Flywheel ring gear

[johnb]

Can anyone tell me exactly how the ring gear is installed on the flywheel on 8v cars? In particular I'm interested in whether there is anything that guarantees the way it's timed, like bolts or screws, or is it just pressed on (like this:

If anyone knows for certain please let me know.

Thanks!

[Tom]

On the 944, the ring gear bolts on. I can dig one out if you need details...

[johnb]

On the 944, the ring gear bolts on. I can dig one out if you need details...

Thanks. So would it be safe to say that the ring gear teeth always have the same relationship to the trigger screw for the reference sensor?

The reason I'm asking is I'm about to drop a new article on the ignition signal timing. There's some code that makes a small (half-tooth) adjustment to the timing depending on the phase relationship between the speed and ref sensor signals. I had always assumed that this was because the ring gear could be pressed on in any orientation, so the ref sensor screw might line up with a tooth, or fall exactly between 2 teeth, or anything in between.

But then I discovered there's actually another source of phase variation, which is the way they're digitized. So I started to wonder if that could be the reason for the adjustment in the code. From what you're saying now I think maybe this is the real reason.

It's not a big deal, no need to go digging things out. But it would be nice to understand exactly why they did that. Of course even if all ring gears have exactly the same timing with regard to the trigger screw, the programmers might just have decided not to rely on that.

[Tom]

I can't remember if the gear can be installed more than one way, or if the bolts only line up in exactly one position. My guess is the teeth end up in the same place regardless. I'll try to take a peek tomorrow... assuming I can find my old parts.

[johnb]

After a little more digging, my tentative prediction is that it's possible to have the ring gear 2 different ways. That would make the code I'm looking at make the most sense.

The other possible explanation I mentioned was that it had something to do with phase change between the signals due to the way they're digitized, but I think I have disproved that now. What made me suspicious was that the reference sensor doesn't trigger at the zero-crossing point, but actually triggers earlier than that based on a voltage threshold that seems to adapt so that it's always around 40% of the peak voltage. But from the testing I did today, it turns out that the phase relationship between the 2 digitized signals stays the same at all engine speeds. So the adjustment in the code can't have anything to do with this.

Interestingly though, a result of this method of digitization is that the ref sensor digital pulse is always ~3.7 degrees (1.35 teeth) before the zero-crossing point. I had assumed that the digital pulse would be exactly at the zero crossing point.

All this gave me a weird case of deja-vu and after some searching I remembered this[1] in which @Tom and I heroically re-discover that the ref sensor screw is 58.64 degrees BTDC (21.5 flywheel teeth).

Now the code does all ignition timing measurements by subtracting the timing value from 22 teeth, which is exactly 60 degrees. But it has the half-tooth correction I mentioned earlier so presumably this accounts for the difference.

That leaves the mysterious 3.7 degrees to explain. I see there is a subtraction of 20 from the raw quarter-tooth values during timing calculations but that comes out to 5 whole teeth, or around 13.6 deg. There must be more to it than this so I'll have to keep digging.

I had hoped to be able to explain the real time part of this without digging into how the final timing values are calculated but alas, everything turns out to be connected together as usual.

[1] https://rennlist.com/forums/944-turbo-a ... -btdc.html

[Tom]

After a little more digging, my tentative prediction is that it's possible to have the ring gear 2 different ways. That would make the code I'm looking at make the most sense.

The other possible explanation I mentioned was that it had something to do with phase change between the signals due to the way they're digitized, but I think I have disproved that now. What made me suspicious was that the reference sensor doesn't trigger at the zero-crossing point, but actually triggers earlier than that based on a voltage threshold that seems to adapt so that it's always around 40% of the peak voltage. But from the testing I did today, it turns out that the phase relationship between the 2 digitized signals stays the same at all engine speeds. So the adjustment in the code can't have anything to do with this.

Interestingly though, a result of this method of digitization is that the ref sensor digital pulse is always ~3.7 degrees (1.35 teeth) before the zero-crossing point. I had assumed that the digital pulse would be exactly at the zero crossing point.

All this gave me a weird case of deja-vu and after some searching I remembered this[1] in which @Tom and I heroically re-discover that the ref sensor screw is 58.64 degrees BTDC (21.5 flywheel teeth).

Now the code does all ignition timing measurements by subtracting the timing value from 22 teeth, which is exactly 60 degrees. But it has the half-tooth correction I mentioned earlier so presumably this accounts for the difference.

That leaves the mysterious 3.7 degrees to explain. I see there is a subtraction of 20 from the raw quarter-tooth values during timing calculations but that comes out to 5 whole teeth, or around 13.6 deg. There must be more to it than this so I'll have to keep digging.

I had hoped to be able to explain the real time part of this without digging into how the final timing values are calculated but alas, everything turns out to be connected together as usual.

[1] https://rennlist.com/forums/944-turbo-a ... -btdc.html

That old thread reminds me I still have the speed and ref sensor test board, with flywheel and ring gear, so will drag that out later this afternoon and take a look. Interesting to see way back then we had speculated about the trigger point not being at the zero crossing point...

Do they both trigger at 40% of peak or just the ref sensor? Or is that what you mean about the phase relationship staying the same (i.e., both triggering at 40%)?

[johnb]

After a little more digging, my tentative prediction is that it's possible to have the ring gear 2 different ways. That would make the code I'm looking at make the most sense.

