procainestart
procainestart Dork
3/19/22 4:13 p.m.

Rainy Saturday, pondering random E36 M3. Like why isn't in-cylinder spark-plug ion sensing a thing?

My Saab 9-5 has in-cylinder spark-plug ion sensing to detect knock/detonation. If'n you weren't aware, between combustion events, current is passed to the plug electrodes. When there's knock/detonation, it creates a unique ion signature in the combustion chamber. Based on the current that crosses the plug gap, the system can very accurately detect knock; the ECU dials back timing before the next combustion event, on a per-cylinder basis (and if the timing doesn't quell it, then boost is bled). Said to be vastly better than a piezoelectric sensor.

GM was licensing the technology. I think some Harleys at some point, and apparently BMW or M-B, too?? But I don't hear about it now.

Just wondering why it never caught on, and what modern forced-induction cars use. The only issue I've heard about is that if you let the plug gap get too wide, it kills the coil-on-plug system, so maybe that's why??

Caperix
Caperix New Reader
3/19/22 5:25 p.m.

My best guess would be cost.  The BMW s85 v10 used ionic current knock detection & the 2 modules were a common failure item that would cause misfires

engiekev
engiekev HalfDork
3/19/22 6:25 p.m.

I would agree with cost. Knock sensors are cheap and proven. On an engine, cost is measured in fractions of a cent due to high volume. Warranty cost is also a huge factor.

Cylinder pressure measurement during development is used to define MBT and knock limited regions and calibrate the spark controls.  So if you can use that data with a well designed knock sensor, there isn't really a need for more sophisticated sensing unit. The benefits probably don't outweigh the costs for a ion sensing device.

There are a few engines that have cylinder pressure sensors in production,I think most are diesel and only have a sensor for one cylinder only (likely for cost).

alfadriver
alfadriver MegaDork
3/19/22 6:47 p.m.

It also wasn't as reliable as one would think.  Not that it would break, but it would miss faults.  It was used on the original Aston V12 in '99, but was taken off later when the crank motion detection was found to be good, and/or better in some cases.  And knock detection is much harder than just sound detection.

Good chance it will be coming back.  

procainestart
procainestart Dork
3/19/22 8:25 p.m.
alfadriver said:

Good chance it will be coming back.  

Despite your prior statement about unreliability, you're saying ion-sensing is getting a second look, yes? Also, I don't know anything about crank motion detection, but I assume it's some sort of algorithm that detects signature angular velocity changes? If the motion detection was at least as good, I wonder why OEs would revisit ion-sensing? Again, I don't know anything about crank motion motion, but if my assumption is correct -- that it's just 1's and 0's moving through a chip, based on a signal from a CPS -- then, with ion-sensing, you have the extra cost of a module (which in a 4-cyl. Saab, was integrated into the coil-on-plug system, which famously failed, primarily because, I'm told, widening plug gap would tax it).

alfadriver
alfadriver MegaDork
3/19/22 8:45 p.m.

In reply to procainestart :

As far as I can tell, it is being looked at, and I don't know why.  Crank angle based misfire is really reliable..

It could be that OEMs are trying to get super extreme with combustion phasing with hybrids to just make clean heat for the catalysts- and using the motor to keep the engine speed constant means you can't detect misfires.  The push to really clean engines is still going on, in spite of the electric development.  And some super extreme clean engines mean that cars like the Mustang has some leeway- and by that, it's not a lot- the worst of the next sets of cars will be better than the average car today.

 

Pete. (l33t FS)
Pete. (l33t FS) MegaDork
3/19/22 10:09 p.m.

Knock sensors can pick up other engine noises, though, like LT4 rocker arm noise or Subaru piston slap.

You ain't driven a slow car until you've driven a 2.2 Subaru that had the ignition timing adaptive pull ALL the timing out.  The funny thing is you could tell where the internal mapping went from one adaptive cell to the next because the engine would wake up like a flipped switch.  This is where it was common for people to relocate the knock sensor to a handy threaded hole on the bellhousing.

Misfire detection is also kind of hazy. For many years, VW would identify any cylinder as #1 misfiring.  And a bad miss might get interpreted as not only the actual cylinder with a problem, but also the next one in the firing order.  Have seen this a lot in Honda V6s, and most recently a 2.5l Nissan with a damaged injector wire on #4 that was also recording misfires for #2.  Probably not an issue in a practical sense, unless the computer shuts down too many injectors as part of a catalyst-saving scheme.

rslifkin
rslifkin UberDork
3/20/22 10:01 a.m.
alfadriver said:

Crank angle based misfire is really reliable..

Unless you're 1990s Chrysler...  With many of the JTEC PCM based vehicles, if it tells you there's a miss, you probably have one, but so minor you can't notice it (or if you can, it'll be at high RPM and just feel a little flat).  And there's a good chance it's not on the cylinder the PCM claims it is.  But if you've got a severe miss (or even a dead cylinder) where it's really obvious to the driver, the PCM will often sit there telling you everything is fine and not pick up on the misfire for some strange reason. 

alfadriver
alfadriver MegaDork
3/20/22 10:37 a.m.

In reply to rslifkin :

It's come a long way in the last 26 years.  26 years before I started working, there were no electronic fuel injected cars on the road.  Heck, computers were the size of rooms that far back.

Even 26 years prior to OBDII there weren't any EFI cars for sale that I am aware of.

Progress is amazing.

Pete. (l33t FS)
Pete. (l33t FS) MegaDork
3/20/22 11:01 a.m.

In reply to alfadriver :

Now you have me wondering when the first D jet cars came out.

Chrysler did have an analog EFI system in 1958. It is really interesting on a technological archeology perspective, but definitely not a mainstream application.

engiekev
engiekev HalfDork
3/20/22 11:04 a.m.

The need for continuous improvement in the auto space (or any product development space) guarantees we'll see more tech like ion sensing coming back and new tech as well. ICE development is far from dead, despite the push for EVs in the marketing materials.

Lots of really cool stuff going on in the combustion and ignition area, like dynamic skip fire from Tula, mahle prejet ignition chambers. Anything to help thermal or mechanical efficiency!

alfadriver
alfadriver MegaDork
3/20/22 11:53 a.m.

In reply to engiekev :

TULA is incredibly over rated.  BTDT.

Pete. (l33t FS)
Pete. (l33t FS) MegaDork
3/20/22 12:31 p.m.

What is TULA?

alfadriver
alfadriver MegaDork
3/20/22 12:47 p.m.
Pete. (l33t FS) said:

What is TULA?

It's a derivative of displacement on demand, where the system dynamically skips cylinders in the theory to add some more dimensions into the number of cylinders being shut off.  So instead of 2 or 4 cyls off, it can be 1-6 or 7 off with finer steps.

Pete. (l33t FS)
Pete. (l33t FS) MegaDork
3/20/22 1:03 p.m.

In reply to alfadriver :

I would think, as a layman with an opinion, that it'd be easier to have a small engine with a turbo than that kind of mess.  Start from a base of efficiency that provides power on demand instead of dragging a powerful engine around and hamstring a percentage of it for the 99% of the time you do not need power.

engiekev
engiekev HalfDork
3/20/22 1:09 p.m.
alfadriver said:

In reply to engiekev :

TULA is incredibly over rated.  BTDT.

It's a zero hardware cost solution, right? Can't expect too much from that. It's funny to see them pivot now to "skip fire" for EV motor control. 

Our Preferred Partners
5wj1dG9iVcM3o7Xh7TOdSkUcFnnJKDp6XPxQ4fZpyEpNnCpXJCSWr8D2NVPNhh4X