Possible uses of our timer resources : ideas soon please!

[AbeFM]

Ah, ok - that's what I was getting at, then, about the "line" at 6 cyl being wasteful. I'd rather see a very dynamic bit banging system which would be more generic, and would work as well on a single cyl 6 stroke as it would a 10 cyl staged injection set up.

My original point behind staged injection was for smooth idle, but I brought in water injection or different fuels as things you can't argue away through injector sizing arguments. Also, no one wants to have this awesome computer then have to buy different injectors any time they want to make a change or play with ratios. I have a friend going through heck right now with a MS with bags and bags full of injectors and VE tables which are total lies trying to get his secondary injectors to come in correctly. It works, but it's a needless hack due to a lack of support from the ECU of choice.

[jbelanger]

Then if we argue about water injection, do we then put it as a 3rd injection channel per cylinder? It might seem overkill at first but for a high boost application that you want as streetable as possible, 2 fuel injectors and water injection would be the ultimate setup.

With bit banged output, this is just another pair of timers so we still use only as many as before with OC. Of course the number of interrupts goes up but I wouldn't rule that out. Also do we now set the max number of injectors to 24 so that you have again the maximum of 3 injection channels per cylinder for an 8-cylinder engine?

[AbeFM]

One thing I was thinking, earlier, is to bank certain injectors, things as water injection, or even secondaries, shouldn't matter so much. You could bank fire secondary injectors on an inline 4 (what does 'sequential' mean anyway at 80% open duty cycles, when the valve is only open 1/4 of that time?), water injection could be on big injector. You'd still want to be able to control the ratio, but the when is not nearly as important as how much.

Still, all in all, I'm a fan of bit banging, driven by timers. To me it seems fast and pretty flexable.

I really doubt you'd need staged injection AND water injection for anything practical. The only reason to run huge injectors you can't idle with is because you're keeping the charge over fueled and cool, which water will do for you. 500 cc/ cyl will work on most any engine and will idle fine. And with water on top of that you're probably not limited by injection.

Then again, what's another byte?

[Fred]

Man, if only you two posted a little less quickly. I'm starting to get RSI as a consequence of a bad seating position and responding to endless posts :-|

Not that I don't want discussion occurring, on the contrary, but I have to keep my finger on the pulse as I'm supposed to be running the show...

Thoughts :

12, 16, 24, fuck it, 90 injector channels, why stop at 24....

Seriously, look at the configuration nightmare that ms2e is.

Core to my idea of the whole point of this exercise is simplicity. Trying to do way too much on a limited bit of hardware is asking for trouble IMO. The core is fastER, it is NOT fast. It is a chassis chip, and a slight hack to use for this in the first place because of budget reasons.

Here is the reality : The more cylinders you have, the more power you can make on smaller injectors and the less cylinder pressure you need to do so. I think it's way out of the scope of this project to want to do 8 primary 8 secondary, 8 water etc.

Don't forget fast integrated logging is another key point. As are "variable sets" i.e. you see a set of things that were used together, not a partially updated set... We need cycles for that stuff too.

Don't forget the board requirements too. Sure you can layer another dozen boards on and do XYZ, but the BASE code should support the BASE hardware config WELL and not do an inferior job of the basics to try to support the sort of dude that obviously has enough $$ for a second FreeEMS for his other bank of injectors on his V8,10,12 or to run half the engine on one EMS etc...

The majority of engines that NEED better injector control are LOW CC and boosted.

At some point you need to stop and say, enough is enough.

Remember the goal at this stage is knock out a GOOD foundation that does accurate injection in a basic way for 1 - 6 in TRUE hardware control and the same for 8 10 12 with semi sequential. A branch can exist with the timer bit banged 8 - 10 - 12 sequential idea in it sure. That is a good thing. BUT, you will idle brilliantly on one injector per port with any of those combinations even with a very large injector (not too much of a % of V8 engines are 500cc per cyl are they!!)

The key is to use our resources to knock out a good basic controller that does an excellent job of a relatively narrow range of configs. See the FreeEMS summed up thread. I'm not saying never do more with it. I'm saying limit to a sensible amount to start with or we will never get to the stage of running something because we'll still be arguing about the pros and cons of this method and that while MS3 rolls out and starts selling...

