Speedup cap - DEcreases injector opening time?

[Philip Lochner]

One approach with low impedance injectors is to install a suitable resistor in series with the injector which limits the max current through the injector and then configuring MS for high impedance to avoid having to use PWM.

I have found using this approach that this also increases the injector opening time and also results in not all the injectors having the same opening times (from 1.5ms to 1.8ms in my case) which then result in different cylinders having rather different AFRs particularly at idle.

Someone mentioned the idea of putting a capacitor in // with this current limiting resistor to effectively short the resistor out for 1ms or less ensuring the fastest possible (and similar) opening time. I can see it working for opening the injector but I'm concerned about this cap having enough time to discharge while the injectors are closed.

Has anyone adopted this approach? Can it work? If so, how to determine the value of the cap?

[Heribert]

Yes , in theory this idea works , and in low power applications in electronic circuit design it is done all the time .

However , to install a capacitor in parallell over the series resistor
( of appr 6 ohms ) to drive an injector in what effect would be a peak and hold mode ( appr 14 mH /2 ohms I think )would require a very large capacitor .
I do not think it is a practical solution and the closing time would also be increased .

I am surprised about the spread from 1,5 to 1,8 ms , I tried out a set of BOSCH loimp injectors ( from a VW 411 , 1970ish) and opening time was stable and identical to within 0,1 ms .

BR

Heribert

[Philip Lochner]

appr 14 mH /2 ohms I think )would require a very large capacitor .

I am surprised about the spread from 1,5 to 1,8 ms , I tried out a set of BOSCH loimp injectors ( from a VW 411 , 1970ish) and opening time was stable and identical to within 0,1 ms .

Thanks for the reply Heribert. In my case, I have 3 injectors (3.91mH, 2.5Ohm each) wired in parallel into a single 5Ohm resistor.

In my own crude method I come to a value of about 67uF for the cap which is not too bad? Please bear with me for a moment, then I'll explain how I got there. (Relevant spreadsheet attached.)

First I calclulated the current through an injector as if it is shorted directly to ground. Then I calculated the current as it would be, through the 5 Ohm resistor. The idea is that the cap should be large enough to absorb the DIFFERENCE in current between these two curves, for ONLY 1ms. Using 20us intervals, I calculated the instantenous current for the 1st 1ms. Since Q = integral of current * dt, I then multiplied the DIFFERENCE in current with 20us and added all these together to find the total value of Q.

SinceC = Q/V, dividing by 18V (the voltage value of the cap - should this not be 12V?) got me to 67uF. Would this approach be valid at all?

I don't see how this cap would increase closing time. Could you explain?

I am more concerned about this cap DIScharging itself before the next charge cycle as surely the only path for discharge would be through the 5 Ohm resistor? If it does not discharge fully it will end up being almost fully charged and hence not serving any purpose at all?

[jakobsladderz]

I plugged a 1.95mH, 0.83333R 'injector', equivalent to your 3 parallel injectors(both resistance and inductance reduce with parallel connection), into a sim package (switcher cad from linear devices, its free to use) and found one thing that might be of interest.. normally you would aim for a 1 amp per injector hold current. with a 5 ohm resistor you are getting about 2/3 amp per injector.. that may be causing some trouble with opening times.. A 3.3 ohm resistor would be better (just under 3A total with a 12V supply..). A 200uF capacitor (with probably unrealistic 1 ohm ESR) builds current to a peak of 3.6 amps for about 3-4 ms before dropping back to 3 amps. discharges sufficiently in the 2us that would be spare at 80%DS and 10ms per pulse (1 per rev, 6000RPM). Probably not worth the effort for such a small current gain though, less than 200mA per injector..

Cheers,
Duncan

[Heribert]

Hi Philip .
Have not checked your calcs , but they are probably correct .
However as to closing time . I may have been wrong on this .
Thinking a bit further below
In injector 0,5*L*I*1 mode , I visualize the "flyback energy" path as follows .
Injector lower terminal ( ie at MOSFET Drain) , a 39 V zener to power supply + and then on to the series resistor with the cap in //.
- on cap goes to the injector upper terminal .

The Lentz Law discharge current will be determined by the voltage drop in this circuit .
And in addition to the Zener drop there will be in series a capacitor charged at some 8 V , so flyback injector ternal voltage will increase
and flyback current will decrease , so it might in fact close the injector marginally faster . The capacitor voltage will mainly discharge thru the
resistor , and tau will be in the region of 400 microsecs, so there may be some importance .
Will try to take some time this weekend to make a bench test on this and see if my meandering thoughts are relevant .

