Help - Search - Members - Calendar
Full Version: Going to Megasquirt
914World.com > The 914 Forums > 914World Garage
Pages: 1, 2, 3
Frank S
QUOTE(bbrock @ Oct 9 2021, 09:51 PM) *

QUOTE(nditiz1 @ Oct 9 2021, 01:15 PM) *

I have only seen the crankfire setup that McMark was selling from Original Customs. With that said what is the adv to running the cam sensor?

If it was easier to install/cheaper big a significant amount/better signal/easier to setup I might entertain the idea. Are any of those advantages present? Also, I know Al Kosmal is currently working on a T4 - EFI setup.


The advantage is that I like to over complicate things biggrin.gif An objective is to get the best economy (especially in town) and lowest emissions I can. A cam sensor let's you go full sequential for spark and injection which in theory can give a little better driveability and economy at lower speeds, reduced emissions, and smoother idle. I know going sequential isn't going to gain me that much, but I like this kind of challenge.

The descriptions for Marios' old NLA cam sync sensor says it must be used with a crank sensor. I'm not sure if the same will be true for the new one which will have a trigger wheel instead of flying magnet. If it were possible to get by with just a cam sensor and no crank sensor, you wouldn't have to drop the engine to do the conversion. I guess that might simplify the install regardless of spark and injection mode.


Mario's cam sync is working togeher with the crank sync only. But you can run a missing tooth sensor in the dizzy for sequential ignition and fuel, so no need for the crank sensor. The dizzy sensors are also much easier to service if the sensor is failing...
JamesM
QUOTE(bbrock @ Oct 9 2021, 06:44 AM) *

- Do I need a crank sensor?


If you want to run any ignition control it is highly recommended. If you want to run full sequential i believe it is mandatory.

I highly recommend Mario's crank trigger setup for multiple reasons, however as you mentioned it does pretty much require you drop the engine in a 914 to install and/or service it. Well, maybe not required but you are going to want to.

More recently I have came across this crank sensor setup that looks like it would be easily serviceable with the motor in the car, but i havent personally run it so i cant speak at all to how well it works. Also its expensive and looks like it probably uses a VR sensor (one of the reasons i prefer Mario's setup is it uses a Hall sensor that gives a nice clear square wave output)

https://www.clewett.com/index.php?main_page...;products_id=31


QUOTE(bbrock @ Oct 9 2021, 06:44 AM) *


- I'd love to hear thoughts on whether modern injectors would be worth the cost if I don't upgrade the fuel rails. Still not sure what I want to do, but curious why modern fuel rails would make a difference.


I see no need to swap out the fuel rails.

These are my preferred injectors when running stock intake runners

https://www.fiveomotorsport.com/a280-a380-h...-fuel-injector/

Lots of advantages in my mind:
They are pretty much plug and play to the stock intake runners and fuel rails.
They have locking EV1 connectors
They are high impedance so no need to use inline resistors or run PWM control
They run at higher pressures for better atomization
They come documented with flow specs and voltage correction datasheet
They are brand NEW

Only potential downside is that you will most likely want to run them with a newer style fuel pressure regulator to run at higher pressures, but there are a few options there that easily install in place of the stock regulator.

I think the deciding factor though should really involve asking yourself the question, do i trust 50 year old injectors and if so, for how long?

A few years ago I was driving behind @StratPlayer out of Salt Lake on the way to RRC and wound up getting my car covered in oil when a sticky stock injector holed a piston in his brand new 2056. Given all the other modernizations you are putting in, i see no reason to stick with old injectors. Not to mention, the stock 2.0 injectors are absolutely HUGE for a 2.0L motor resulting in very small injector pulse widths at idle which produces less accurate fueling. In fact, back in the VERY early days (pre MS2) I had to run a special "hi-res" code variant to gain the resolution needed for proper idle control with the stock injectors. If you do want to run d-jet injectors the ones off a 1.7 are much more appropriately sized for the application.
ClayPerrine
Add a 36-1 wheel behind the fan replacing the spacer, and use a Vanagon hall effect distributor for the cam sensor.

The vanagon distributor is a drop in for a Type-IV.

