Two stroke load measurement idea

[Nath88]

Hi,
I planning to use microsquirt with ms2extra, on a single cylinder 2 stroke. I will be at the bench testing phase soon, waiting on a few fuel system parts. I will be starting with alpha-n strategy due to the weak intake vacuum signal. I've been told that I may run into low load throttle response issues when running alpha-n, due to the variations between cycles of the 2 stroke engine, the airflow predicted by the throttle position is not reliable like a 4 stroke engine (where the airflow is consistently induced by the falling piston).
Standard theory is that the airflow through the 2 stroke engine is induced by the pressure in the crankcase vs. the pressure (vacuum) in the exhaust port, since both ports are open simultaneously. So measuring this with a differential pressure sensor should accurately predict the airflow...
So my question is, would it be possible to use this pressure differential as the 'MAP' input to Speed-density?
The 'MAP' pressure would be sampled at the peak point, presumably around 125 degrees ATDC (after the transfer ports open). Pressure differential could be up to 100kPa.

[billr]

I expect that may give a usable signal, you just gotta experiment a bit. How about just measuring MAP in the crankcase, using the MAP sampling-window settings to get the best MAP-type signal there?

[dontz125]

Where you put the sensor will make a HUGE difference, especially with a reed-valve intake. I would be fascinated to see some actual datalogs to demonstrate this notion of a 'weak MAP signal'; it strikes me as counter-intuitive, as the engine has to draw in the same amount (or more!) of charge as an equivalent 4T cylinder, in the same (or less!) crank angle.

Here's a thought - I've seen comments from turbo guys trying to measure their wastegate back pressure; they use 12-18" lengths of stainless brake line (tube, not braided) followed by a hose connection to a MAP sensor. It suddenly occurs to me that measuring the header pressure - as close to the exhaust port as possible - timed for exhaust port closure would give you a MUCH more accurate measure of trapped pressure, much like the MAP sampling on a 4 stroke is normally set to measure the intake tract pressure just as the intake valve closes.

[Nath88]

I think crankcase pressure would be better than in the intake for sure. I was going to go that way at first, and I probably still will, before I try to incorporate the exhaust pressure effect.

DonTZ, clever. The YZ has some extra exhaust ports above the main, wouldn't be difficult to drill in and pick up the pressure from there. The only problem i see with measuring the trapped charge is that you'd need cylinder injection to get the fuel into the measured air... It would be perfect for a direct injection set up.

[Nath88]

I've read in a few places that the two stroke intake signal is similar to high overlap cams. Theres a vid on youtube of a banshee idling at 85kPa. I have another log somewhere of an EFI RZ350 around the same. 80kPa on overrun.

[dontz125]

You don't spray all the fuel into the measured air in a 4-stroke; why would you try to do it with a 2-stroke? With the standard MAP measurement, you're measuring / calculating the mass of fuel & air ALREADY in the cyl, and using that as a basis for predicting what will go in next cycle. Same idea with the XAP (eXhaust Absolute Pressure ) - you're measuring / calculating the heated & compressed mix that's been trapped in the chamber, and using that to decide how much will probably arrive over the next 360 degrees. Assuming any semblance of predicability and regularity, I see no reason why you couldn't tune an SD map using XAP.

Banshee / RZ MAP - where is the MAP sensor? When is it being measured? I expect the vacuum upstream of a reed valve can get pretty fluffy, plus those two machines have a balance tube. The crankcase of a rotary valve engine gets pulled down HARD just before the intake opens.

[Nath88]

My thought was to cater for cycle to cycle variations, assuming the previous cycle doesn't predict the next very well... Would all depend how much it varies. At the times where the motor is 4 stroking it could be a lot. Either way it would require getting the fuel in less than half a cycle so probably a bit advanced for anything but high pressure direct injection. XAP sounds the best and easiest so far. My plan will be to get up and running on alpha-N with throttle body injection (throttle body from a CRF250) as proof of concept. Later I'm going to run an injector in one of the transfers. Good to know there shouldn't be an issue running an unconventional pressure signal.

The Suter 500 run at the IOM this year was having trouble with response on corner exit, closed throttle back to open throttle at high rpm, apparently the rider was opening the throttle on corner entry to get a response by the exit. In steady state conditions the fuelling was OK. Guessing it was the whole resonant exhaust thing throwing off the prediction.

Those other set ups were port injection and map sensor upstream of the reeds, don't know about sample time, but I agree after the reeds would give a nice pressure or vacuum signal depending on the sample timing.

[dontz125]

The YZ has some extra exhaust ports above the main, wouldn't be difficult to drill in and pick up the pressure from there.

I was actually thinking about drilling into the header of the expansion chamber. Getting the inner lip of the tube as flush with the inner wall of the header should give the cleanest static pressure,with the least interference from pressure waves or gas flow. There are strap-on thermocouple mounts that require no welding; these would be the easiest install. The actual sensor might need to be flexibly mounted with a proper connector to allow easy chamber removal, especially for a racer.