But , as a general idea , I do not favor 50-100 microF electrolytics in this kind of application . Prejudice rules!!

Best regards

Heribert

[zsilinszkyz]

Have you realised that you have less than 2V on your injectors when you connect them like you've described? Isn't it possible, that this is the cause of your problems (I suppose you're injectors are rated for 12V)?

Regarding your calculations, you've ignored the effect of the injector's inductance, which might also cause ringing. There are circuit simulation programs that could be used to simulate the response of the circuit. For instance there's a free version of Tina from TI.

Zoltan

[Philip Lochner]

Have you realised that you have less than 2V on your injectors when you connect them like you've described?

I think there is something you don't understand Zoltan. I've attached the setup I have.

[Heribert]

with a 5 Ohm series resistor is likely to be the cause of spread in opening time between injectors .
Zoltan is absolutely right.
Series resistor 5 ohms . 3 // injectors , each 2,5 ohms, 3,9 mH . Supply 12 V
Injector steady state voltage ( >90%), which will occur after some 3,2 ms ( L/R = 0,78 ms) will be 1,7 V , equalling 0,68 A per injector is probably on the lower side of what the injectors require for accurate , fast opening .
You will do well to reduce the 5 ohm series resistor for the 3// group to 3,5 ohm . Even better with individual resistors , I think ( appr 6 ohms each)
If you have a good scope , you can study the opening time very accurately by measuring the injector current rising flank . To get sensible results , use only one injector with a variable series resistor ( which you can also let do double duty as the curent measuring shunt)
Scope the rising flank , which will be an exponential curve where the rise time will be determined by the circuit L/R ( R will be Rinjector+Rseries resistor +Rsource . R source will probably be insignificant , at least if you have a good power supply or a car battery)

Observe the rising current trace carefully , and you will find a quite small superimposed exponential curve with an amplitude of maybe 50 mA and a duration of < 100µs. This "quirk" occurs when the mmf generated by the coil is high enough to lift he injector pintle from its seat ( counterforce by a spring and fuel pressure) and travel to its open position . The reluctance of the magnetic circuit changes during this operation and Lenz law results in some " driving current" distortion. This takes place at each opening operation and opening time is determined by L/R and required force ( ie current)
And if the driving current is just about enough to reach the required mmf
the spread in opening times will be quite large .
Also closing time can be measured the same way .
I have dug into this to some extent , and on Eric Fs homepage ( not too fast?)there are some curves with explanations

To get good results from your install , determine the required opening current of your injector , and make sure that you aim for a series resistor that gives at least twice that current at 20 Centigrade . This in order to have a fast risetime of the current at the opening current value .
Injector coil is copper wire , and will have appr 20 % higher resistance at 75 C , and copper coils are often allowed to operate at even higher temps , maybe even 95 C .

Longwinded explanation , soory abt that , but hope can be of help .

Heribert

[Philip Lochner]

with a 5 Ohm series resistor is likely to be the cause of spread in opening time between injectors .
Zoltan is absolutely right.

This has been my conclusion as using the scope over this 5Ohm resistor I could actually see 3 little dents in the curve which I presume is the 3 injectors opening. This is also how I concluded that opening times were very slow and ranging from 1.5ms to 1.8ms.

To get good results from your install , determine the required opening current of your injector , and make sure that you aim for a series resistor that gives at least twice that current at 20 Centigrade .

Well, I did not follow such a good engineering approach but I did wire various resistors in // to the 5Ohm and I could see how the opening times became less and closer to each other until eventually at 1.4ms, the 3 "dents" in the trace had merged into 1. This was at the point where I had a 2Ohm res in // with the 5Ohm giving 1.42Ohm effectively. This is also how I'm driving the car now without any problems... yet.

The reason for posting this thread is that I am a little concerned about the heat dissipation in the injector as I have not yet activated PWM. I am deliberately trying to avoid PWM due to the problems experienced (and solved) by Justin (Turbo4V). (although I have tried it a little and found that something goes wrong quite quickly). So, I thought, if the cap could work, I could keep the 5Ohm res (to keep heat dissipation low) and use the cap to speed up the opening time. Also, I'm trying to avoid building a new injector harness by finding a solution with the existing which has this "3 inj into one res" setup.

The original ECU has an additional injector driver connected between the injector and the 5Ohm resistor which shorts the injector directly to ground for a short while allowing the injectors to open quickly and then the driver through the 5Ohm res takes over holding it open.

It seems my "gut feel" (that if there was merit in this cap approach, someone would have done so by now) is being confirmed .

Thanks for your input chaps. Again, the forum has demonstrated it worth!