MSD Coil Ballast Resistor Needed?

[rick-l]

I forgot the 0.6 ohm resistor. That drops it a bit -- even more amazing.

80.108 mJ

3.6 Watts avg at 900 RPM
16 Watts at 4000 RPM

Only 66 Watts peak

Edit: That was just a LTSpice simulation. I don't know specs of the Darlington driver so it is just a modern transistor.

[scarceller]

But are you still saying that the coil reaches saturation in about 2.4ms at 14vdc? that's not what I see in car or in bench testing.

[rick-l]

Originally Posted by scarceller

But are you still saying that the coil reaches saturation in about 2.4ms at 14vdc? that's not what I see in car or in bench testing.

If you add the 0.6 ohms in series with the inductor it hits the current limit at 3 mS, at 0.9 it takes the entire 4 mS.

[scarceller]

And the darlington transistor in the DME has two 0.1ohm resistors in parallel from ground to the emitter, that adds another 0.05ohms, right?

Your calcs based on an assumed inductance of 3.6mH still indicate that the coil may actually be/have a higher inductive value, right?

I know for sure that at 14vdc that coil hits saturation at about 4ms.

[ischmitz]

I think the SPICE simulation completely omits the fact that the Darlington transistor has a considerable voltage drop across its NPN junction. No way the coil ever "sees" 14V. What does the o-scope read when you measure between GND and pin 1 of the DME ?

[rick-l]

Sal
I was watching your video again and is the bottom trace that you refer to as ground through a resistor so it indicates current?

I am not simulating inductor saturation. This is where beyond a certain flux density the permeability of the core is reduced and causes the inductance to drop. I don't have any data to predict that.

I'm just looking at where it hits the current limit that I modeled in spice with an ideal inductor.
Below is current with 0.8 ohms in series with the inductor in brown, power in red (50 mJ integrated) and the collector voltage of the cobbled together Darlington in blue.

If anyone wants to play with the LTSpice file (.asc) PM me your address.

[mysocal911]

Originally Posted by rick-l

Sal

I am not simulating inductor saturation. This is where beyond a certain flux density the permeability of the core is reduced and causes the inductance to drop. I don't have any data to predict that.

Yes, you've forgotten that in your simulation, i.e. a very critical parameter!

Once saturation of the core (the point at which the flux no longer changes), the voltage
across the coil goes to zero and the darlington driver in the ECU 'sees' the battery voltage thus entering a high power mode, but the current will be limited by the internal current limiting circuitry.

[mysocal911]

Originally Posted by ischmitz

I think the SPICE simulation completely omits the fact that the Darlington transistor has a considerable voltage drop across its NPN junction.

Not really, probably about 1.5 - 2.1 volts.

Originally Posted by ischmitz

No way the coil ever "sees" 14V.

Correct.

Vcoil ~= 14.0 - Vsat (darlington)

[mysocal911]

Sal should be able to easily determine the coil's saturation point by incrementally increasing
the dwell time while monitoring pin 1 of the DME ECM uses an EPROM emulator, i.e. tweaking
the microcode.

[rick-l]

So you let everybody spew three pages of L di/dt calculations over a period of days all the time knowing that this was not the limiting factor for a DME ignition?

Originally Posted by mysocal911

Yes, you've forgotten that in your simulation, i.e. a very critical parameter!

Once saturation of the core (the point at which the flux no longer changes), the voltage across the coil goes to zero

[scarceller]

Dave,

I've done this already, with my EPROM emulator I can increase dwell time and watch voltage at DME pin 1, exactly as you say. The result is the voltage starts to go way up at DME pin 1 (knees upward) right around 3.8 to 4.0 ms this is exactly how I find optimal dwell time for any coil in car. I tweak the dwell map to do this.

Basically I've confirmed peak saturation time in car at about 4ms and it also matches the 964 bench test using the 964 ignitors at 4ms. In the end I usually don't tend to care all that much of the inductance value of the coil, I simply test in car by increasing dwell till the coil saturates.

I've also measured voltage drop across the coil's '-' and '+' posts, this also works. So long as I use a isolated ground scope test lead.

Originally Posted by mysocal911

Sal should be able to easily determine the coil's saturation point by incrementally increasing
the dwell time while monitoring pin 1 of the DME ECM uses an EPROM emulator, i.e. tweaking
the microcode.