Clay
bbrock
This is all great info everyone! Looks like I'll be adding a crank sensor which was the original plan anyway, but seeing that his new cam sensor will have a trigger wheel (I assume missing tooth) got me thinking.

Thanks @JamesM for pointing me toward those injectors. I wasn't very clear before but I had already decided to go with new injectors. The question was whether to upgrade the fuel rails which I'd rather not do to keep the stockish appearance. I agree that going to all this trouble and running old injectors doesn't make much sense. I figured I'd need to replace the stock pressure regulator too.

You raise an interesting question about injector size though. I read somewhere that for running full sequential injection, it is better to err on the larger size for injectors. The rationale was that larger injectors allow for a shorter pulse width that allows for more of the squirt to go into an open intake valve and therefore could spread the benefit of sequential a little higher up the rpm range. Made sense to me but wonder what others think.

The vanagon distributor is an interesting idea. Didn't know that was an option.
JamesM
QUOTE(bbrock @ Oct 9 2021, 04:17 PM) *


You raise an interesting question about injector size though. I read somewhere that for running full sequential injection, it is better to err on the larger size for injectors. The rationale was that larger injectors allow for a shorter pulse width that allows for more of the squirt to go into an open intake valve and therefore could spread the benefit of sequential a little higher up the rpm range. Made sense to me but wonder what others think.



I think the theory is sound, shorter pulse width will buy you more time firing on an open valve. You could probably run some math to figure out what the exact rpm difference would be between a larger and smaller injector. Looking at the math before though was what made me loose interest in sequential in the first place, at least for my autox car. Only gains are in emissions/economy at idle/lower RPM and once past a certain injector duty cycle and depending on your cam duration you are firing on a closed valve either way. Given how my autox car is normally driven, wouldnt have had any advantages for me.
Also with these motors usually having to be tuned on the rich side everywhere, including idle anyways i don't know if it would wind up making any measurable difference at all.
bbrock
Okay, here's some fun with math. I calculated the theoretic time injectors on my engine need to be open to spray a dose as a percentage of the time the intake valve would be open. Not 100% sure my math is right but I got help on calculations here:

https://www.team-integra.net/threads/calcul...se-width.46738/

and here:

https://www.physicsforums.com/threads/findi...ns-open.584241/

The target AFR numbers are very tentative based on some earlier threads on the subject. I'm assuming I'll want the leanest mix at idle and cruising around 3000 rpm. Between idle and 3000 rpm would mostly be mild to moderate acceleration under load, and above 3K will mostly be WOT. I've barely dipped my toe into learning about managing mixtures, so those numbers might change as I learn more about AFR and how megasquirt can adjust timing and fuel dose in response to CHT. I think they are reasonable for this thought exercise though.

[Data Corrected]

Anyway, here are the results:

Click to view attachment

Click to view attachment
Superhawk996
I didn't check your math. I'm impressed though. No one bothers to do the math. Engineer at heart I guess! happy11.gif

I sure hope it's snowing out there . . . otherwise you're wasting valuable driving time doing injector math. biggrin.gif Just sayin'.

popcorn[1].gif Great discussion.
bbrock
QUOTE(Superhawk996 @ Oct 10 2021, 04:15 PM) *

I didn't check your math. I'm impressed though. No one bothers to do the math. Engineer at heart.

I sure hope it's snowing out there . . . otherwise you're wasting valuable driving time doing injector math. biggrin.gif

popcorn[1].gif


Predicting a foot of it tonight. Trying to get a project done before it gets cold enough to freeze in our greenhouse so no driving this weekend.
bbrock
Found a major error in my calculations. I accidentally included a wrong column in the pulse width calculation. The corrected results are more interesting. It looks like with 26 lb injectors, it would start spraying part of the charge onto the backs of closed valves starting at about 1,500 rpm but with 36 lb injectors, that wouldn't begin until about 2,200 rpm. At 5k rpm about 29% of the charge from a 26 lb injector would spray into an open valve and that increases to about 40% with a 36 lb. injector.

Based on that, it seems like maybe larger injectors would be the way to go, but there must be a down side. Thoughts?
JamesM
QUOTE(bbrock @ Oct 10 2021, 05:03 PM) *

Found a major error in my calculations. I accidentally included a wrong column in the pulse width calculation. The corrected results are more interesting. It looks like with 26 lb injectors, it would start spraying part of the charge onto the backs of closed valves starting at about 1,500 rpm but with 36 lb injectors, that wouldn't begin until about 2,200 rpm. At 5k rpm about 29% of the charge from a 26 lb injector would spray into an open valve and that increases to about 40% with a 36 lb. injector.

