DIYEFI.org Forum

Hi all

I've had a bit of a dialogue with Wiglaf on the topic over PM, and now we decided to make a public thread out of it. This is for anyones benefit as well as an opportunity for anyone to contribute to the discussion.
Initially, i've asked Wiglaf if he would be willing to help with design/prototyping of the high voltage part of the system since he mentioned he has lots knowledge in that area.

here is a part of Wiglaf's last PM:

Wiglaf wrote: Is the end goal for this to be integrated into the computer, or would it be standalone that just sends out an AFR and spark advance signal? Just curious, as you know my project will demand a different computer than what most of the guys here are working on so if it was somewhat modular then that would help out.

How difficult the circuitry is kind of depends on how accurate it needs to be. Do you know what the actual current range we are looking at is? Also, 400V seems like almost enough to cause it to arc if it's in there with compressed air and gas. Does it only run the measurement during the power and exhaust stroke? Less than .04" spacing across surface will arc at 400V, air is better but obviously not when it's loaded with gas and high temps. This is a bit of a gray area for me.

To answer the first paragraph, IMO that the amount of processing power needed to do the number crunching for the pressure peak estimation will necessitate another MCU or even a specialized DSP (perhaps FPGA based). So, i'd say that the digital circuitry will need to be separate from the rest of EMS, but have a communication channel to the EMS to constantly update it with the estimated pressure peak information for later decisions by EMS (for example on how to adjust timing advance on the next ignition events). So i would say it is a good idea for it to be modular in this sense (i.e not monolithic with the rest of the EMS)

There are some caveats with that though: the ion sensing module (ISM, just to call it something) will need to know about the angular position/speed of the crank (to be able to reference the ion current data against the crank angle) and the time window when to sample the ion current. So ISM will probably need to get this info from the EMS over some fast interconnect.

In addition, i realize that Wiglaf is building an engine with non-standard valvetrain (to put it lightly ), and his EMS needs may be different from rest of us, but i believe that he can still leverage a lot of stuff from the freeEMS while making an EMS for his project Really, from what my info is about Wiglaf's project, everything is gonna be the same except for the lack of throttle body (i.e TPS) and consequent lack of reference-able vacuum (i am guessing he's gonna use a MAF to estimate load and airflow) and extra control outputs for controlling the valvetrain actuators.

In regards to accuracy of the high-voltage part, i vaguely remember the current being in milliamp range, but please review the references (next post) to verify.

I will pull up my references about IS and post them up here for everyones benefit. Please add more if you have good ones.
There were a few discussions about this on msefi forums, but they've disappeared last time i've checked. (Lance's pruning?)

Cheers

Here are some references:

Lars Eriksson's phd paper

Very cool (but old) article about someone's successful DIY effort

A high-voltage circuit update for the system in the previous link

Delphi's SAE paper on IS and PPE

VEMS IS references (many links are outdated)

An insightful discussion on NASIOC with some good links

will add more as i go.........

good links, i'm pretty glad that they went as far as to include some circuits... although I'll admit that I still would have a hard time making heads or tails of the actual data. I need some time to read all of this..

I would like to bump this thread to see if Wiglaf or anyone else have some fresh ideas, in particular about the high-voltage hardware part of it.

here are some ideas i've came up with my limited knowledge of coils/transformers/high-voltage stuff:

- i have a coil-on-plug (COP) setup on my car, which perhaps makes it a bit harder to feed the necessary 400VDC directly to the plugs. So, i've been thinking of just feeding a high-frequency low-voltage (high-current?) VAC into the primary windings of the existing COPs during the 'pressure peak measuring window' of the piston cycle. Parameters (frequency/voltage) will have to be calculated to ensure about 400VAC on the plug, with the frequency that would be appropriate to sample the ion current with. Having the phase info for this VAC, a backend computer can be analyzing the signal 'in-sync' with the VAC phase (sampling at peaks and/or trophs, when 400V potential is achieved on the plug electrodes)

i need people experienced with this to criticize the above idea. Will this melt the coils? Any other coils i should use to make this setup easier?

Thanks
Alex

That seems like a pretty good idea to me :-)

Provided you never let the coils saturate you shouldn't heat them much. You will need quite a high frequency AC to do that though or you will end up making many small sparks ;-)

Fred.

Fred wrote:... or you will end up making many small sparks

Fred.

which is fine IMO as the flame front would have long left the spark spot at that time window, and only burned gases would remain at that spot.

I have a feeling that small sparks is exactly what's going on during ion sensing the way manufacturers do it. You basically induce a constant (or periodic, according to my idea) 'spark' (i.e electrical current through the ionized medium of gases that dynamically change their properties, such as temperature, pressure, chemical composition ...), and measure the current with a current mirror and other circuits, hopefully converting it to nice high-impedance 0-5v easily measurable signals.

Hmm, I don't think so as I think they leave the low HV applied all the time except when the spark is being fired. Furthermore, a spark takes a significant potential to form across a gap like that. Some current leakage is not a spark, a spark needs quite a bit of push. You would have to be selective about when you did and didn't pwm it if you were making sparks/constant burn. I strongly doubt that that is occurring. I could be wrong though :-)

Fred.

That's odd... You can do that? You would think that since the coils are only a few ohms (maybe 50?) that you'd get quite a lot of current there, unless you're counting on the inductance to keep it from building up. But that doesn't sound super realistic, since... Since inductors should flow current like caps flow voltage.

Silly thought.

Precisely. Inductors resist change in current. Change from none to some, change from some to none, change from fwd to reverse. You can rely on them to restrict current flow, but you would need to reverse polarity fairly quickly. Measuring it might be tricky though.

Has there been any progression on this?

I'd like to have a go at building the ion sensing circuit outlined in http://vvnet.fi/ville/ion/DIY-Ion-Sensing-2.pdf. However, I'm not entirely proficient when it comes to electronics (I'm a programmer), esp. the HV supply is a little out of my league.

While it would be nice to use the ion signal as real time feedback to the ECU, I think the easiest solution for the moment is as ababkin outlines to analyse the signal with a laptop, in an autotune fashion. You need a laptop in order to tune the car anyway. Having the ion signal would in any case allow us to start working on algorithms--if we manage to optimise the algorithms enough, they may eventually be able to run on the ECU itself. Possibly a little optimistic, but I figure it's worth a try.