24 FETs/IGBTs on the main O board is a lot already. (12+12/12+6+6) and will keep 99% of people VERY happy. I mean, look at how happy most MS users are now. I'm happy with it. I'm not happy about the politics or future direction. The intent of this is a step up from ms2/v3/ms2e at the same sort of price with more performance and options to actually use the features because of pin count.

We don't want to try to conquer the F1 scene with it...

Majority of ms2e users (that post at least) : 4 cyl
Next : v8 (only by sheer numbers of em in the US of A)
next : 6 pot

who do you personally know with a jag v12? ok, rule them out, what about ANY other v12 or v10? do they want to put a solder it yourself EMS in their flash car? probably not. If we cater for the cylinder counts in a proportional manner :

12 cyl done ok (where ok = semi seq or bit banged as bit banged may be good when its timer done, but its not brilliant)
10 cyl done ok
8 cyl done ok
6 cyl really well
4 cyl really well
lower cyl counts ok

we please everyone pretty much and the largest group a lot.

I would like to hear Sean's thoughts in this thread. He is our only "V8 guy" at the moment, does he need 24 outputs for fuel+ign+water? i strongly doubt it. would he be ok with using two cores to do that sort of level of control on such a MONSTER build? probably. Seriously, though we all aim for More Power!! :-) when is enough enough in terms of support? basic timer semi sequential or bit bang seq could get you 800hp from most V8's with boost and or revs without too much hassle electronics wise. 800hp makes even a 3 tonne tank move damn fast!!

My ute is good for about 13.0 as it stands on low boost with a "lousy 2 litre" and it weights between 1200 - 1500kg (dont know accurately) its making 400crank hp roughly with 1 squirt simultaneous and 550cc injectors maxed. it idles fine. I'm thinking that just upping the boost will get me a low 12, and I DON'T CARE about 1/4 mile times AT ALL, just circuit use.

No matter what we do within reason, it's going to kick arse for the price for any reasonable amount of power from any semi normal engine setup. I think its severely shooting ourselves in the foot to try to do too much.

I don't have time or wrist life left to discuss timers in detail, but this was supposed to be about OTHER uses for times, triggers and inj and ign were a given...

This isn't really the place to discuss how many we want to aim for, but because you guys have (prompted by me) it can all stay i guess...

I could have sworn I nailed down 12+12/12+6+6 weeks ago and we were supposed to be figuring out the best way to achieve it. That is a <extremely large amount> of injector/ignition control as it is. more is just gettin silly IMO.

Sean, post your thoughts! :-)

Admin.

[jbelanger]

I agree that there are hardware limitations (both CPU and board) which will can be the main driver for what will be done. As long as the firmware is not to limitation for doing more then I don't have a problem. If/when xgate is used some of those limitations will become less restrictive and it would be good to have only some adjustments to do on the code to make use of this.

To come back to the original topic, the point is that there is a need to have 4 timers for ignition for certain engines and 4 timers for bit-banging sequential and staged injection. And if the 12+12 can be reconfigured to 8+8+8 then it would support all realistic applications without going overboard. And I agree that using the 6 OC for sequential or semi-sequential injection where possible is better than bit-banging outputs.

As for the other timer uses, as I mentioned I think it would be useful to have one or 2 IC for speed measurement. But some of it could be inferred or done with outside logic. Other than that I can't think of anything else and you mentioned the PWM uses earlier.

Just to close the loop on the requirements vs design issue, I think that the kind of reverse logic of defining the requirements to fit the hardware are more justified here due the practical reasons such as very low volume and resources and the fact that the hardware used is not grossly limited for the task. This refers to this post viewtopic.php?f=13&t=52

Jean

[Fred]

As long as the firmware is not to limitation for doing more then I don't have a problem. If/when xgate is used some of those limitations will become less restrictive and it would be good to have only some adjustments to do on the code to make use of this.

I think it is unreasonable to make a major architectural change like starting to use the XGATE and expect to have to do very little. If the code is modular (as much as it can be) and clean etc, then reworking all of those parts won't be too hard at all. It's only when the code is tangled with itself that you run into issues, i.e. no clear interface between logical blocks.

