Page 3 of 7

Re: Ion Sense

Posted: Mon Nov 12, 2018 10:36 pm
by puff
I've read once again the explanaion above and don't see any magic in those 'magic' pins of that magic box anymore.
Pin 3 of the magic box sets high, when it sees knock. (btw, try start moving on idle with a handbrake - won't it start knocking?)
Pins 4&11 are needed for synchronization (You just fire a spark in two cylinders simultaneously - wasted spark, and the magic box will track ion sense and will set high eiher pin4 or pin 11, depending on in which of the two cylinders the spark was not acually wasted)

trying to figure out the part numbers - they are quite pricey I should say. but they seem to allow much better fault diagnostics. Wondering, what's inside that magic box, is it difficult to reverse engineer it? will these work on any sparkplug, on any engine wihout any additional fine-tuning?

Re: Ion Sense

Posted: Tue Nov 13, 2018 11:27 am
by kb1gtt
Ion sense is fairly well known. We expect the ion sense voltage is around 80V to 300V. The magic is mostly a reference to our lack of understanding about what this black box is doing. Does it generate a signal based on detected detonation? Does it generate a signal proportional to the cylinder temperature? If it generates a signal, how many volts does it generate? As well what are the signal delays or signal manipulations? There are many analog things which we do not know about this black box. So until it's better understood, it's magic.

Re: Ion Sense

Posted: Thu Nov 22, 2018 4:09 pm
by AndreyB
russian wrote:
Sat Nov 03, 2018 3:37 pm
looks like Mazda uses very similar looking coils? http://www.mcx5.org/ion_sensor-1281.html
So that PE2018100A coil is used on a lot of different recent Mazdas - 3 6 CX-5 2013-2017 probably more. Part of skyactiv maybe?
These ones seem to handle everything inside the main ECU, not with an external block like Saab. http://nema.club/2014wsm/service%20highlights/books/n6w01/html/id0140g0004400.html

Re: Ion Sense

Posted: Sat Nov 24, 2018 2:20 am
by AndreyB
All the components inside the Ionization Detection Module 12787708 box are surprisingly small - nothing big for high voltage?

Smaller chips seem to be saying S8TJ 22A0 and S84J 21H0 - google knows nothing.

The bigger chip seem to say H56968JFP or N56968JFP or M56968JFP. M56968JFP seems to be related to RENESAS in google.

(image removed - couple of posts below there is a better picture)

Re: Ion Sense

Posted: Sat Nov 24, 2018 9:42 am
by kb1gtt
Looks like silicon goop. That has a much higher dielectric strength than an air gap. As well that dielectric strength is much more stable than air over a wide range of temperature and humidity. This means you can get high voltage with very small gaps.

I don't see via's, and this looks to be a single sided board. AKA they used 0R jumpers instead of punching via's through the PCB. I suspect those green devices are 0R jumpers. Also those black traces are likely "printed" carbon traces which are resistors. Those are low quality resistors with wide tolerances. In this case the goop gives them much better humidity and temperature stability. Those traces are like a 20% or 30% tolerance resistor. The SMT resistors are likely 5% or 1% resistors.

Mineral spirits can likely help with removing the goop. You may want to try soaking it for a couple hours in mineral spirits.
While these may be standard chips with standard datasheet's, it is common that OEM's will have the chips customer stamped to prevent people from being able to do what you are doing. It's worth checking the chip numbers and looking for datasheets, but I doubt you'll find anything. Also I do not know if the mineral spirits will dissolve the carbon traces. I have had good luck with it on PCB's and it doesn't seem to effect caps, resistors, etc.

Can you trace the MCU pins and see what kind of schematic you get for those circuits? Can you trace any other circuits? I don't know when things are resistors, or what values they might have. You could take measure while tracing the circuit and drawing a schematic which includes ohms values.

Re: Ion Sense

Posted: Sat Nov 24, 2018 3:52 pm
by AndreyB
On that picture most of the goop is already removed - I've used flux remover liquid.

I doubt I will have time (and skills) to reverse engineer the schematics, would you be interested in me shipping you think block?

Re: Ion Sense

Posted: Sat Nov 24, 2018 5:05 pm
by kb1gtt
Sure ship it to me. I can probably poke it and make a schematic in the 10 minute windows after the kids go to sleep.