[rick-l]

The voltage going up at DME pin 1 would indicate the start of current limit, see V (col) in the last simulation. Core saturation would be indicated by an accompanying increase in current rate of change.

[scarceller]

Rick,

I think Ingo was correct to assume you'll never have full system voltage across the coil because of the darlington and 0.05ohm resistors within the DME. I think this is why that actual in car dwell time is somewhat longer. When I get a chance I'll scope pin 1 again to see what exactly the voltage drop shows. But as I recall we have like 1v drop even at very low current from ground to pin 1 while the coil charges.

What I see in car at pin1 is that the voltage ramps nice and linear, then it takes a sharp bend upward as the coil saturates, then it finally plateaus as the current limiting kicks in. I also see this in bech testing the 964 coils with the ignitors. Look at my video again, you'll see the sharp knee followed by the plateau.
https://www.youtube.com/watch?v=a7dFdIdsZeU
See the 3min25sec mark in the video, you can see the knee happen followed by current limit.

Most hi-end EFI systems like MoTec and BoschMotorSports outline this very method for finding peak dwell times. They make a big point to say 'it must be done in car' so it accounts for all the voltage drops in the final coil configuration. Now that statement I read really make sense.

[mysocal911]

Originally Posted by rick-l

The voltage going up at DME pin 1 would indicate the start of current limit, see V (col) in the last simulation. Core saturation would be indicated by an accompanying increase in current rate of change.

Yes.

[rick-l]

If the lower scope trace is the equivalent to pin 1 on the DME note the similarity to the v(col) plot in post #66. The voltage gradually rises as the current through the transistor goes up until the current limit starts, it rapidly rises, and hits a plateau. Is that what the lower trace is, the collector of the drive transistor?

Any way to add a 0.1 Ohm resistor to measure current?

EDIT: if that is the collector voltage notice that the voltage (which indicates current) is pretty linear until it starts to current limit which would indicate the inductance is not changing much before it hits the knee. I would think the core would be made of a cheaper iron material and would soft saturate, not hard saturate.

[mysocal911]

Originally Posted by rick-l

I
Any way to add a 0.1 Ohm resistor to measure current?

Just open the DME ECU and monitor the current limit resistor voltage.

[ischmitz]

Just a note of caution: Measuring across the 0.05 Ohm shunt with an o-scope is not trivial. Most bench-top o-scopes are grounded to neutral and thus you can't simply hook the ground wire up to one side of the resistor. Your benchtop power supply is likely also grounded to neutral.

I use a Fuke 123 when I have a task like this at hand - it is battery-powered and thus I can connect the GND lead anywhere I want without having to pay attention. With a bench-top you either want to use a good current-probe ($$$) or do a 2-channel differential measurement (noise-prone).

Ingo

[scarceller]

That lower trace is the voltage at the coil's '-' (from ground to the '-'). In that bench test video I was using the 964 BIM ignitor so we are measuring the voltage across the BIM (from ground to the BIM's coil output that goes to coil's '-'). Once voltage starts to go up significantly across the BIM it means the coil is at or very near saturation.

I do the same type of test in car at DME pin 1 and see same type of behavior.

[scarceller]

Agree, I use a hi-quality Techtronix 2465 scope that has isolated ground on the input. Or I can do as you say 'diff between 2 inputs', either works on this scope.

Originally Posted by ischmitz

Just a note of caution: Measuring across the 0.05 Ohm shunt with an o-scope is not trivial. Most bench-top o-scopes are grounded to neutral and thus you can't simply hook the ground wire up to one side of the resistor. Your benchtop power supply is likely also grounded to neutral.

I use a Fuke 123 when I have a task like this at hand - it is battery-powered and thus I can connect the GND lead anywhere I want without having to pay attention. With a bench-top you either want to use a good current-probe ($$$) or do a 2-channel differential measurement (noise-prone).

Ingo

[scarceller]

Dave,

Can do this also, those 2 resistors are 0.05ohms so at:
1amp = 50mv
2amp = 100mv
4amp = 200mv
8amp = 400mv

Certainly within the ability of the o-scope to measure this.

I have a test DME, I can easily solder a wire to that location and scope that point.

Originally Posted by mysocal911

Just open the DME ECU and monitor the current limit resistor voltage.