DIY Fuel injection thread.

[Roman]

I think the reason the regular traction control setup uses ignition and fuel cut %, is because it can respond instantly. 
Using something like boost control or e-throttle and there's latency involved and so weird shit can happen.
Plus with the normal traction control you can setup a slip limit, your tyre can put down its best grip while it's slipping at say 10% faster speed than non driven wheels. 

I think the best way to set things like e-throttle angles and boost control, is to control the torque output proactively rather than reactively, then use reactive measures like ign cut to trim it from there.
So develop a table that represents how much power the motor is allowed under different conditions.

I've had some ideas about speed based boost control though, basically you'd build a standardized table that represents how much horsepower the car can put down without wheel slip under "standard conditions" 
Horsepower has a fairly basic relationship against mass air flow so you could have an allowed mass air flow number at a given speed.
So for example to output the same torque amount to the wheels you'd need twice as much HP/air at 200kph than 100kph.
Then the load axis could be a trim pot or some maths channel automagic that perhaps looks at ignition cut % and trims the table up or down if you're wheel spinning too much or not enough.
Then boost control would work by having an "air per cyl estimated" target for a given speed and trim value. 
So "theoretically" if you started wheel spinning at 80kph you could adjust your trim pot and it would rescale the rest of the map to still give you highest torque at the wheels that the car is capable of at each speed

/Just get slicks haha
 

[ajg193]

Also all four wheels can be travelling at different speeds without slip - like when going around a corner. You'd need to account for steering angle if you wanted to know the 'exact' moment of excessive slip initiating.

[Yowzer]

Largely a bit technical, only in more recent years with the introduction of stability control does the traction control bother with steering angle, so it can clamp the brake on the individual wheel. Around a corner your combined front wheel speed will be a touch higher than the rear wheels anyway so you can work it no worries. It's not like you're incorporating individual wheel braking or anything.

[Honda Ass Dragger]

On 24/02/2021 at 13:23, Roman said:

I think the reason the regular traction control setup uses ignition and fuel cut %, is because it can respond instantly. 
Using something like boost control or e-throttle and there's latency involved and so weird shit can happen.
Plus with the normal traction control you can setup a slip limit, your tyre can put down its best grip while it's slipping at say 10% faster speed than non driven wheels. 

I think the best way to set things like e-throttle angles and boost control, is to control the torque output proactively rather than reactively, then use reactive measures like ign cut to trim it from there.
So develop a table that represents how much power the motor is allowed under different conditions.

I've had some ideas about speed based boost control though, basically you'd build a standardized table that represents how much horsepower the car can put down without wheel slip under "standard conditions" 
Horsepower has a fairly basic relationship against mass air flow so you could have an allowed mass air flow number at a given speed.
So for example to output the same torque amount to the wheels you'd need twice as much HP/air at 200kph than 100kph.
Then the load axis could be a trim pot or some maths channel automagic that perhaps looks at ignition cut % and trims the table up or down if you're wheel spinning too much or not enough.
Then boost control would work by having an "air per cyl estimated" target for a given speed and trim value. 
So "theoretically" if you started wheel spinning at 80kph you could adjust your trim pot and it would rescale the rest of the map to still give you highest torque at the wheels that the car is capable of at each speed

/Just get slicks haha
 

This is some good food for thought - I've noticed a lot of Emtron users are using a ETB to bleed off boost (I think as a method of boost control)

I'll have to do some further thinking on this

 

 

[ajg193]

These look to be based on an early mega/microsquirt:

https://www.ebay.com/itm/motorcycle-ATV-scooter-buggy-moped-UAV-EFI-engine-Electronic-Fuel-Injection-kit/273127447029?hash=item3f97aa19f5:g:PHwAAOSwaudbUUTp

Tuned with a ripped off version of MegaTune

[tortron]

 

[tortron]

 

[tortron]

 

[Roman]

Just hooked up an electric water pump to control via ECU. 
So uses a spare output with varying duty cycle at 500hz to control the pump speed.
Then the pump has another pin where it can output the pumps actual speed back to the ECU. 
So for testing I hooked it up to vary the speed with throttle position.
So looks like below 20% and pump stays off, then above 70% it's flat lined at full speed. 
It looks like if it thinks there's an error/pump jammed (as seen below 20% part) it zings out a high pump speed number.
So will see what options I have to hook these up to a "pumps fucked" light.