Based on that, it seems like maybe larger injectors would be the way to go, but there must be a down side. Thoughts?



With larger injectors you lose some level of precision in fuel metering. I experienced this first hand when a early version of the MS1 code was unable to control the larger injector well enough for a decent idle.

At higher duty cycles both batch and sequential are putting the same amount of fuel though the open valve just as a result of how long the injector is firing for, so the only real difference is going to be seen below the point where the sequential starts firing on the closed valve.

So when comparing 26lb to 36lb injectors with sequential injection we are basically looking at a trade off between fueling precision across the entire operating range vs possible improvement in the 1500-2200 RPM range due to firing 100% on an open valve, and honestly how much time do you spend driving in that range? I think I would go with the smaller injector either way.

You really took me seriously on the math there! I cheated when i looked into it and just used recorded duty cycles across the RPM range for my setup from my datalogs, wound up with a similar result though, if I recall in my case the the cutoff was going to be somewhere around 1800rpm which for my purposes wasn't worth the effort/cost.

You give up the potential for some features going with Microsquirt but i think that platform has some advantages as well and in my mind it is the better compromise.
bbrock
QUOTE(JamesM @ Oct 11 2021, 02:21 AM) *

With larger injectors you lose some level of precision in fuel metering. I experienced this first hand when a early version of the MS1 code was unable to control the larger injector well enough for a decent idle.


I was reading about that last night. It sounded like this has been improved in later code versions. One discussion indicated it becomes a problem with pulse widths of 2 ms. Seems like it would be worth doing some more research.

QUOTE
At higher duty cycles both batch and sequential are putting the same amount of fuel though the open valve just as a result of how long the injector is firing for, so the only real difference is going to be seen below the point where the sequential starts firing on the closed valve.

So when comparing 26lb to 36lb injectors with sequential injection we are basically looking at a trade off between fueling precision across the entire operating range vs possible improvement in the 1500-2200 RPM range due to firing 100% on an open valve, and honestly how much time do you spend driving in that range? I think I would go with the smaller injector either way.


If my logic is right, batch firing delivers a charge in two squirts per cycle so a minimum of 50% of the charge will be sprayed onto a closed valve. That suggests some potential benefit up to 2000-3000 rpm depending on injector size because you'd have at least 25% more of the fuel charge spraying into open valves. Batch and sequential wouldn't be fully equivalent until at least 50% of the sequential charge goes on closed valves.

Engine speeds up to 3000 rpm covers pretty much all city driving, which has been about a third of the driving I've done in the car so far. Most of that is speeding up or slowing down and there isn't much chance to cruise a 3K rpm. That's exactly where I expect the most potential to improve fuel economy. Looking at my old mileage records for this car when driving on pure leaded gas, I consistently got mid to upper 30s on tanks of pure highway driving (remember it was 55 mph speed limit then) and don't expect much room for improvement there. But I got low to mid teens in town. Pretty horrible for such a small car really. I don't know if it is realistic, but low to mid 20s in town would be nice.

QUOTE
You really took me seriously on the math there! I cheated when i looked into it and just used recorded duty cycles across the RPM range for my setup from my datalogs, wound up with a similar result though, if I recall in my case the the cutoff was going to be somewhere around 1800rpm which for my purposes wasn't worth the effort/cost.


Ha ha. It was a fun little exercise. Glad to hear we came up with similar results. I wasn't sure how reliable this process for calculating would be.
jd74914
QUOTE(bbrock @ Oct 11 2021, 12:59 PM) *

QUOTE(JamesM @ Oct 11 2021, 02:21 AM) *

With larger injectors you lose some level of precision in fuel metering. I experienced this first hand when a early version of the MS1 code was unable to control the larger injector well enough for a decent idle.


I was reading about that last night. It sounded like this has been improved in later code versions. One discussion indicated it becomes a problem with pulse widths of 2 ms. Seems like it would be worth doing some more research.