The other possible explanation I mentioned was that it had something to do with phase change between the signals due to the way they're digitized, but I think I have disproved that now. What made me suspicious was that the reference sensor doesn't trigger at the zero-crossing point, but actually triggers earlier than that based on a voltage threshold that seems to adapt so that it's always around 40% of the peak voltage. But from the testing I did today, it turns out that the phase relationship between the 2 digitized signals stays the same at all engine speeds. So the adjustment in the code can't have anything to do with this.

Interestingly though, a result of this method of digitization is that the ref sensor digital pulse is always ~3.7 degrees (1.35 teeth) before the zero-crossing point. I had assumed that the digital pulse would be exactly at the zero crossing point.

All this gave me a weird case of deja-vu and after some searching I remembered this[1] in which @Tom and I heroically re-discover that the ref sensor screw is 58.64 degrees BTDC (21.5 flywheel teeth).

Now the code does all ignition timing measurements by subtracting the timing value from 22 teeth, which is exactly 60 degrees. But it has the half-tooth correction I mentioned earlier so presumably this accounts for the difference.

That leaves the mysterious 3.7 degrees to explain. I see there is a subtraction of 20 from the raw quarter-tooth values during timing calculations but that comes out to 5 whole teeth, or around 13.6 deg. There must be more to it than this so I'll have to keep digging.

I had hoped to be able to explain the real time part of this without digging into how the final timing values are calculated but alas, everything turns out to be connected together as usual.

[1] https://rennlist.com/forums/944-turbo-a ... -btdc.html

That old thread reminds me I still have the speed and ref sensor test board, with flywheel and ring gear, so will drag that out later this afternoon and take a look. Interesting to see way back then we had speculated about the trigger point not being at the zero crossing point...

Do they both trigger at 40% of peak or just the ref sensor? Or is that what you mean about the phase relationship staying the same (i.e., both triggering at 40%)?

Good question - it's hard to tell! The speed sensor pulses are triggered at the 40% threshold on one side and exactly at the zero-crossing on the other side. But I can't tell which side is which because it's a differential signal and I'm not sure which side I should be looking at! With the ref sensor it's easy - if you probe one end and you see it going negative first, you're looking at the wrong end. But the speed sensor is there all the time so it's harder to tell if it went positive or negative first. In theory I might be able to catch it at the start with a slow time base, maybe I'll try that.

With that said my best guess currently is that for the speed sensor, the rising edge of the digital pulse is the zero crossing and the falling edge uses the 40% threshold.

In these images the speed sensor (red channel) is inverted in the software, which is how I think it should be.

Green is the ref sensor digital pulse at the 8051 INT0 pin. It's inverted after the S100 before the 8051. Falling edges trigger the 8051 interrupts for both signals. Note how the ref sensor pulse is right in the middle of the low part of the speed sensor pulse - I verified that it stays like this as engine speed increases. But the code handles the case where the speed sensor signal is high at this moment, by delaying ignition for an extra half-tooth. So presumably such a condition is possible on some cars.

Screenshot_2025-10-26_16-40-46.png (103.93 KiB) Viewed 354 times

Screenshot_2025-10-26_16-39-48.png (101.79 KiB) Viewed 354 times

[Tom]

I got my test rig out and checked the ring gear. It has three sets of three identical bolt holes -- meaning the ring lines up with the holes in any of three positions. Looking closely, however, I think all the teeth still end up in the exact same place regardless. The turbo has 132 teeth, though, which is divisible by 3, so it's possible the teeth get shifted some on non-turbo cars depending on how its clocked (or they do more to key it)? If I can do anything helpful with my test rig, let me know.

ring-gear-2.jpg (555.99 KiB) Viewed 346 times

ring-gear.gif (603.93 KiB) Viewed 346 times

[johnb]

Good point about the non-turbo cars. I think they have 130 teeth so maybe that's why the code is the way it is.

That's a cool setup - what were you using it to test?

Anyway since you have it out now it would be good to know how exactly the screw does line up. Is it lined up exactly with a tooth, or in between? From the signals I captured, it looks like the zero crossings line up pretty closely, close enough that the difference could be just variations in how the sensors sit or whatever.

Also looking at the teeth, it looks to me like the trough between them is bigger than the tips of the teeth, which tends to confirm my suspicion that my signal should be inverted (as it is in the pic I posted). This way, the positive half is narrower, which matches with the positive half being caused by the approaching tooth. In that case, I can confirm that the digital pulse of the speed sensor goes high exactly at the zero crossing on the rising edge, and the falling edge corresponds to ~40% of the negative peak.

So ultimately I think all this confirms that the extra 1.35 teeth (3.7 degrees) advance caused by the ref sensor threshold detection is not compensated for anywhere, so it must be in the actual timing. The -5 teeth thing is a bit of a mystery to me too. That would be around 13.6 degrees, so the net result is that the map values are ~3.65 teeth (10 degrees) advanced from the true values. Why? I can't think of any good reason for that.

[Tom]

When you ask how the screw lines up -- you mean the relationship between the ref sensor pin and the ring gear teeth? Other than accommodating other models, nothing obvious comes to mind about the half-tooth delay. The phase seems to be fixed by the hardware, so I can't think up an scenario where it would change on a running car...? I'll have to read your write up to understand the other quirks you mention...

On the test rig, I was just trying to get a better handle on the speed and ref signals, and used it to help make the Carmonica (i.e., a little test light for the signals). But in the process I stumbled across a little phantom pulse generated by the diagnostic pin, which could explain some of the starter kick-back issues...

ref pulse 1.webp (409.23 KiB) Viewed 340 times

ref pulse 2.webp (417.06 KiB) Viewed 340 times