Looking for DME BIN files

[Tom]

Here's what I don't get though. Why are the dwell times so long for low rpm? Worst case, at ~14v it takes no more than 6.5ms to fully charge. But the dwell map calls for more than 10 times that at the lowest rpm. What does the extra dwell time achieve?

Is there something happening at low rpm besides the voltage drop that makes it charge more slowly?

A couple of considerations: Firstly, during cranking the fuel-air mixture is generally not well mixed, and can be difficult to ignite. To get the best chance, we need to ensure the spark has the maximum energy possible.

It isn't that it needs that much time to charge. At 6.0 volts, the coils is basically as high of current as it can get (a little less than 6 amps) after 40ms, the longer time doesn't improve spark energy.

Background info:
The DME updates its rpm measurement every T1 interrupt (you calculated 11.5ms earlier). It is a count of how many teeth have passed since the last interrupt. At 40rpm, there is only 1 tooth counted for the 11.5ms duration. A small increase in actual crank speed will dramatically increase the DME's measured rpm. A count of 2 teeth is 80rpm; 3 teeth is 120rpm, etc.


Ok, now when first cranking the engine, the starter motor is accelerating the crankshaft. The DME might first calculate 40rpm, but due to the extreme quantization, that measurement can quickly change from 40rpm to something much higher. By having a long dwell time, the DME is ensuring that even if the rpms dramatically change, the coil will have had plenty of time to charge.

Additional factor, is that if you look at the crankshaft speed during cranking, it is not consistent. It slows when a cylinder is in the compression stroke, and accelerates after that cylinder hits TDC (this should be true even without spark or fuel).

This is guarding against the accelerations that happen during starting / low-rpm, to ensure the coil has plenty of time to build energy. We can see this as the DME has uses generally the same number of ('half')teeth required for the lower rpms:

DME_Dwell_raw.PNG

To charge and fire the coil confidently, we want multiple teeth to have passed. The table's minimum of 10 is equal to 5 teeth, which at 40rpm is approximately 57ms.

Finally, the 'on-time' of the coil vs 'off-time' at low rpms is pretty small. So, even though it is on for a long time, the off time is extremely long, giving a lot of time for the driver & coil to cool off. And cranking operation is not considered as a continuous operating mode, rather a transient.

That all make sense and aligns nicely with general internet discussion on the topic. It's worth noting that 40rpm is barely cranking and you can imagine if it's turning that low, even between chugs of the starter, the voltage is probably way down there too. Some of those cells seem to exists solely for transitions and idle fluctuations... Can't really see any sustainable, no-transitory, situation where the motor is cranking at 40rpm with 14 volts for example...

One nit, on the 86, 88S, and 89 factory images I have, unless I'm missing something, they all show a minimum of 7 (rather than 10) half teeth in the BIN.

dwell7min.jpg (245.13 KiB) Viewed 654 times

[Tom]

Well I tried measuring the charge-up time of the stock coil on my car today, using a current clamp probe on the scope. My alternator charges a little over 14 volts, though I did do a few tests with the headlights and blower fan on to bring the volts down in the mid 13's. It's hard to pin anything down exactly on a running motor because everything is moving at once -- voltage, rpm, and commanded dwell all bouncing around -- so consider this 'general direction' type results... This coil seems to take between 4.4 and 4.9 ms to hit peak 9 amps at typical operating voltages in the mid 13's to low 14's -- so not as fast as John's, and not as slow as Josh's. I'm pretty sure I have another factory coil in the garage somewhere, so will try some more tests this weekend. And if I can figure out how, I'll try bench testing too, to get the coil away from all the metal in the engine compartment. I'm beginning to think not all factory coils are alike....

14p2-720rpm.jpg (752.26 KiB) Viewed 633 times

Here, in an effort to hit cells with dwell times longer than charge-up requirements, the idle is slowed down to 720 (commanded) and volts are 14.2. The total dwell is btween 5 and 5.5ms, pretty much exactly what the dwell table says it should be. Just eye-balling it, I'd say it gets to 9 amps in about 4.6ms.




Here's a short video, also at a commanded 720rpm and you can see it's clearly hitting the max 9amps during the dwell time (bouncing between 5 and 5,5ms), again around 4.6ms to get to 9 amps... +/-

I did rev the motor and could see the dwell diminish to 3.6ms +/- at higher RPMs, as the tables suggest, so think it's all set up right -- but open to comments on my testing set up/accuracy. Maybe john's coil best reflects what Bosch expected? Mine's not 'too' far behind (and never misses) but I suddenly have coil envy. I'll do more testing this weekend....