I'm not sure this is really a 'code' issue. I would place it more in the loop execution time and hardware drivers. From what I've seen over the last decade or so, it doesn't appear the internal code is getting more efficient, the processers are just getting faster. So with that, best course of action for increased time resolution with MS looks to be going with newer hardware (ie: MS3).

At super low pulse widths you end up with this injector latency problem which can also pose some issues. On the 'cheap' injector side you aren't finding too many latency and flow matched sets. You could test that though...
bbrock
QUOTE(jd74914 @ Oct 11 2021, 12:38 PM) *

I'm not sure this is really a 'code' issue. I would place it more in the loop execution time and hardware drivers. From what I've seen over the last decade or so, it doesn't appear the internal code is getting more efficient, the processers are just getting faster. So with that, best course of action for increased time resolution with MS looks to be going with newer hardware (ie: MS3).

At super low pulse widths you end up with this injector latency problem which can also pose some issues. On the 'cheap' injector side you aren't finding too many latency and flow matched sets. You could test that though...


Good info and that makes sense. I'm wondering if going sequential helps with this problem. Since each injector squirts only once per cycle rather than twice as in batch injection, the pulse widths are double for a give fuel demand. Would that not help with problems of low resolution for large injectors at low speeds? Looking at my tables, it looks like the 36 lbs/hr injector would still have a pulse width of 9.6 ms at idle which doesn't seem too small.
jd74914
Pulses of ~10 ms aren't too bad.

But...

I don't see where you add an engine pumping efficiency (VE) into the air flow calculation. I didn't check your calcs from a PV=nRT sense so I'm assuming those are correct in terms of air/fuel mass flows with perfect cylinder filling.

The engine is only going to be pumping maybe 10-30%* 'efficiently' at idle making your actual fuel flow requirements much less. Pulses of 0.96 ms are too short. Most injectors have 1-1.5 ms latency (voltage and fuel pressure dependent).



*Edit: Found a post showing Falcor76 has ~20% VE @ idle so I think these numbers are somewhat reasonable.
Frank S
Fuel Injector Size:
https://thedubshop.com/pages.php?pageid=44
I'm running 21lb/hr and they are fine for a 120 HP Engine.
Don't go to large as this will be generate problems at idle and in overun conditions with to low pulswidth (out of the linear range of the injectors).
MS3 can deal much better with small puls width as the Dead Time Voltage Corretcion is not linear and MS3 is addressing that problem (which alone is a clear advantage if you compare with MS2 or Mircrosquirt).

Again Crank and/or Cam sensors:
A Crank sensor alone can only be good for Batch injection and Wasted Spark ignition.
Why? Because a engine cycle is 720° Crank rotation, so the ECU does not know if the engine is at #1 or #3 TDC. If you add a single tooth Cam Sensor in addition the ECU will be enabled to define #1 TDC, so you could run sequential fuel and spark.

If you run a Toothed Wheel Cam/Distibutor Sensor, this sensor allone will enable the ECU to define #1 TDC. This is because 720° Crank rotation = 360° Cam/Distributor rotation.

Sequential Modification:
You don't need to modify the MS2V3 board for sequential Fuel and Spark if you use the MS3. The MS3 daughtercard has all the outputs available, you just need to connect them to the JPB Board you are planning to use.
bbrock
QUOTE(jd74914 @ Oct 11 2021, 03:00 PM) *

Pulses of ~10 ms aren't too bad.

But...

I don't see where you add an engine pumping efficiency (VE) into the air flow calculation. I didn't check your calcs from a PV=nRT sense so I'm assuming those are correct in terms of air/fuel mass flows with perfect cylinder filling.

The engine is only going to be pumping maybe 10-30%* 'efficiently' at idle making your actual fuel flow requirements much less. Pulses of 0.96 ms are too short. Most injectors have 1-1.5 ms latency (voltage and fuel pressure dependent).



*Edit: Found a post showing Falcor76 has ~20% VE @ idle so I think these numbers are somewhat reasonable.


Excellent! Yes, there's a little obscure note in the chart that says VE is calculated at 100%. This is all new to me so I was going to refine the values to include more realistic VE numbers later. Any guidance on some standard VE values to plug in? So far I've only found typical values at max HP.