To come back to the original topic, the point is that there is a need to have 4 timers for ignition for certain engines and 4 timers for bit-banging sequential and staged injection. And if the 12+12 can be reconfigured to 8+8+8 then it would support all realistic applications without going overboard. And I agree that using the 6 OC for sequential or semi-sequential injection where possible is better than bit-banging outputs.

I don't recognise that need in either case. For the staged injection you can and for anything except your A series engines you should end at the same time. start each staged injection with the PIT timer, finish both main and staged with the main timer. rotary ignition... i may have this wrong, but even if the events are close, why cant the timers be reused?

PITISR1(void){
clear flag
bang preconfigured output
configure next output
configure timer
exit stage left like the pink panther
}

As for the other timer uses, as I mentioned I think it would be useful to have one or 2 IC for speed measurement. But some of it could be inferred or done with outside logic.

I can't say I recognise that either, a normal ISR and a counter checked regularly with the modular down counter or the RTI will give you all speed inputs in a reasonable fashion updated plenty often. It requires no where near the accuracy that engine position does. Not even close. Just a counter/time each. Done.

I have to ask you Jean, why don't you do the ecotec swap or similar? I understand the love of old cars, and adore my beetles, but if it wasn't for the impractical nature of running coolant to the front, I would be the first to be swapping a well designed modern engine into one. I've done extensive work on minis, and they are GREAT fun little cars, but engineering wise they are like a b&g board :-p long stroke, small bore, micro brakes, glass jaw gearboxs etc. What is the attraction to keeping the A series? If you are pushing boost etc then it's obviously not a vintage classic. Just curious to understand :-)

Admin.

[jbelanger]

As long as the firmware is not to limitation for doing more then I don't have a problem. If/when xgate is used some of those limitations will become less restrictive and it would be good to have only some adjustments to do on the code to make use of this.

I think it is unreasonable to make a major architectural change like starting to use the XGATE and expect to have to do very little. If the code is modular (as much as it can be) and clean etc, then reworking all of those parts won't be too hard at all. It's only when the code is tangled with itself that you run into issues, i.e. no clear interface between logical blocks.

I agree that the changes will be significant but as you say only to the parts affected. What I meant is that the part doing the computations should be done such that changing the number of supported cylinders should be a minor thing or even involve no changes at all because it's all parameterized.

To come back to the original topic, the point is that there is a need to have 4 timers for ignition for certain engines and 4 timers for bit-banging sequential and staged injection. And if the 12+12 can be reconfigured to 8+8+8 then it would support all realistic applications without going overboard. And I agree that using the 6 OC for sequential or semi-sequential injection where possible is better than bit-banging outputs.

I don't recognise that need in either case. For the staged injection you can and for anything except your A series engines you should end at the same time. start each staged injection with the PIT timer, finish both main and staged with the main timer. rotary ignition... i may have this wrong, but even if the events are close, why cant the timers be reused?

PITISR1(void){
clear flag
bang preconfigured output
configure next output
configure timer
exit stage left like the pink panther
}

I can see that timers could be reused for ignition because there probably wouldn't be any overlap. However, there could be inversion of order (theoretically, I have no idea if it is actually possible) which means special cases and the 2 events could happen on different sides of a wheel event. These are all special cases to be handled either in the ISR or the computation if you use different timer, everything is simplified because all those special cases disappear with separate timers.

As for injection, ending at the same time is not good for the A-series engine, you're right, but that's not the only case. The injection end is not an angle event it is a time event. True you want to end at a certain angle but if the speed of the engine changes during the time you compute the start angle and the time you end then you don't finish at the same angle. That's true for both injector but if they have different pulsewidths, you don't start at the same time so the error is not the same. It is possible that this error is negligible. But what may not be is in the case where the 2 injectors are at significantly different distances from the intake valve. You then want to be able to change the timing at which they end to account for the different travel distances. Also, in addition to the special cases mentioned for ignition, in the case of injection I think that widely different pulsewidths are possible for primary and secondary which means there could be overlap.

As for the other timer uses, as I mentioned I think it would be useful to have one or 2 IC for speed measurement. But some of it could be inferred or done with outside logic.

