Ignition timing question, CIS vs EFI

[tabasco]

When we look at the advance curve from the pre-Motronic era, the advance usually keeps going up, eventually reaches max advance and stays up there until redline (with some exceptions due to vacuum advance or retard).



Now, on the 964 & 993 maps that Steve W shares on his website, we can see that at full throttle, max advance is reached around 3500~4000, and then it goes down.

https://www.911chips.com/ignition.htm

See at the bottom of the page, the 964 and 993s ignition curves reach around 25 degrees around 3000~3500rpm, then the timing goes down to around 15 degrees and eventually back up to only 17 close to redline.

Can someone explain why the curves look like this? Why does the 3.6L need that much more timing at mid-range than high up?
Intuitively, it would seem that higher revs would require more advance.

[snbush67]

To continue advance at higher revs may result in detonation.

[tabasco]

Originally Posted by snbush67

To continue advance at higher revs may result in detonation.

OK. But can you explain why?

[phelix]

And don’t forget that 964 and 993 are duel plug so don’t need as much advance…

[mysocal911]

Originally Posted by phelix

And don’t forget that 964 and 993 are duel plug so don’t need as much advance…

Actually, the 964/993 have advance curves that run into the low 40 degrees ranges. They also have knock control systems using DSP
(digital signal processing) technology to monitor for detonation and retard the timing.

[Schulisco]

I'm no motor specialist. All I can say from what I've read over the time - ignition and fuel mapping (if possible, depends on the engine management) is always a compromise between hardware requirements, horsepower, torque, emissions, driveability, fuel economics, temperature management etc.

A leaner mixture tends to knock earlier than a rich mixture. Knocking tend to occur more on lower revs with much load on acceleration than on higher revs. So depending on the fill grade, required time for burning and exhausting the cylinder will force to retard the ignition.

The larger the displacement the longer the flame front takes time to burn the gas, that's why bigger and race engines have twin spark ignition.

Also there is one specialty on Porsche engines - the valves are bigger than other vendors to let them "breathe" better which gives more torque on lower rpm ranges. That's why the optimal position for the spark plug in the middle of the cylinder head was not possible. And 4 valves per cylinder were too costly and too complex for that type of cylinder heads they were using.
Remember - the 959 had water cooled cylinder heads, because of the high cylinder head temperatures. This was also a reason why Porsche decided to move to water cooled engines, beside gas and noise emissions, more power and fuel economy.

Thomas

[AndrewCologne]

A leaner mixture tends to knock earlier than a rich mixture.

No, leaner mixture needs a more advanced ignition to have the same power/pressure pressure at the same time compared with richer mixture.
Principles of Tuning Programmable EFI



Thats why in ignition maps mixtures at lower/leaner loads will be ignited faster, means with more ignition advance. And thats also why a high "ported vacuum" at the dizzy (which does occur on very low load) needs to be ignited faster where the vacuum unit of the dizzy advances with approx. 5°.

It does not burn faster but its combustion will result hotter due more O2, here often exists confusion

Knocking tend to occur more on lower revs with much load on acceleration than on higher revs.

knocking especially at higher revs is dangerous, as you dont hear it the way you directly recognise it at lower revs.

[tabasco]

Originally Posted by mysocal911

Actually, the 964/993 have advance curves that run into the low 40 degrees ranges.

They never reach 40 degrees at full throttle, only in mid/high revs low load (deceleration).
And that's expected, the usual trend for timing is going up (edit) as load decreases, and going up as RPM increases.

In this thread, Jamie mentions high timing in deceleration helps engine braking.

They also have knock control systems using DSP (digital signal processing) technology to monitor for detonation and retard the timing.

I wonder if the knock sensor is active throughout the whole rpm range. Some engines only use it at lower revs because otherwise the signal is too noisy.
If they used a DSP maybe they can efficiently filter out the noise and get it to work all the time.

Without knock control, I could understand the engineers would stay on the safe side, and only push the timing close to the limit when the knock sensor is active.
If the mixture gets richer at high revs, it would require less timing as it burns quicker.
Maybe a combination of both?

I don't think I've seen any other maps with such a reduction in timing from mid to high rpm at full load.