My thought was to cater for cycle to cycle variations, assuming the previous cycle doesn't predict the next very well... Would all depend how much it varies. At the times where the motor is 4 stroking it could be a lot.

There really shouldn't be that much cycle to cycle variation; even chopping the throttle should still result in a smooth and predictable (if steep) gradient.
Four-stroking means it's WAYYYY too rich, and all the pulse timing in the world won't help.

The Suter 500 run at the IOM this year was having trouble with response on corner exit, closed throttle back to open throttle at high rpm, apparently the rider was opening the throttle on corner entry to get a response by the exit. In steady state conditions the fuelling was OK. Guessing it was the whole resonant exhaust thing throwing off the prediction.

Back when I was racing TZs, I had to crack the throttle while under braking to make sure the off to on transition was smooth. Sounds like they haven't finished dialling in the bottom.

[Nath88]

I'll probably just drill into the expansion chamber as you suggest. Ground flush and probably at the top so it doesn't fill with oil. For the crankcase pressure I was thinking a barb fitting into a tapped hole in the side of the reed block housing area would do nicely.

I've experienced the same thing, opening the throttle just to get it firing going before opening up to 100%, so it's not completely unique to EFI.
With piston port/diaphragm carb kart motors, for a given RPM there is a certain amount of throttle required before the motor responds at all. Like the throttle dead zone increased with RPM. At full rpm it would need probably 85% throttle to keep it firing. I suspect the airflow drops dramatically under this throttle position, fuelling needs to reduce accordingly rather than reduce linearly like a 4 stroke. The banshee guy on here had a similar problem and cured it with Decel fuel cut, because it was overfuelling on the throttle chop between gears and took a second to clear when cracking throttle back open. Probably just need to lean out the decel area of the VE table like you suggest.

I'd like to see a graph of actual airflow across all RPM and TPS sites to confirm. Considering rigging up a MAF to tune the fuelling on decel from actual airflow.

Going to dyno the bike this week in carbed form as a baseline before I get started.

[dontz125]

If you have a MAF rigged up, you no longer need to worry about SD or AN tuning ...

[Nath88]

While that would make life easy, I don't think I can fit one in the bike, it would have to be a temporary setup.

[dontz125]

0 280 218 113 - 96mm x 70mm OD MAF, 10-480 kg/hr.

[Nath88]

Thanks Don, another question you might be able to help with... I'm now thinking of using a missing tooth wheel and hall effect sensor to get more accurate fuel delivery.
So now I'm thinking of controlling ignition also. I'd like to trigger the stock CDI through the pickup coil input, and simulate the input from the microsquirt ignition output with some external circuitry if necessary. Ever done anything like that? What sort of signal does the CDI look for? The CDI trigger is a standard type 'long tooth' on the ignition rotor with VR sensor.

Reasons for keeping the stock CDI: self powered, self contained, already has a good base curve, already has a place on the bike and I can revert to normal operation by switching the stock pickup back in.

[dontz125]

It's expecting a VR-type signal, well-spread. Whether it will accept a 5v DC pulse, or if you'll need to add some conditioning components to trick the input circuits, is something you'll need to experiment with.

[Nath88]

I did some research which indicates the CDI triggers on the zero crossing, after a negative or positive peak, normal VR stuff. I want to try using the MS ignition output into a transformer, wind the output equivalent to the original pickup coil, then set the ignition dwell in the MS so it's equivalent to the length of the tooth on the rotor (around 0.3ms at 9000rpm), then wind the input as required to get a nice output on a scope. Easy, right?
Timing should be more accurate than the stock ignition setup, since it will be getting the sharp zero crossing it only gets at high rpm at all rpm.

[dontz125]

Probably be easier to come up with an RC or RLC filter to mimic the filter. Measuring the coil resistance is easy, measuring the inductance less so without the right meter (or without published values in the service manual).

Download something like LTSPICE for free, then play with different values to get the wave form you want. Note that it will very likely not be necessary to get a perfect duplicate of the VR wave to operate the stock box; close enough may actually be quite coarse and simple.

Edit - I would be astonished if the stock unit triggers on zero crossing; the whole problem the MS family has with long tooth triggers is that it is impossible to predict when a zero crossing will register. I find it far more likely that that this is either an edge or voltage-threshold trigger.

[Nath88]

So I've looked at the sport devices CDI diagram and a few others similar. Would you say this setup is typical of motorcycle CDIs?
If so, I could block the leading pulse from the stock pickup with a diode, and produce that from the MS, modifying the stock curve. The trailing pulse can stay for ignition until the MS syncs and redundancy.

Or I can produce the trailing pulse and fire the ignition any time.