[johnb]

Well I tried measuring the charge-up time of the stock coil on my car today, using a current clamp probe on the scope. My alternator charges a little over 14 volts, though I did do a few tests with the headlights and blower fan on to bring the volts down in the mid 13's. It's hard to pin anything down exactly on a running motor because everything is moving at once -- voltage, rpm, and commanded dwell all bouncing around -- so consider this 'general direction' type results... This coil seems to take between 4.4 and 4.9 ms to hit peak 9 amps at typical operating voltages in the mid 13's to low 14's -- so not as fast as John's, and not as slow as Josh's. I'm pretty sure I have another factory coil in the garage somewhere, so will try some more tests this weekend. And if I can figure out how, I'll try bench testing too, to get the coil away from all the metal in the engine compartment. I'm beginning to think not all factory coils are alike....



14p2-720rpm.jpg
Here, in an effort to hit cells with dwell times longer than charge-up requirements, the idle is slowed down to 720 (commanded) and volts are 14.2. The total dwell is btween 5 and 5.5ms, pretty much exactly what the dwell table says it should be. Just eye-balling it, I'd say it gets to 9 amps in about 4.6ms.




Here's a short video, also at a commanded 720rpm and you can see it's clearly hitting the max 9amps during the dwell time (bouncing between 5 and 5,5ms), again around 4.6ms to get to 9 amps... +/-

I did rev the motor and could see the dwell diminish to 3.6ms +/- at higher RPMs, as the tables suggest, so think it's all set up right -- but open to comments on my testing set up/accuracy. Maybe john's coil best reflects what Bosch expected? Mine's not 'too' far behind (and never misses) but I suddenly have coil envy. I'll do more testing this weekend....

Interesting!

I actually did another test a few days ago as a kind of sanity check - I wasn't going to post anything because I got basically the same results as before:

coil_10.png (97.01 KiB) Viewed 628 times

I tested it up to 2000rpm, it's consistently around 3.5ms for me.

This is probably the same coil I had back when I posted the comments on Rogue's rennlist thread. As it happens, I've been running into stumbling issues under load when the engine is warm so I have actually ordered a brand new Bosch coil anyway. Regardless of whether that ends up having anything to to do with my issue I'll do another test when it arrives.

[Tom]

I did more testing today and thought I'd share results. To eliminate variables (which I'll get back to), I opted to install a stock Bosch DME rather than the Ftech DME I had been using. I was getting odd results with the Ftech (see below) so wanted to take that out of the equation. Bottom line, with the stock Bosch DME and stock coil, my results are right in line with @johnb's. The results varied a slight amount between factory coils, but not by much.

Fig-1.jpg (654.75 KiB) Viewed 584 times

Above: Bosch DME; Bosch OEM coil, maybe 10-15 years old; 3.2ms @14.2 volts

fig-2.jpg (716.2 KiB) Viewed 584 times

Above: Bosch DME; Bosch OEM coil, maybe 10-15 years old; 3.4ms @14 volts
(headlights/blower on high to lower voltage)

fig-3.jpg (712.25 KiB) Viewed 584 times

Above: Bosch DME; original coil that came with the car in 1986; 3.8ms @ 14.2 volts

fig-4.jpg (724.9 KiB) Viewed 584 times

Above: Bosch DME; original coil that came with the car in 1986; 4.0ms @ 13.8 volts
(headlights/blower on high to lower voltage)



Now here's the odd part, I was getting odd and inconsistent readings with the FTech 9 DME. At times it would look like the stock DME and other times the amps would keep climbing the entire dwell cycle. After doing the above tests, where the probe seems to be working fine, I put the Ftech DME back in and saw the same odd behavior -- the amps are not leveling off at 9 (and the dwell seems a tad shorter despite same rpm and volts as other tests). I'm a bit stumped on this and hoping @FTECH9 sees this and might be able to shed some light?

Here's the same everything, except with the Ftech DME installed....

ftech-dme.jpg (597.28 KiB) Viewed 584 times

[Rogue_Ant]

Thanks for getting some more data points!

To get 9 amps that fast, the coils must have quite a bit lower inductances than what I measured. They need to be something in the range of 3.0 - 3.5 mH, depending on resistances.

I wonder exactly what the coil part number was that I tested? It came from my 1986 car, but I no longer have it or that coil.

[Tom]

Thanks for getting some more data points!

To get 9 amps that fast, the coils must have quite a bit lower inductances than what I measured. They need to be something in the range of 3.0 - 3.5 mH, depending on resistances.

