injection, but a lot shorter (like 0.001ms!) for ignition if you don't want to skew timing by more than a degree or so and be able to run at high revs.

A digital switch, like a Hall-effect, optical, or VR sensor, won't need debouncing. These should generate a nice, clean waveform (sine or sine-like for VR).

the stock ECU has a debouncing circut that the contacts feed into...so it would seem like I'd need to go from digital (the dizzy) to analog (to mimic the "dirty" signal of the contacts) unless a digital signal can be made to "fake" it...

This maybe be stupid or brilliant. Can you somehow "record" the waveform and "play" it back synced with the tach signal?

okay you EE types out there.....

want to check out this circuit?

There will be 2 opto's each with a "Load1" going to Pin 21 and Pin 22 of the ECU to trigger the injectors.......

The opto isn't useful in that circuit. It's not isolating Q1 from anything coming down the left-hand +12.

I assume the left-hand +12 is some sort of switch signal you expect to be nasty, and the upper +12 is constant power. If so, then run the collector of the opto's transistor to the constant +12 up top. Be sure the opto's ground is run back to the ground shared by the nasty signal generator, whatever that is. The idea behind an opto is you have a clean power source and a nasty switching signal. The signal goes on the LED side, and the clean power source on the transistor side. The signal switches the clean power on and off, so you get clean power "out", to the thing you're isolating. As drawn, you'll get the left-hand +12 running straight to Q1, using whatever ground potential exists between Q1 and the switch signal. The opto might as well not be there at all.

What's to the left of this circuit?

this circuit was posted on an electronics forum....I asked a question and this is what was replied...totally clueless whether it would work or not

Left hand +12 is from the vehicle power source (ignition activated to turn on)

Here is the typed response along with the circuit:

QUOTE

Hi, I put a circuit with this to perform your function. Any interrupter optocoupler or combination of IRLED and phototransistor will work for the detector. Also, about any N channel hexfet (or power mosfet) will sub for the 2N700. If you get the interrupter, it should be marked E +/- and D +/-. That's Emitter and Detector with polarity marked. For reliable operation, the IRLED needs to run at about 50 mills. It doesn't matter to the IRLED, but the phototransistor has to have lots of photons to go into conduction. If the collector of the phototransistor doesn't pull low enough, increase the value of R2. In operation, the phototransistor will be on with the base illuminated. The collector voltage will be close to ground, keeping the hexfet off (no current through the load). When the illumination is interrupted, the phototransistor will turn off. The collector voltage will rise to +12, and the hexfet will conduct, allowing current through the load.

OK, I understand the explanation, but not the circuit, still. If both +12 inputs are constant power, then where does the "interruption" come from? If this isn't an optocoupler (aka optoisolator), but an optical sensor with a shutter between the LED and the transistor, then it all makes sense.

The OPTO1 would be either a reflective or transmissive optical switch or interrupter module. These are an LED and phototransistor pair in a plastic case, and the light beam is either interrupted with a slotted disk or similar device on the rotating shaft, or small reflectors on the rotating shaft bounce the light beam from the LED into the phototransistor. It wouldn't make sense to use an optoisolator, these are used to couple signals between two electrical circuits with different ground references.

The circuit would work, whoever posted is just saying that other power N-channel FETs would work, it doesn't have to be that exact one. The LED needs to have 50 mA current but the spec sheet for the part would tell you how much current. You pick a resistor value to give you the right current in the transistor. Use the biggest value to allow the FET to turn off and to keep the power dissipation down. The spec sheet will call out the optimum phototransistor current. Here's a spec sheet for an example optical switch.

Just tested the circuit on my workbench...it works with the following components:

Optical Interrupter Switch

and

FET

only reason I used these is that we happened to have these sitting in a draw...we used to stock more, but we sold off most of the small components since we don't buildup PCBs anymore....time to do some more research and start ordering parts....

I need to add:

the circuit works as is, I have NO idea if it'll trigger the ECU and ground pins 22 and 23 as needed....