This is obviously critical to the analysis. So assuming 10% VE, the idle pulse width should be around 1.3 ms with 26 lb injectors and, as you said, 0.96 ms with 36 lb. That puts the nail in the coffin for the larger injectors it seems. It seems even the 26 lb injectors might be pushing the limit. Assuming 20% VE at idle, I get a 2.7 ms pulse width with the 26s (1.9 ms with 36s).

Playing with VE a little and assuming a static 85% through the range above idle stretches out the theoretic band where sequential could potentially benefit. I've added curves for % of charge sprayed into open valves below with a reference line at 50% to show the point where sequential is no different from batch. Once I have more realistic guestimate numbers for VE, I'll post updated tables.

Click to view attachment
crash914
Call Mario.

Ask about the MAX EFI. will integrate O2 sensor, cam sync, CNP, full sequential.

It works well.
bbrock
Well no reason to stop this insanity now, so here's more fun with math. I found a good discussion about VE vs engine speed for a stock 2.0L on Pelican. It didn't provide a nice table of VEs I would have liked, but led to enough info to make some WAGs to get closer with my chicken scratch estimates. I learned that VE typically follows the torque curve with peak VE being at about the same engine speed as peak torque. I also learned that typical peak VE values for 4 cylinder, 2 valve, overhead cam engines is 75-80% So I plotted a graph using 10% VE at idle to represent worst case in terms of small pulse width and max of 80% at about 3800 rpm which looks about where the 2.0L peaks (I was too lazy to look up the actual peak), and I dropped it to 60% @ 5800 rpm - again, just trying to match the shape of the torque curve. Then I had Excel calculate a polynomial equation for the shape and used that to plug in VE values in my tables. Whew - we're into some nerd shit now!

Click to view attachment

The results are interesting.

Click to view attachment

With this adjustment, both injectors extend the range where sequential would spray more of the charge into open valves than batch. Really intriguing is that the 36 lb injectors would still be spraying all or nearly all of the charge into open valves into the low 3ks which would be potential efficiency gains even into highway cruising speeds. In fact, the 36 lb never dips below 50% of charge going into open valves. But... there is still that pesky 0.94 ms pulse width @ idle. It would be nice to know what that minimum VE would actually be because even at 20% VE, the minimum pulse width climbs to 1.9 ms which might just be doable with MS3.

Click to view attachment
bbrock
QUOTE(Frank S @ Oct 11 2021, 03:49 PM) *

Fuel Injector Size:
https://thedubshop.com/pages.php?pageid=44
I'm running 21lb/hr and they are fine for a 120 HP Engine.
Don't go to large as this will be generate problems at idle and in overun conditions with to low pulswidth (out of the linear range of the injectors).
MS3 can deal much better with small puls width as the Dead Time Voltage Corretcion is not linear and MS3 is addressing that problem (which alone is a clear advantage if you compare with MS2 or Mircrosquirt).


I think I need to look a little more into MS3 or maybe the MAX EFI option @crash914 mentioned to see if they would let me run a smaller pulse width. MS3 would add about $200 to the project ($100 more than MS2 for the board kit plus $100 for the MS3X board to get the sequential output channels.

I've looked at Mario's and a couple other injector calculators and they all followed the same equation. Looks to me like they are just sizing the injector to be able to provide enough charge per cycle for the HP to be generated. Makes sense but it looks like squeezing efficiency our of sequential injection really depends on the timing of when the charge is sprayed to get as much as possible into the open valve for better atomization. Hence, the obsession over the larger injectors.

QUOTE
Again Crank and/or Cam sensors:
A Crank sensor alone can only be good for Batch injection and Wasted Spark ignition.
Why? Because a engine cycle is 720° Crank rotation, so the ECU does not know if the engine is at #1 or #3 TDC. If you add a single tooth Cam Sensor in addition the ECU will be enabled to define #1 TDC, so you could run sequential fuel and spark.

If you run a Toothed Wheel Cam/Distibutor Sensor, this sensor allone will enable the ECU to define #1 TDC. This is because 720° Crank rotation = 360° Cam/Distributor rotation.


Good info!

QUOTE
Sequential Modification:
You don't need to modify the MS2V3 board for sequential Fuel and Spark if you use the MS3. The MS3 daughtercard has all the outputs available, you just need to connect them to the JPB Board you are planning to use.