Re: Ion Sense

Posted: Sat Nov 24, 2018 11:08 pm
by AndreyB
Better picture with less goop and pinout commited to GIT https://github.com/rusefi/rusefi/tree/master/hardware/ion_sense_saab

This one is "first generation" 12787708 from 2003-2004

Image

Re: Ion Sense

Posted: Sun Nov 25, 2018 1:07 am
by kb1gtt
That's a much better picture. I see "output" has a low quality pull down resistor and some kind of ESD protection at the harness connector. So that's no big deal. It's basically directly connected to the magic chip. This likely means this signal is either a digital or analog signal, but probably it's analog. I think we already knew that.

I see the CD24 and CD13 are also just as vague. Little bit of electrical protection, and direct connected to a magic chip. Appears it could be either analog or digital, but I think we expect it's digital. It has a series resistor, which might be of interest once we know if this is analog or digital.

So basically from this picture, I don't think we know anything we didn't already know.

I suspect the best way to learn about this is to get a functional car and scope those signals. Right now we simply don't know the specific details about what those magic chips are doing. Perhaps get a SAAB ECU, connect it to a STM and simulate the engine signals. Then probe the signals.

Re: Ion Sense

Posted: Sun Nov 25, 2018 2:32 am
by AndreyB
kb1gtt wrote:
Sun Nov 25, 2018 1:07 am
Perhaps get a SAAB ECU
There is a lot of Saabs for sale cheap saying "run until key broke, would run great after easy key fix" ($1500-$2000 to replace key + immobilized + ECU). It looks like there is a very little chance of playing with all this on a bench, only if I buy (or rent?) a running Saab 9-3.

Re: Ion Sense

Posted: Sun Nov 25, 2018 3:09 am
by AndreyB
Surprise surprise, there are two version of this module. The one I've opened first was 12787708 listed as being for 2003-2004.

On the videos we actually see my car using 55352173, that one is listed as being for the 9-3 from 2005 to 2011. Here are pictures of the insides of 55352173 (I've renamed files in GIT so if you've downloaded previously your file name is wrong, please download again)

Image

And the connectors are keyed differently.

Re: Ion Sense

Posted: Mon Nov 26, 2018 7:22 pm
by Thommm
High potential feature for any engine management system. Have you guys read this:
https://www.newtis.info/tisv2/a/en/e60-m5-lim/wiring-functional-info/power-train/engine-electronics/ignition/ionic-current-control-unit-bank-2/1qP23f6

Bmw uses Ion sense in at least the M5 E60. Im still looking for a COP with most of the required stuff on the inside, I think these exist.

Re: Ion Sense

Posted: Mon Nov 26, 2018 10:45 pm
by kb1gtt
Hmmm, where to get. That picture seems to indicate some really nice things. AKA bar to mA is really helpful.

Re: Ion Sense

Posted: Mon Nov 26, 2018 11:10 pm
by AndreyB
Interesting how both Saab and BMW have placed these units really close to coils, right on the head. Is there a reason for this?

Re: Ion Sense

Posted: Tue Nov 27, 2018 12:20 am
by kb1gtt
I believe it's mostly because the wires are antenna's, they emit RF with is bad from a regulatory stand point. There might also be some stuff about longer wires could create low pass filters, and similar issues, but I think the major reason is the concerns about emitting RF.

Re: Ion Sense

Posted: Sun Dec 02, 2018 11:43 pm
by AndreyB
Now with four Saab coils but we were unable to get Miata engine to detonate while doing wide open throttle pulls :(


One of the potential issues with this experiment - I am using 2001 Miata spark plugs which in original application are used with coils-on-plug so they are probably resistor spark plugs, but I also use 1991 miata wires just to connect things and the wires are also resistor, so twice resistor overall. I should try with 1991 spark plugs to remove one of the resistances next time.

Next step would be to add some diesel into the mix.

Also open question why I was unable to start the car with this setup - I had to use original Miata coils to get engine warm and only at that point I was able to start on Saab coils. Why?

Two coils were operated by 5v ECU signal two coils were operated by 12v ECU signal. Need to confirm that spark is equal in both cases.

red signal is coil #3 control
yellow signal is knock detection pin #3
blue signal is pin #11 - Cyl #2 or Cyl #4 combustion detected, with 1.5K pull-up to 5v



logs and tune attached

Re: Ion Sense

Posted: Mon Dec 03, 2018 12:50 am
by kb1gtt
What is the ohms of a detonating cyl? AKA if it's to hot, then the standby voltage leaks some current. What is that voltage and current? I think I recall 900V and 1mA in one application from some years ago. Once you know that you could perhaps wedge an 0603 1M ohm resistor across the gap and see if that generates the analog signal. AKA resistor simulates the ionized electrical path.

