DIYEFI.org Forum

I checked out the freeEMS pin assignment spread sheet and noticed that sequential injection will only be available on engines up to 6cyl. Does this mean that the 144 pin version of the hcs12 will be likely be used in future boards? I didn't a search on the subject and didn't find anything, but maybe I missed something.

-Sean

Short answer : the hardware does not allow (no matter what the pin count for the xdp512) proper control of sequential 8 cylinder operation, however, because V8's naturally make a lot of power, semi sequential mode where two timed shots 360 crank degrees apart are injected such that one waits in a port and the other is ingested directly should provide VERY pleasing results. The reason being that each cylinder will always get the exact same charge distribution. Sequential and single shot only really becomes critical when injector size is excessive. 550cc injectors should idle brilliantly using two shot semi mode on a v8, and supports up to 800hp, which is more than likely enough for you?

Having said that, there is a way of doing sequential on 8, 10, 12 cylinder engines with almost the same precision as using hardware timers directly, read about it and discuss it here :

http://www.diyefi.org/forum/viewtopic.php?f=8&t=76

It's hard to know if such code with its slightly reduced accuracy would generate a better running engine than the perfectly timed two shot arrangement with the best possible accuracy.

When the XEP100 becomes commonplace we can switch to that being the main platform and do sequential on 12 with its 16 timer channels :-)

Another option that could be discussed in this thread is to use external hardware to route and multiplex the timer pins to two cyls each.

Admin.

Admin wrote:Short answer : the hardware does not allow (no matter what the pin count for the xdp512) proper control of sequential 8 cylinder operation, however, because V8's naturally make a lot of power, semi sequential mode where two timed shots 360 crank degrees apart are injected such that one waits in a port and the other is ingested directly should provide VERY pleasing results. The reason being that each cylinder will always get the exact same charge distribution. Sequential and single shot only really becomes critical when injector size is excessive. 550cc injectors should idle brilliantly using two shot semi mode on a v8, and supports up to 800hp, which is more than likely enough for you?

Yes that is enough for my application, but I have built some 1600HP setups that could really benefit from sequential. Batch fire/staged injection has always worked well for me in demanding applications. But I think it would be nice to trim each cylinder individually. Thermo couplers are a cheap way of dialing in each cylinder.

Admin wrote: Having said that, there is a way of doing sequential on 8, 10, 12 cylinder engines with almost the same precision as using hardware timers directly, read about it and discuss it here :

http://www.diyefi.org/forum/viewtopic.php?f=8&t=76

Very interesting! and I think we should try to do seq on up to at least 8 cyl. If it proves to be a big headache and the code is somewhat "modular" continuing efforts on the XEP100 wouldn't be too big a deal in the future.

Admin wrote: It's hard to know if such code with its slightly reduced accuracy would generate a better running engine than the perfectly timed two shot arrangement with the best possible accuracy.
[/quote="Admin"]

As you know injector timing isn't nearly as critical as ignition timing and as long as it injects fuel some where along the 720 deg cycle(assuming 4 stroke) you can still have the ability to trim cylinders(for what its worth ).

Admin wrote:
Another option that could be discussed in this thread is to use external hardware to route and multiplex the timer pins to two cyls each.

Admin.

Another option indeed. Digikey should have my resistors/leds/photo-coupelrs/caps here next week hopefully soon after that I can get more up to speed.

-sean

seank wrote:As you know injector timing isn't nearly as critical as ignition timing and as long as it injects fuel some where along the 720 deg cycle(assuming 4 stroke) you can still have the ability to trim cylinders(for what its worth :) ).

Yes, it's not so much the timing, but rather the length that is critical, as you also know i'm sure :-)

And yes, at the very least, it shouldnt be hard to write a module to do the alternative scheme, I'm sure you would be happy to test it :-)

Hopefully i'll get some work done over the weekend (got NONE done today while answering the goals thread posts) and post up a timer based LED flasher for you to go forth and conquer all things nissan and gm with :-)

I can work on denso/36-1/X-Y and you can work on nissan and we can share and discuss ideas publicly such that others can chime in :-) should be sweet.

No point you doing the same ground work that I have already done though, best you focus on application of the principles to your specific problem.

Anyway, this belongs elsewhere :-)

Admin.

To improve on the precision of the bit-banging of the injector pins I was thinking of using the XGATE to do the pulse width timing. I'll start a new thread on this but here is an overview of what I've thought about.

The timer to start injection would still be needed but there would no longer be a need for the timer for the end-of-pulse. Also, it would be possible to have PWM for low-Z injectors (even though I'm not a fan of this method and the hardware required). It would also allow mapping the injection channels to different pins and multiplex them for siamese or TBI solutions if someone wanted such a configuration. I'd like to have 16 injector pins to allow staged sequential injection for up to 8 cylinders and up to 16 sequential single injectors (if someone wants such a beast). Of course, that requires the main code to be able to handle those numbers but that shouldn't be a big issue: it's the same algorithm but it requires more memory for tables (or whatever is used) and more time for computing the additional channels.

I'll have to start playing with my XEP100 demo board and the XGATE assembler before I can do anything useful which might take a while but in any case I want to define the requirements and interface correctly before I start the real code.

By the way, I wouldn't hold my breath for the XEP100 to become available. It seems Freescale is not interested to make it available to anyone but big OEMs with the corresponding big quantities. So using the XGATE looks like a more promising solution than waiting for a chip with the needed timers.

Jean

Thanks for that Jean, while I was typing my last post that thought crossed my mind, but just before I was going to type it it evaporated :-)

I look forward to seeing your ideas :-)

I think it would be a good thing to have true Timer control as a bench mark in any case so we can really tell how much we lose (or not) by going these other routes.

Admin.

A simple question I keep putting off asking: What is XGATE and how does it work? I was ok glossing over it, but "use xgate so you don't have to use the timers" does seem to rely heavily on knowing what xgate is.

An 80MHz co-processor that runs independently of the main cpu core with a limited RISC instruction set and full access to most of the hardware. It might be fast enough to bit bang things semi accurately.

Fred wrote:Both VEMS and MS3 will do V8 sequential which we won't, but both VEMS and MS3 will cost 2 - 5 times as much as a bare FreeEMS solution.

Was decision to limit support to six injector channels due hardware limitations, software design or to keep cost low? As owner of 16 plug V8 it's a bit disappointing to hear about such limitation. While doing wasted spark and using four channels to drive eight injectors is acceptable doing it "right" would of course feel better.

Regarding target audience for FreeEMS I believe there would be market for V8 owners as well. Many people, including me, have sourced cheap late model 5.7L/6.1L Chrysler HEMI engines from junkyards with intention to use them on classic mopar muscle cars or hotrods. Currently cheapest solution is going to carburetor with custom spark controller and intake manifold which means $1500 investment. There's two setups I know running with megasquirt. One of them using Ford EDIS and I think another was based on MS1/Extra. Using FreeEMS with stock sensors (MAP for air and hall effect for CKP/CMP), injectors (8 pcs) and coils (8 COP each firing two plugs) would be dream come true for many building car on limited budget. Running 5.7L engines from wrecked trucks are already going as low as $500, but currently anything else than megasquirt controlling it equals to extra $1500 to $5000.

Is running 8 sequential with 8 COP in the territory of the X Gate stuff?
or can it be achieved with what is already planned, should the V8 guys have to settle with WS and semi sequential?
Could this be cause for a poll to see if there is a serious requirement for V8 functionality?
PS: did I mention my next car is a V8 BMW 535iS? and I want to be able to tweak it!