But then also it will be interesting to see what axes on a table will be meaningful to control the pump speed. 
I've added a radiator temp sensor too, so maybe an ECT / rad temp table 

Or maybe ECT / injector duty cycle, because the more fuel you're using either means high load at low rpm or high rpm.
Both of which likely require more pump speed

 

[Testament]

@Roman wouldnt you want a minimum speed, rather than off completely at low/no load?  unless thermosiphon morris minor stuff goes on?

[Roman]

Yeah that was why I needed to find out minimum duty cycle to keep it running. So I can put values in the table between 20% and 70% rather than 0 and 100.

[gibbon]

Is megasquirt version 1 still viable? I hear it's "obsolete" but so is my car

Seems like the mk2 has more processing grunt but not sure how much that matters. All I need is the basics, fuel spark and knock control, with couple of additional on/off outputs 

Additionally there's a Russian guy selling ms2 knockoffs on eBay which are pretty cheap, wonder if they're worth looking at 

[kws]

Any reason you would be looking at old MS vs a Speeduino? More or less the same function but current and continuously developed.

[gibbon]

The only thing turning me off Speeduino is no knock sensor support (that I'm aware of) 

[kpr]

Cant say i've used it on ms.  but just because it supports a knock sensor / detection.  doesn't mean its anygood.   and its up to the end user to  set it up correct.   in most cases to will have to run your engine into light knock to map out the knock thresholds. Its not something that will help you tune your engine.    or would i rely on it  if you tuned your car for say 98 and threw in some 95.   personally i dont use it on any of my stuff.  just tune it for the fuel you are going to use and be done with it

[gibbon]

I'm not opposed to picking up a Speeduino, would you recomment .3 or .4? I'm completely new to the concept, I see that the preassembled kits are "Arduino not included" - is there a specific version of Arduino to look out for?

[Roman]

+1 to knock sensor overrated. 
One thing is that you can set it up perfectly on a dyno to detect knock in a really non noisy environment (car sitting on rollers) 
But then you go somewhere with real life vibrations (ripple strips, gravel, whatever) and you get lots of false triggers because the noise floor has raised.
Then sometimes in real situations the signal/noise ratio of actual knock is really marginal.
Personally I have it setup for observational purposes but could happily take it or leave it. 

[kws]

Arduino Mega 2560, I used ones i got from Aliexpress and had no issues but there are risks around quality if you go too cheap.

The Speeduino manual is very useful, https://wiki.speeduino.com/en/Getting_Started

I have used the 0.3 board in my Rover and it worked well, but the main difference is that you cannot easily drive a stepper motor idle controller with the 0.3 board and the wiring connections all terminate at screw connectors on the board. The 0.4 has space for a stepper controller, and the wiring all comes from a 40 pin IDC connector, so its a lot neater. I still have a 0.4 here that I will put into a project at some point.

Im a big fan of the Speeduino for cost and features, and the support from both the creator and forums is huge.

[ajg193]

Microsquirt works well enough, but is probably overpriced for what it is. The biggest drawbacks are that the fueling models incorporated in MS aren't really that great and take a lot of fiddling to get perfect

If I were to do EFI again I would probably go for a Link Atom or maybe speeduino if it wasn't a daily driver

[gibbon]

17 minutes ago, ajg193 said:

 maybe speeduino if it wasn't a daily driver

What makes Speeduino unsuitable for daily driving duties?