[AndrewCologne]

Originally Posted by tabasco

They never reach 40 degrees at full throttle, only in mid/high revs low load (deceleration).
And that's expected, the usual trend for timing is going down as load decreases, and going up as RPM increases.

In his post he does not mention that such an advanced timing is reached at WOT.

And in your last paragraph youre mixing up load and RPM, as ...
higher load = less advance
lower load = more advance
Higher rpm = more advance (up to a certain rpm)
Lower rpm = less advance

I wonder if the knock sensor is active throughout the whole rpm range. Some engines only use it at lower revs because otherwise the signal is too noisy.
If they used a DSP maybe they can efficiently filter out the noise and get it to work all the time.

A DSP with an optimal bandpass filter build to the specific engine chracteristic is obligatory – cause a knock detection at only low/mid revs would be no real help.

[Schulisco]

@AndrewCologne:
https://en.wikipedia.org/wiki/Engine_knocking#:~:text=enriching%20the%20air%E2%8 0%93fuel%20ratio%20which%20alters%20the%20chemical %20reactions%20during%20combustion,%20reduces%20th e%20combustion%20temperature%20and%20increases%20t he%20margin%20to%20detonation

Detonation can be prevented by any or all of the following techniques:

• retarding ignition timing
• the use of a fuel with high octane rating, which increases the combustion temperature of the fuel and reduces the proclivity to detonate
• enriching the air–fuel ratio which alters the chemical reactions during combustion, reduces the combustion temperature and increases the margin to detonation
• reducing peak cylinder pressure
• decreasing the manifold pressure by reducing the throttle opening or boost pressure
• reducing the load on the engine

I read the opposite: Leaner mixtures burn slowier than stochiometric or richer ones. I did not know that modern direct injection and superchargend engines are more subsceptible than older ones.

[tabasco]

Originally Posted by AndrewCologne

In his post he does not mention that such an advanced timing is reached at WOT.

The topic of this thread is advance at WOT.

And in your last paragraph youre mixing up load and RPM, as ...
higher load = less advance
lower load = more advance
Higher rpm = more advance (up to a certain rpm)
Lower rpm = less advance

Correct! That's what I meant, thanks for catching it, I'll edit the post.

A DSP with an optimal bandpass filter build to the specific engine chracteristic is obligatory – cause a knock detection at only low/mid revs would be no real help.

Yes, that's goal, but sometimes valve train and other engine noises at high RPM prevent from reaching this goal. I've read that some knock sensors only work until mid range, which is still helpful as knock can happen at full load in the lower RPM range. Like flooring it on the wrong gear. And that's where the most engines spend most of their time anyway,
If Porsche achieved to use the knock sensors on the full RPM range, that's great.

[spuggy]

Originally Posted by tabasco

The topic of this thread is advance at WOT.

I can't really comment on this, because my ignition map is MAP vs RPM. I do use a relatively high number (37 degrees or so, twin-plugged) for off-boost/light load. It snarls along nicely at cruise, and throttle response seems pretty good...

Originally Posted by tabasco

Yes, that's goal, but sometimes valve train and other engine noises at high RPM prevent from reaching this goal.

Functional aftermarket knock detection has been available a long time (90's?).

My experience with it is that you run an aircooled motor up with WOT/wastegate open to redline, an abrupt throttle close (eg for the shift) will cause knock detection to "hear" metal-to-metal from the chain/pistons etc for a fraction of a second, no more. Considering it vanishes on the re-application of throttle or when the RPMs drop a bit - and that timing gets pulled when the throttle is already shut, it doesn't seem like much of a problem to me.

Some seem to think an aircooled 911 motor is too "noisy" for knock detection, but this would seem to ignore that the factory didn't think so for the 964/993. Works great for me - I've been running it on my 930 for more than a decade and still running it now, with a full EFI (Motec) setup. In preference to Motec knock detection.

Originally Posted by tabasco

I've read that some knock sensors only work until mid range, which is still helpful as knock can happen at full load in the lower RPM range. Like flooring it on the wrong gear..

Safeguard works from 2000 RPM up to whatever.

I found that a "sort-of" stock 930 with factory timing will knock around 3000 RPM with load & West Coast premium gas - eg climbing hills on 1/3 throttle in the correct gear.