Re: Ion Sense

Posted: Mon Dec 03, 2018 12:53 am
by kb1gtt
About starting when hot vs cold. My guess is that the spark is weak. Your coil drivers do not like the capacitance of your plug wires, this likely causes less energy in the spark. Once warm, it's simply easier for the plugs to ignite the fuel, so a weaker spark is sufficient.

Re: Ion Sense

Posted: Mon Dec 03, 2018 1:42 am
by AndreyB
kb1gtt wrote:
Mon Dec 03, 2018 12:53 am
About starting when hot vs cold. My guess is that the spark is weak. Your coil drivers do not like the capacitance of your plug wires, this likely causes less energy in the spark.
Any ideas how to solve this? Are you saying changing spark plugs would not improve things?

Re: Ion Sense

Posted: Mon Dec 03, 2018 2:38 am
by AndreyB
russian wrote:
Mon Dec 03, 2018 1:42 am
Any ideas how to solve this?
How funny would be to use SPARK PLUG EXTENSION TUBE PEUGEOT 404 or https://www.ebay.com/itm/282620583248
or FERRARI TESTAROSSA SPARK PLUG EXTENDER 1X 74MM BLACK 124792 https://www.ebay.com/itm/282620583248

Damn more $ into this guessing game :( And I have already tried to use two https://www.ebay.com/itm/152600474700 Coil-on-Plug extension leads (HT308) but these were pretty loose both on the spark plugs and in coils.

Re: Ion Sense

Posted: Mon Dec 03, 2018 10:28 am
by kb1gtt
Those solid extensions would likely be better than the flexible wire extensions. However right now I would wager a guess that your bigger issue is the long-ish ground path for the spark. If you could some how add a GND strap from your harness to the engine block, that would likely help allot. Keep in mind that when the spark plug fires, the energy that jumps the gap has to go some where. I believe it currently goes to the block, then across some GND strap to the frame, then through some GND strap back to the ION sense box, then across the harness wiring back to the other side of the coil. You want to make that current loop as short as possible, and with as fat of a wire as reasonably possible. So bolt the ION magic boxes directly to the engine block, and use a short fat copper wire. After that loop is minimized, you could consider shortening it more if you use those solid extensions.

Hmmm, what is the Miata compression ratio, also what is the Saab's compression ratio? The plug gap is likely important. Plug gaps can range from around 0.020 to 0.060, with most being around 0.030 to 0.035. You likely want your Miata plug to be close to the Saab gap. You also likely have a different compression ratio, so you may need a slightly different gap. Your coil wants to dump the energy at a specific voltage. Your voltage is determined mostly by the compressed air between the gap, and the distance of the gap. I'll assume the Saab is a turbo, so it likely has a lower compression ratio then the Miata while cranking. Keep in mind the turbo does not generate significant boost while cranking. Lets say the Saab wants to jump at 35kV. With a lower compression ratio, the Saab can have a larger gap than you have in the Miata to get that voltage. So if your compression ratio is higher, and if the Saab gap was say 0.035, then try a gap of perhaps 0.030 for your Miata. For a slightly less eyeball approach, look at the graph on the below page. You an determine Saab's air pressure based on the compression ratio and ambient air pressure. For your situation just just seal level pressure. Multiply that pressure by the Saab's gap. Then use that number to find the voltage on this graph. You can get your Miata compression pressure by measuring it with a gauge in the spark plug hole. Use the Saab's predicted voltage divided by your pressure to determine the gap.
http://www.dragonfly75.com/motorbike/gapvoltage.html

Hmmm, the above seems blah, as that appears it was for a specific test fixture. I'm not sure if that test figure is simulating your typically compression pressures or not. It might be for much lower compression. I expect your voltage to be around 30kV, which is reasonably far to the right on that graph. Any how, it might work for this application.

Also consider this. The coil has a peak voltage it can sustain. Probably something like 40kV. Your plug gap may be set for something like 30kV. However your long leads add impedance to the current loop that your spark current have to follow. Your coil has to provide voltage for the gap plus the impedance. This means that your 30kV gap may require 40kV or more from the coil. If this happens your coil may start to arc internally, or you may have some voltage limiting circuit activated which keeps the voltage limited and may limit the energy in the spark. Capacitance is a huge source of impedance, as is small long-ish wire. If you can remove the capacitance by shortening the wire, or if you can decrease the ohms by using fatter wire, those will decrease the coil voltage, such that a 30kV gap will need more like a coil voltage of 31kV to jump the gap. If you have exceeded say 40kV, you have likely caused arcing in the coil, and the coil may be damaged. So after you shorten the impedance, and after you have adjusted the gap, if it still does not fire as desired you may need to replace your coils.