I can't say I recognise that either, a normal ISR and a counter checked regularly with the modular down counter or the RTI will give you all speed inputs in a reasonable fashion updated plenty often. It requires no where near the accuracy that engine position does. Not even close. Just a counter/time each. Done.

I would need to do the computation but I'm not convinced that you would not get to much jitter from that to not have quite a bit of noise if you wanted to extract HP measurements from the data. But if you say so, then it's fine.

I have to ask you Jean, why don't you do the ecotec swap or similar? I understand the love of old cars, and adore my beetles, but if it wasn't for the impractical nature of running coolant to the front, I would be the first to be swapping a well designed modern engine into one. I've done extensive work on minis, and they are GREAT fun little cars, but engineering wise they are like a b&g board :-p long stroke, small bore, micro brakes, glass jaw gearboxs etc. What is the attraction to keeping the A series? If you are pushing boost etc then it's obviously not a vintage classic. Just curious to understand

Admin.

First, I don't even have one anymore (due to personal reasons that required me to get rid of most of the "toys"). That may change in the future and I may consider an engine swap but the fact that it is there and port injection hasn't been done successfully by anyone but Rover is reason enough to do it. And the A-series engine just fits. Even though it was already years if not decades old when the first Mini came out in 1959.

Jean

[SleepyKeys]

As for injection, if we want to extend support to up to 16 injectors channels and staged injectors (which will cover 8 cylinders sequential plus staged injectors or up to 16 single injectors sequential) then there would be a need to use 4 timers: one for the start-of-pulse for the main injectors, one for the end-of-pulse for the main injectors, one for the start-of-pulse for the staged injectors, one for the end-of-pulse for the staged injectors.

I think seq on at least 8 would be a big plus. You could batch fire a second set of injectors and as long as you didn’t max out your seq controlled set, you can still have individual cyl fuel trim control.

Does anyone think there would be a need for 12 cylinders staged?

Not likely, 10 cyl maybe (ford/dodge v10s are somewhat common in some countries, but most people that need big power chose V8s).

Actually, if you need the speed then you want the timer. And you may actually want to have 2 timers for speed measurement if you want to do traction control: one on the driveshaft or a driven wheel and one on an undriven wheel. You can use rpm only for traction control but direct speed measurement can give other information such as gear, clutch slip, wheel spin, ... And those can be used for modulating boost control and some other power control as well as help in power measurement using the logged data.

An other point about knowing actual speed is for idle control. If you're moving slowly, you don't necessarily want to do the
same thing with the IAC as if you're just idling.

Traction control would be a nice feature especially for real race cars. Having an idle rpm/MPH table would be neat, that's one reason I let the stock ecu control my iac valve. With vehicle speed we could do transmission controls too, a typical auto trans controller sells for several hundred US dollars.

Core to my idea of the whole point of this exercise is simplicity. Trying to do way too much on a limited bit of hardware is
asking for trouble IMO. The core is fastER, it is NOT fast. It is a chassis chip, and a slight hack to use for this in the first place because of budget reasons.

Im a big fan of doing more with less, but how much more would it cost for a fully qualified MCU and when you factor in how
much you spend on the average project does it really cost that much more?

Remember the goal at this stage is knock out a GOOD foundation that does accurate injection in a basic way for 1 - 6 in
TRUE hardware control and the same for 8 10 12 with semi sequential. A branch can exist with the timer bit banged 8 - 10 - 12 sequential idea in it sure. That is a good thing. BUT, you will idle brilliantly on one injector per port with any of those
combinations even with a very large injector (not too much of a % of V8 engines are 500cc per cyl are they!!)

It would seem to me that the seq/semi-sqe model can be extended (much of the code resued) in the future on hardware that is more capable. When that times comes it seems like all other methods may become "effort better spent elsewhere" just a thought.

I would like to hear Sean's thoughts in this thread. He is our only "V8 guy" at the moment, does he need 24 outputs for
fuel+ign+water? i strongly doubt it. would he be ok with using two cores to do that sort of level of control on such a MONSTER
build? probably. Seriously, though we all aim for More Power!! when is enough enough in terms of support? basic timer semi
sequential or bit bang seq could get you 800hp from most V8's with boost and or revs without too much hassle electronics
wise. 800hp makes even a 3 tonne tank move damn fast!!