US 930 dizzy advance curve & timing is less aggressive/lower than RoW specs - still knocks & causes the Safeguard to pull timing when climbing hills.

I used to have little shiny pockmarks in the carbon on the piston crowns. Haven't had that for a long time now.

I some things to be not what I expected once I actually had instrumentation:

• My 930 didn't knock under boost. At all. With sane timing/boost/fuelling. Ever. Not once.
• My 930 knocked in the mid-range with some load (but out of boost), with conservative (factory) timing. Every time.

https://forums.pelicanparts.com/911-930-turbo-super-charging-forum/611409-my-j-s-safeguard-paid-itself-last-week.html

Oh, and actually trusting the Safeguard to detect knock & pull timing (after it demonstrated that it did precisely that when my distributor advance plate stuck fully-advanced!) allowed advancing the timing 5-6 degrees over stock (and bleeding the extra off as boost rises to arrive back at factory timing).

Could probably have gone much further - but that yielded a very useful improvement in mid-range tractability/response. Looks to me as though the factory ended up leaving a lot on the table, timing-wise, for safety. It's kind of amazing to me that CIS cars run as well as they do, TBH... Those guys at the factory really knew their stuff.

[AndrewCologne]

Originally Posted by Schulisco

I read the opposite: Leaner mixtures burn slowier than stochiometric or richer ones.

Yes, and that's why you advance the ignition on leaner mixtures and lower loads so the pressure at combustion gets its optimal timing.

Remember when removing the feet from the gas pedal at higher revs where the mixture at deaccelerating suddenly leans out and the spark isn't fast enough to fully ignite the mixture in the cylinder so it makes the popping exhaust sound. That's why.

Other fact: If connecting the retard vacuum hose on a 3.0 US 930.16 engine like used as factory default, you MUST check the deceleration valve for proper working, so it adds "metered" air to the manifold. If not and with such an retarded ignition timing in combination with a closed throttle at such high revs and such an upcoming high vacuum in the manifold, this will lean out the mixture significantly and a huge popping sound will come out of the tail pipe

Originally Posted by tabasco

Yes, that's goal, but sometimes valve train and other engine noises at high RPM prevent from reaching this goal. I've read that some knock sensors only work until mid range, which is still helpful as knock can happen at full load in the lower RPM range.

Perfect adjusted DSPs will exactly hit the exact frequency where knocking on that specific engine will occur. We're talking about 6 khz in case of a 95mm bore engine at all revs. I installed such a knocking sensor from Bosch in my 3.0 SC engine and it's interesting what you hear without bandpass filtering and ... with bandpass filter at approx. 6khz when a slight test knock occurs. This peak signal will be directly recognized by a DSP and the ECU retards the ignition timing on that point of load and rev.

[mysocal911]

Originally Posted by spuggy

I can't really comment on this, because my ignition map is MAP vs RPM. I do use a relatively high number (37 degrees or so, twin-plugged) for off-boost/light load. It snarls along nicely at cruise, and throttle response seems pretty good...



Functional aftermarket knock detection has been available a long time (90's?).

My experience with it is that you run an aircooled motor up with WOT/wastegate open to redline, an abrupt throttle close (eg for the shift) will cause knock detection to "hear" metal-to-metal from the chain/pistons etc for a fraction of a second, no more. Considering it vanishes on the re-application of throttle or when the RPMs drop a bit - and that timing gets pulled when the throttle is already shut, it doesn't seem like much of a problem to me.

Some seem to think an aircooled 911 motor is too "noisy" for knock detection, but this would seem to ignore that the factory didn't think so for the 964/993. Works great for me - I've been running it on my 930 for more than a decade and still running it now, with a full EFI (Motec) setup. In preference to Motec knock detection.



Safeguard works from 2000 RPM up to whatever.

I found that a "sort-of" stock 930 with factory timing will knock around 3000 RPM with load & West Coast premium gas - eg climbing hills on 1/3 throttle in the correct gear.

US 930 dizzy advance curve & timing is less aggressive/lower than RoW specs - still knocks & causes the Safeguard to pull timing when climbing hills.

I used to have little shiny pockmarks in the carbon on the piston crowns. Haven't had that for a long time now.