I looked into the daughter board and you still have to add the MS3X board to get the sequential output channels. Seems kind of weird MS3 doesn't have at least 4 injector channels built in, but I haven't found anything to indicate otherwise. The MS2 mod is really just removing a few components and adding some jumpers to repurpose unused features as output pins for injectors. If you are building the board from kit, the mod is more just skipping a few steps on assembly. Still, if MS3 can handle short pulse widths, that would be a good reason to use it.
Frank S
[quote]Sequential Modification:
You don't need to modify the MS2V3 board for sequential Fuel and Spark if you use the MS3. The MS3 daughtercard has all the outputs available, you just need to connect them to the JPB Board you are planning to use.
[/quote]
I looked into the daughter board and you still have to add the MS3X board to get the sequential output channels. Seems kind of weird MS3 doesn't have at least 4 injector channels built in, but I haven't found anything to indicate otherwise. The MS2 mod is really just removing a few components and adding some jumpers to repurpose unused features as output pins for injectors. If you are building the board from kit, the mod is more just skipping a few steps on assembly. Still, if MS3 can handle short pulse widths, that would be a good reason to use it.
[/quote]

I wasn't clear about the board. I thought about adding this:
http://jbperf.com/quad_ign_inj/index.html
Instead of the MS3X board.

Regarding your fuelinjector thoughts, you don't want to spray into the open valve.

MS3 is providing so much more value over MS2 that it is fully worth the extra $$ and I think if you would talk to the guys at DYI Autotune, they would also sell a MS2 kit without a MS2 daughtercard and adding a MS3 daughtercard therefore, which will remove some cost at least.

The MS3 dauthercard will also fit into the MS2 case if you slightly modify it and the MS2 case will also fit into the original D-Jet housing if you want to stay with the original look.
JamesM
It sounds like you really want to go with the 36lb injectors biggrin.gif

Larger injectors will always have less resolution smaller ones BUT 36lb injectors wont be an issue with an MS2 or newer CPU, it was only MS1 hardware that lacked the resolution under the stock code base to control them properly at idle. If I recall it wasn't so much an issue with the speed of the CPU as it was the limited CPU register capacity which is why they were able to address it with a custom code version that gave up other features in order to gain extra register space. We are talking ancient history here (15 or so years ago when all the MS code was written in assembly) so im not 100% on the specifics, but operationally hasn't been an issue since the early days, I ran the stock 2.0 (36lb) injectors for years, no big deal.

That being said, because you do loose some resolution with the larger injectors you will want to be sure all your injection related settings, especially the injector dead time and voltage correction factors are set as exact as possible (something that is impossible with stock injectors) as having the larger injectors amplifies mixture issues due to minor voltage fluctuations.
Given that fact along with what you seem to be going for here 'Frank S' brings up a good point about MS3 actually having a full table for injector voltage correction vs the MS2s linear correction. I had completely forgotten about that feature difference but am now reminded that the linear correction was one of the things that slightly annoyed me once I changed over to injectors that actually had documented voltage correction tables with them. Given you shouldn't see more than a 1-2 volt delta under normal operation conditions the linear settings get you pretty damn close (i can kick on my headlights with no AFR change noted) so for me it was splitting hairs, but if you really want to optimize it with the larger injectors, having the full correction table may be the way to go.

As for PW at idle, ill have to see if i have any old logs, but if i recall on a stock motor with the large injectors I was sitting somewhere in the low to mid 2ms range. Issue with the original MS1 code though was that it could only control PW in .1ms increments and the AFR difference between a 2.2ms and 2.3 ms injection pulse at idle with a 36lb injector was surprisingly large. Thankfully that is no longer an issue.

I know you haven't built your system yet, but based on what i have seen you doing so far I'm going to call it now and say you are going to wind up with the most precise (and thoroughly planned out) MS build to date.
first.gif
jd74914
QUOTE(bbrock @ Oct 11 2021, 08:45 PM) *

QUOTE
Again Crank and/or Cam sensors:
A Crank sensor alone can only be good for Batch injection and Wasted Spark ignition.
Why? Because a engine cycle is 720° Crank rotation, so the ECU does not know if the engine is at #1 or #3 TDC. If you add a single tooth Cam Sensor in addition the ECU will be enabled to define #1 TDC, so you could run sequential fuel and spark.