Re: Ion Sense

Posted: Fri Dec 07, 2018 8:15 am
by nighthunter
Hi friends, now i am reading this thread. I have a Z20NET in Opel Vectra C it is the same engine rebranded from Saab B207 which uses the same ionsense technology. Should i measure you the signals? I can measure up to 4 analogs on my DS1054.

Re: Ion Sense

Posted: Sat Dec 08, 2018 3:22 am
by AndreyB
nighthunter wrote:
Fri Dec 07, 2018 8:15 am
I have ... which uses the same ionsense technology
YES!!! YES!!! Please-please-please!

Which year is your Vectra? There were two versions of the box.

How are we going to get to the copper in the wires? Can you buy a piece of harness and replacement box to scavenge the connector so that you can sniff signals? Or maybe just a replacement box to open it and attach to posts inside the box? Are you comfortable cutting or stripping your wires on your harness? You probably do not want to open your real box.

To begin with I would love to see how your pin #3 and pin #11 signals look like.

Re: Ion Sense

Posted: Sat Dec 08, 2018 10:31 am
by nighthunter
I have theese needle probes. Image

But i can even cut the wires if current probing is needed. Anything for science :-). The sense lines are shielded, so they cant all be needle probed, so they need to be cut near conectors. It is an Z20NET 2003 M/T GTS Irmscher Vec C. But now it is far from specs. TD04L14 turbo installed. IDS+ suspension, B207R CAMS, balancers taken off, ceramic clutch. 3" exh and downpipe. Stock CAT. Only fine tuning of the T8 is needed (learning in progress).
Write down wich is needed. Today i will take a look wich type of ionsense module i have installed. I think there were actually 3 (3 different PNs) versions of the module.

Re: Ion Sense

Posted: Sat Dec 08, 2018 1:17 pm
by AndreyB
That's a sweet ride!

Probably same wire colors? Let's start by just sniffing idling or revving GND #13, one of coil control wires like #1, #3 knock signal Yellow/Blue is the main signal and one of combustion detected like #11.

https://rusefi.com/wiki/index.php?title=Vehicle:Saab_9-3_2005#Knock_detection

Also with IDM disconnected can you figure out pull-up value in combustion detected like #11 wire?

Re: Ion Sense

Posted: Sat Dec 08, 2018 5:38 pm
by nighthunter
Ok, tomorrow youll have all the signals scoped here. But today i got a spare tank with a spare pump. I was on to test an 15% diesel blend. But can it be that the iosense also senses peak pressure? Tomorrow we will see. Because almost instantly when the blend hit the running engine MIL came on with ignitor/spark plug bad DTC.
Today i was too busy to test it. Tomorrow Ill take scope to my workshop and record it. For today here are some pictures taken from the module i have installed. Looks like the module has also a pin legend stamped on. Cute.

Re: Ion Sense

Posted: Sat Dec 08, 2018 6:23 pm
by AndreyB
So yours is 12787708, I couple of these here to try reproduce your results.

Re: Ion Sense

Posted: Sun Dec 09, 2018 2:31 pm
by AndreyB
"A comparison of three ion sensing circuits" attached - found this while trying to google what other OEM applications used ion sense. So far we know Saab, Mazda SkyActiv 2011+ and BMW S85 v10 engine which was recently featured as "Best Car You Should Never Own" https://www.youtube.com/watch?v=bCp_4uVtT-Y

Re: Ion Sense

Posted: Mon Dec 10, 2018 7:59 pm
by nighthunter
Hi,
first batch of data is here
CH1 - E1
CH2 - E4
CH3 - I1
CH4 - I4
Accquired with Rigol DS1054Z.

the pass is rusefi4russian2018waves

On the I signals, there were spikes of around neg. 70V refferenced to vehicle GND.
Also all signals are referenced to vehicle chassis/GND. GND was taken from the Ionsense/coilpack common ground point on right top of the head.

Re: Ion Sense

Posted: Mon Dec 10, 2018 10:30 pm
by AndreyB
I do not have a Rigol so I am not sure which exact software would open these wave forms. Anyone? Is that going to be something from http://int.rigol.com/Support/SoftDownload/2? Which one?