Thanks for the intro. I would be very happy with 18 outputs(8ign, 8 seq,1 batch fuel, 1 batch meth), but I dont have a problem with semi seq model at this time. I'm definitely not the most experienced hardware/software guy on this board but , I would rather see this project grow with the hardware rather than trying to do too much with too little.

-sean

[AbeFM]

Ah, good to see another set of comments in here.

I have one question I want to preface this all with: Is there anything inheriently wrong with doing *everything* as a bit bang? Then the number of channels becomes purely a software issue - 4 cyl, 10 staged sets is the same thing as a v-98 or whatever. You write one, simple, code, and I-4's and V6's all run the same basic set up.

That said, I still see plenty of use in 8 "real" channels (8 seq, or 4+4), and 12 would cover basically any realistic set up you could imagine:
V12 seq
I4 seq + 4 seq staged + 4 seq water (seems overkill, but you could do it)
V6 seq + 6 seq staged/water
V8 seq + 4 staged (semi-seq per pair per side)
V10 + 2 staged
And more complicated stuff like
V6 seq + 2 staged + 2 banked water + ???
V8 seq + 2 staged + 2 banked water

I think (what I call batched and you call semi-sequential) is fine for secondary type injectors (water, high flow rate fuel) - sequential is really only a concern at low speed low power.

I think seq on at least 8 would be a big plus. You could batch fire a second set of injectors and as long as you didn’t max out your seq controlled set, you can still have individual cyl fuel trim control.
...
Traction control would be a nice feature especially for real race cars. Having an idle rpm/MPH table would be neat, that's one reason I let the stock ecu control my iac valve. With vehicle speed we could do transmission controls too, a typical auto trans controller sells for several hundred US dollars.

Traction control that only noticed that the wheel spun 4 rotations when it was supposed to spin 2 isn't doing you much good. If you want to know where to put more timers, use 4 for bit-banged injection as Jean suggested, and put the rest watching generic wheels like that.

I have another project I want to discuss offline, but it relates.

Thanks for the intro. I would be very happy with 18 outputs(8ign, 8 seq,1 batch fuel, 1 batch meth), but I dont have a problem with semi seq model at this time. I'm definitely not the most experienced hardware/software guy on this board but , I would rather see this project grow with the hardware rather than trying to do too much with too little.

-sean

He goes even further than I would, asking for one batched for fuel. It's probably ok. I think ideal for a V-8 is 4, running each side like a pair-banked 4 cyl. But in reality I'm sure 1 injector channel would be fine.

[Fred]

What I meant is that the part doing the computations should be done such that changing the number of supported cylinders should be a minor thing or even involve no changes at all because it's all parameterized.

Parameterised, the perfect word for it :-) that's what I was getting at with rota/2stroke/siamese etc, if they can't be cleanly integrated into that solution in a generic way, then IMO they need a branch each.

Same for the timer/bang vs timer code. Once you start splitting like that you end up with a different kind of mess though : wheel types X control strategies X engine types

Or you end up with a horrid set of code in the device trying to figure out how to do everything at once for everyone and configure it at boot time or on the fly based on config.

Which is why I'm keen to narrow the field a bit.

I can see that timers could be reused for ignition because there probably wouldn't be any overlap. However, there could be inversion of order (theoretically, I have no idea if it is actually possible) which means special cases and the 2 events could happen on different sides of a wheel event. These are all special cases to be handled either in the ISR or the computation if you use different timer, everything is simplified because all those special cases disappear with separate timers.

That is a really good point. Another reason for trying to keep it simple and do one thing well. Once again, there is no reason why more can't be done later. Later being the key point here.

The injection end is not an angle event it is a time event. True you want to end at a certain angle but if the speed of the engine changes during the time you compute the start angle and the time you end then you don't finish at the same angle. That's true for both injector but if they have different pulsewidths, you don't start at the same time so the error is not the same.

I see no good reason to try starting them at different relative offsets. you calculate both start points when you calc the longest one, then you when the longest (timer) one starts, you calc the combined end point based on known start points and flows etc.

But what may not be is in the case where the 2 injectors are at significantly different distances from the intake valve. You then want to be able to change the timing at which they end to account for the different travel distances.