I wonder exactly what the coil part number was that I tested? It came from my 1986 car, but I no longer have it or that coil.

It certainly wouldn't be the first time someone sold a near-miss part for the 944 to save a buck... (or just because they don't know any better). Still a highly informative discussion whether Bosch fully charges the stock coil or not. And if you had a coil that took 6.5ms to charge, chances are you're not the only one.

On my last point above, I sent a note to Joe, but it dawns on me that you @Rogue_Ant may know -- do the IGBT drivers in the Ftech DME limit current to 9amps like the factory DME? I did more testing, and the current just climbs the entire dwell cycle with my Ftech DME (but caps it right at 9 when using a Bosch DME). Is that by design, since the IGBT's can handle more current and the dwell maps keep it from ever going "too" high? Or should I be worried?

[Rogue_Ant]

It certainly wouldn't be the first time someone sold a near-miss part for the 944 to save a buck... (or just because they don't know any better). Still a highly informative discussion whether Bosch fully charges the stock coil or not. And if you had a coil that took 6.5ms to charge, chances are you're not the only one.

It sure looked like a factory coil, and I'm sure I tested more than 1. I wonder if there was a change during the production years...

[Rogue_Ant]

On my last point above, I sent a note to Joe, but it dawns on me that you @Rogue_Ant may know -- do the IGBT drivers in the Ftech DME limit current to 9amps like the factory DME? I did more testing, and the current just climbs the entire dwell cycle with my Ftech DME (but caps it right at 9 when using a Bosch DME). Is that by design, since the IGBT's can handle more current and the dwell maps keep it from ever going "too" high? Or should I be worried?

IGBTs are optimized to be an on/off device, not often intended for explicit operation in the 'ohmic' region (saturation region in a BJT, or linear region in a MOSFET). Generally current limiting with an IGBT is done via rapidly turning it on/off. (E.G. turn it off for 2us, then on for 40-50us, back off for 2us, etc.)
However, in addition to the current-limiting logic, it would likely require an active clamp during the current-limiting time to prevent errant sparking. This really makes the circuit a lot more complex. I don't know what the latest Ftech9 solution is, but when we were first develop the Rogue version, there was no current-limiting feature with the IGBTs.

More info - the IGBT should also have a lower voltage drop than the BJT Darlington used in the factory DME. So, everything else being equal, the IGBT solution should charge the coil slightly faster than factory. And with a little bit less loss (heat generation).

Lastly, given enough coil-on-time the inductance no longer impacts current. The coil will ultimately hit current dictated by V=IR (remember to include all resistances, and the VCE drop across the switch). This is likely around 20 amps for 13.8v.

[Tom]

IGBTs are optimized to be an on/off device, not often intended for explicit operation in the 'ohmic' region (saturation region in a BJT, or linear region in a MOSFET). Generally current limiting with an IGBT is done via rapidly turning it on/off. (E.G. turn it off for 2us, then on for 40-50us, back off for 2us, etc.)
However, in addition to the current-limiting logic, it would likely require an active clamp during the current-limiting time to prevent errant sparking. This really makes the circuit a lot more complex. I don't know what the latest Ftech9 solution is, but when we were first develop the Rogue version, there was no current-limiting feature with the IGBTs.

More info - the IGBT should also have a lower voltage drop than the BJT Darlington used in the factory DME. So, everything else being equal, the IGBT solution should charge the coil slightly faster than factory. And with a little bit less loss (heat generation).

Lastly, given enough coil-on-time the inductance no longer impacts current. The coil will ultimately hit current dictated by V=IR (remember to include all resistances, and the VCE drop across the switch). This is likely around 20 amps for 13.8v.

Thanks for all that! Can I interpret that to mean you wouldn't be surprised or concerned to see the amps keep climbing up past 9, 10, 11, etc. whenever there is enough dwell time to get there? I'm still just a bit stumped why the first time I tested it, it seemed to be limiting current to something around 9 or 10amps -- as seen in the video I posted in #232 above...

[Rogue_Ant]

Thanks for all that! Can I interpret that to mean you wouldn't be surprised or concerned to see the amps keep climbing up past 9, 10, 11, etc. whenever there is enough dwell time to get there? I'm still just a bit stumped why the first time I tested it, it seemed to be limiting current to something around 9 or 10amps -- as seen in the video I posted in #232 above...

Yes, I would expect the amps to climb throughout typical dwell times.

In your video, the VPP was exactly 10.0V. Likely, it was just the limit of the probe/scope input with however you had it configured.