Custom stimulator with VVT!
#23
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,977
Total Cats: 356
Ken, give me more clues like:
1) Pins to make it easy to attach oscilloscope probes for the CKP/CMP signals
2) Prefererences on how to hook up the megasquirt on my stim: DB37, etc...?
3) I'll add a 2x16 LCD to it, do you want any other data on the LCD?
4) Currently my stim is not using timers to generate the pulses, so above about 3000rpm, the MS is losing sync. Is 3000rpm ok, or do you need to go higher? I'm thinking that 3000rpm should be ok.
5) Pullups for the CKP/CMP signals on the stim or on the MS?
6) Something else that I can't think of right now?
Dimitris
1) Pins to make it easy to attach oscilloscope probes for the CKP/CMP signals
2) Prefererences on how to hook up the megasquirt on my stim: DB37, etc...?
3) I'll add a 2x16 LCD to it, do you want any other data on the LCD?
4) Currently my stim is not using timers to generate the pulses, so above about 3000rpm, the MS is losing sync. Is 3000rpm ok, or do you need to go higher? I'm thinking that 3000rpm should be ok.
5) Pullups for the CKP/CMP signals on the stim or on the MS?
6) Something else that I can't think of right now?
Dimitris
#24
2) Prefererences on how to hook up the megasquirt on my stim: DB37, etc...?
3) I'll add a 2x16 LCD to it, do you want any other data on the LCD?
4) Currently my stim is not using timers to generate the pulses, so above about 3000rpm, the MS is losing sync. Is 3000rpm ok, or do you need to go higher? I'm thinking that 3000rpm should be ok.
5) Pullups for the CKP/CMP signals on the stim or on the MS?
6) Something else that I can't think of right now?
Don't rush getting it b/c I have plenty to get through before I get to VVT. I just wanted to make sure it's known that this will happen in 1.1.
Ken
#25
When the RPM is rising, the crank angle prediction tends to lag (retard), when it's dropping, it is advanced. So the worst from a timing error point of view would be a full-power clutch drop shift.
#26
Aside from that one example, the amount of inaccuracy with a 12+1 is going to be negligible.
I think on the scope during events like those I measured at most a few degrees off. I mean you have to REALLY try hard to get it to be off significantly with a toothed wheel that has more than 8 teeth.
My main concern is that with a turbo engine that's in gear, it starts to hurt power in first and second gear if you have a wheel with too few teeth b/c of the ignition timing retarding.
Also, your examples cover "heavy acceleration" but I was mainly concerned by "heavy acceleration under heavy load," which only the clutch drop covers.
Ken
I think on the scope during events like those I measured at most a few degrees off. I mean you have to REALLY try hard to get it to be off significantly with a toothed wheel that has more than 8 teeth.
My main concern is that with a turbo engine that's in gear, it starts to hurt power in first and second gear if you have a wheel with too few teeth b/c of the ignition timing retarding.
Also, your examples cover "heavy acceleration" but I was mainly concerned by "heavy acceleration under heavy load," which only the clutch drop covers.
Ken
#27
fwiw with the AEM reading only 2 teeth, revving in neutral from idle, I saw 4* of retardation, very briefly at low RPM. It very quickly diminished to 2 then 1 degree.
Remember it's worse at low RPM. And I have a 12 lb flywheel, so engine acceleration in neutral will be way quicker than most any car in 1st gear.
With the AEM now reading 6 teeth, the worst I could see was 2*, and it diminished quickly to <1*.
J_Man reading 12 teeth barely sees half a degree.
He and I did another test, idling then turning one injector off to force a misfire. I could see 1* of error on one cylinder, he could see none. And he barely has any flywheel at all.
Based on this, I reckon 12+1 will be plenty - provided the crank position prediction algorithm is robust - barring a full power clutch drop shift example - but then again, that's at high RPM where errors are lower.
Remember it's worse at low RPM. And I have a 12 lb flywheel, so engine acceleration in neutral will be way quicker than most any car in 1st gear.
With the AEM now reading 6 teeth, the worst I could see was 2*, and it diminished quickly to <1*.
J_Man reading 12 teeth barely sees half a degree.
He and I did another test, idling then turning one injector off to force a misfire. I could see 1* of error on one cylinder, he could see none. And he barely has any flywheel at all.
Based on this, I reckon 12+1 will be plenty - provided the crank position prediction algorithm is robust - barring a full power clutch drop shift example - but then again, that's at high RPM where errors are lower.
#30
I mean if the engine is at steady state, no load or RPM changes, there's always a little jitter in RPM due to mechanical factors, etc...
That's why you rarely see RPM perfectly flat, there's always some fluctuation in the 10's of RPMs even when cruising at steady RPM. That little bit of fluctuation can be magnified by a prediction algorithm depending on how it's done. At idle, the 1st deriv option in ms3/ms2-extra causes a tiny bit of jitter. I've seen similar things on other manufacturers' EMS's too.
We're porting over the ms2 standard firmware's "alpha-beta-gamma" prediction algorithm in ms3 1.1 though which is supposed to be much better all around than simple 1st or 2nd derivative prediction.
I usually turn off prediction for anything with 12 or more teeth.
Ken
That's why you rarely see RPM perfectly flat, there's always some fluctuation in the 10's of RPMs even when cruising at steady RPM. That little bit of fluctuation can be magnified by a prediction algorithm depending on how it's done. At idle, the 1st deriv option in ms3/ms2-extra causes a tiny bit of jitter. I've seen similar things on other manufacturers' EMS's too.
We're porting over the ms2 standard firmware's "alpha-beta-gamma" prediction algorithm in ms3 1.1 though which is supposed to be much better all around than simple 1st or 2nd derivative prediction.
I usually turn off prediction for anything with 12 or more teeth.
Ken
#33
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,977
Total Cats: 356
A 2-3 digit LCD is not something readily available, standardized, etc. If you have the space, go for a 2x16. If you don't, get 3 7-segment led displays. You will drive them at high frequency, one at a time. So for 3 7-segment displays, you need 7 pins for the leds of the segment and 3 selection pins.
This is what I did to my own dash, sometime ago: AFR instead of the dummy oil temp gauge, connected to my LM-1:
This is what I did to my own dash, sometime ago: AFR instead of the dummy oil temp gauge, connected to my LM-1:
#35
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,977
Total Cats: 356
Yes, there are serial LCD displays, but I can't think of any off the top of my head, that would fit in there.
You can use any HD44780-based LCD in 4-bit mode, and get away with 10 wires: 2 for power/gnd, 1 for lcd contrast, 1 for backlight, 2 for control signals and 4 for the data (hence the term 4-bit mode), but you drop the speed to less than half.
Dimitris
#39
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,977
Total Cats: 356
Yes, you would need to fabricate a two-layer pcb to help you with that, or two single sided board and join them via headers. I generally use veroboards and eurocards as much as possible, especially if I am doing anything that requires less than 20x of the same PCB.