I some things to be not what I expected once I actually had instrumentation:

• My 930 didn't knock under boost. At all. With sane timing/boost/fuelling. Ever. Not once.
• My 930 knocked in the mid-range with some load (but out of boost), with conservative (factory) timing. Every time.

https://forums.pelicanparts.com/911-930-turbo-super-charging-forum/611409-my-j-s-safeguard-paid-itself-last-week.html

Oh, and actually trusting the Safeguard to detect knock & pull timing (after it demonstrated that it did precisely that when my distributor advance plate stuck fully-advanced!) allowed advancing the timing 5-6 degrees over stock (and bleeding the extra off as boost rises to arrive back at factory timing).

Could probably have gone much further - but that yielded a very useful improvement in mid-range tractability/response. Looks to me as though the factory ended up leaving a lot on the table, timing-wise, for safety. It's kind of amazing to me that CIS cars run as well as they do, TBH... Those guys at the factory really knew their stuff.

The key, though, is that the knock control firmware needs to be designed specifically for each engine configuration,
which typically is not the case for most all aftermarket engine builds, e.g. using MoTech.
A DSP chip such as the TMS320 can provide the necessary processor power once the time domain data
(knock sensor signals) are converted to the frequency domain for processing.

The 964 DME ECM used a separate chip (Motorola 6800 - 8 bit) for knock control, where the 993 DME ECM
had that function integrated into the main Intel processor.

[Jonny H]

Let's not forget that the mechanical distributor cannot bring the advance back down due to the way it works (weights 'thrown' out with centrifugal force).

Even on older engines, there is a an advantage to more advance in the mid-range. Using electronic advance, I've seen an RS motor take 45 degrees 'on cam' without knocking, falling to 36 at high RPM to avoid detonation.

[snbush67]

With vacuum signal hooked up, I have a 123 distributor set up to advance to 47 degrees at lower rpm’s with light load, as RPMs and load increase the MAP timing is backed out.

There are two curves one is determined by RPM’s and the other by MAP/ vacuum.

[Schulisco]

Originally Posted by snbush67

With vacuum signal hooked up, I have a 123 distributor set up to advance to 47 degrees at lower rpm’s with light load, as RPMs and load increase the timing is backed out.

There are two curves one is determined by RPM’s and the other by vacuum.

Sounds interesting. I suppose on your '78 911?

Would you share the curves / files with us? 47degrees adv sounds pretty heavy. Do you used a knocking tool? Which gas? Is the engine still stock?

Thomas

[Jonny H]

Originally Posted by snbush67

With vacuum signal hooked up, I have a 123 distributor set up to advance to 47 degrees at lower rpm’s with light load, as RPMs and load increase the timing is backed out.

There are two curves one is determined by RPM’s and the other by vacuum.

Originally Posted by Schulisco

Sounds interesting. I suppose on your '78 911?

Would you share the curves / files with us? 47degrees adv sounds pretty heavy. Do you used a knocking tool? Which gas? Is the engine still stock?

Thomas

That is quite normal on 'light load' - it's an attempt at better fuel economy but should not be confused with advance on full throttle - completely different set of circumstances!

My RS example above of 45 degrees was WOT on a dyne with full monitoring of knock so quite impressive but not for the tinkerers without monitoring equipment!

[PeteKz]

This is a great discussion about timing. Let me add that more timing is not necessarily a good thing. I gotta wonder about 47 degrees of advance.

I can't speak about the advance curves in the Motronic and later EFI systems, as I still run CIS. However, there is point where too much advance loses power and increases tendency to detonation. That typically has to be determined on a dyno at a range of RPMs and loads, and that's how the factory or performance tuners figure out the basic advance maps. Then the knock sensing backs off that timing if knock is detected. That's how engine computers compensate for lower octane fuel or high temperatures, or other deviations from ideal.

[AndrewCologne]

The "tricky" thing is, that many 123 users do simply use the known "ported" vacuum connection which leads to the dizzy. And here this gives a total wrong signal as ported vacuum is only available at a slightly open throttle. With this kind of vacuum you will never get the whole "common" MAP vacuum range needed for a proper timing at "all loads".

So ... to get a full MAP signal ...
For 911 SC till '79 without ECU this is the same vacuum line like connected to the WUR.
For 911 SC from '80 with ECU its the same vacuum line connected to the cruise control and deaccel. valve.