I tend to think that staged sequential injection is a bonus, and that provided that the timing is consistent, you are only bringing in staged injectors at upper rpms/boosts and therefore sequential becomes less important. If you are still sequential, whether you are slightly late or early is splitting hairs. You can just tune the combined stop point for both as a good compromise for both (or more likely, the correct timing for low revs idle/cruise conditions where it matters.

Also, in addition to the special cases mentioned for ignition, in the case of injection I think that widely different pulsewidths are possible for primary and secondary which means there could be overlap.

I intend that they always overlap in a predictable fashion, to do otherwise is to introduce massive extra complexity for a tiny gain on a setup that hardly anyone uses.

I would need to do the computation but I'm not convinced that you would not get to much jitter from that to not have quite a bit of noise if you wanted to extract HP measurements from the data. But if you say so, then it's fine.

I'm not sure if i detect tone in that last bit, but I wasn't trying to say its as good, just that it is good enough.

Please try to remember we are aiming to do really good fuel+ign and then see what is left for accessories. It seems silly to assign timers to traction control to get that massive power down when you could be using them to get more of the massive power and let something else take care of the traction if it is that bad.

Additionally, you are then getting into the "you can have this feature with your setup, but you cant with yours" type of thing which I would strongly like to avoid. If a feature is present, anyone should be able to use it. Using timers for traction = I can't have traction in my future RB30DET stagea build with full timer sequential.

Features should take a clear and definite back seat/second place to good basic control covering the majority of normal hipo users.

First, I don't even have one anymore (due to personal reasons that required me to get rid of most of the "toys"). That may change in the future and I may consider an engine swap but the fact that it is there and port injection hasn't been done successfully by anyone but Rover is reason enough to do it. And the A-series engine just fits. Even though it was already years if not decades old when the first Mini came out in 1959.

In that case I am even more inclined to ignore it in the short term in special configurations at least.

Firstly we wanted 4 channels with timing control :

This would be very useful for siamese-port engine with dual injectors per port since each of the 4 injectors could require different timing. There are ways of doing it with the 2 current MS2 injector drivers but it would allow more flexibility.

Now I'm thinking that you were vague AND I misinterpretted it as one per cyl rather than staged, but anyway...

You need a 3d timing map to configure it regardless, you also want mid point injection, then you want separate timing on the staged channels compared to the main rather than using larger mains or one per cyl etc. None of these things are required at all for any normal engine. I understand the desire to do it, and I'd be keen to be involved with trying to make it happen, but the code comnplexity costs and config costs seem very high for the support of a rather small minority group if you include them for all users. It seems to me that a branch that for example only supported 36-1/+1 trigger type and special tables/special algorithms etc to support it properly without compromising it's operation to fit with the norm, or the norms comfort/performance to fit with it. Do you see where I'm coming from?

It seems analogous to special needs groups getting govt grants for their religions and cultural practices when normal people get no such grants (effective tax rebates) because they are normal. i.e. pretty unfair.

If you can show me a detailed design that works good saimese control (NOT average) in with GREAT normal sequential control in a clean config freeish way, I'm all for it. I just can't see how it can fit together when the requirements are SO different.

The thing is, there are other blocks of code that should get all of our work that all of us need :
basic structure/tables/settings/interfaces
serial comms
etc etc
Being aware of making it compatible with siamese while designing is a good thing (provided it does not force massive unnecessary compromise) but at the end of the day, it's not all that much of a big deal to retrofit an extra timing table or two, add a few extra calcs, and vars to read from and change the wheel code to suit.

Anyway, I think basic non staged 2 injector proper siamese injection is a reality without any funky compromises regardless. What are the upper echelons or power for a 1300cc A engine anyway? How large do you have to go injector wise to support it? I'm guessing less than 300hp, probably less than 260hp. if you can only use 50% of the injector and carry only 2 injectors, you need 1200cc injectors to do it singly. They might idle ok with proper timer control, but maybe not i guess. Then again, the price of extreme specific power is always compromise in idle etc unless you go to matching extreme lengths to counter it (staged etc). I'll probably put 750 - 1000 cc injectors on my fe3 when I get home to try to max out the holset. I expect that it might be slightly rich at idle with it like this. That is the compromise of massive power :-)

If I'm being unreasonable, please tell me.

Next...