If you run a Toothed Wheel Cam/Distibutor Sensor, this sensor allone will enable the ECU to define #1 TDC. This is because 720° Crank rotation = 360° Cam/Distributor rotation.


Good info!

I actually have a toothed cam wheel with hall effect trigger somewhere in the garage. Never used it, but let me see if I can dig it out to show how it was done.

jd74914
QUOTE(bbrock @ Oct 11 2021, 08:45 PM) *

I've looked at Mario's and a couple other injector calculators and they all followed the same equation. Looks to me like they are just sizing the injector to be able to provide enough charge per cycle for the HP to be generated. Makes sense but it looks like squeezing efficiency our of sequential injection really depends on the timing of when the charge is sprayed to get as much as possible into the open valve for better atomization. Hence, the obsession over the larger injectors.

What you're seeing is typical of automotive aftermarket...just general equations without much value besides really rough approximations.

I'm really enjoying seeing your calculations-thank you!

QUOTE(bbrock @ Oct 11 2021, 08:45 PM) *

I think I need to look a little more into MS3 or maybe the MAX EFI option @crash914 mentioned to see if they would let me run a smaller pulse width. MS3 would add about $200 to the project ($100 more than MS2 for the board kit plus $100 for the MS3X board to get the sequential output channels.


Personally, I'd push to an MS3...but its easy to spend other people's money. With an MS2 loop speed of 0.33 ms, assuming engine speed of 3000 rpm, you're looking at the engine moving 5.94 degrees per loop execution time...MS3 @ 0.2 ms gives you 3.6 degrees. Pretty decent difference. Caveat here-I'm not really sure how the code is structured. You would hope they would prioritize some things like injector/ignition switching as well as engine position calculation, but I'm not sure.

As a side note, The extreme high end stuff has hardware angle clocks (typ. FPGAs) to keep these kind of calculations off of the main processor. The use the main processor to schedule events to happen at definitely crank angles vs. times. Subtle difference that doesn't really matter for a 50 year old engine, but maybe interesting.

QUOTE(JamesM @ Oct 12 2021, 02:17 AM) *

Given that fact along with what you seem to be going for here 'Frank S' brings up a good point about MS3 actually having a full table for injector voltage correction vs the MS2s linear correction. I had completely forgotten about that feature difference but am now reminded that the linear correction was one of the things that slightly annoyed me once I changed over to injectors that actually had documented voltage correction tables with them. Given you shouldn't see more than a 1-2 volt delta under normal operation conditions the linear settings get you pretty damn close (i can kick on my headlights with no AFR change noted) so for me it was splitting hairs, but if you really want to optimize it with the larger injectors, having the full correction table may be the way to go.
first.gif

agree.gif

I think this correction table is very important. Not a direct comparison, but when tuning high-strung bike motors we found having full voltage correction was exceedingly important. Those were engines with poor charging capability and weight optimized harnesses so voltage drop was a thing. As James said, likely not a huge issue for you but when really pushing it...

Now, you've been looking at all of this in terms of fueling during valve opening. This isn't necessarily ideal when you start looking at air/spray flow velocities and mix transit times into the cylinder. The delay is real, particularly due to the variation in intake velocity with crank angle as the valves open. For real optimization you need to tune the injector end angle at all engine speeds and load points...

https://www.motec.com.au/forum/viewtopic.php?t=769

I've been fortunate enough to have a load bearing dyno to do sweeps, but this should be doable on the road. Just needs some more logging and drive time. Optimized injection timing minimizes wall condensation and actually can yield some pretty meaningful efficiency gains. You do need good engine position resolution to do this well which really pushes towards the crank sensor (say 36:1 or 60:1 wheel).
Superhawk996
You're becoming a real Geek. laugh.gif

I've had these sitting on my shelf since college. I think you need them more than I do. PM me your address and they are on their way to you.

Click to view attachment


bbrock
You guys are awesome! smilie_pokal.gif

I was reading through the MS3 tuning manual last night and it does look like MS3 will be worth the added coin. I've officially gone from Microsquirt all the way up to MS3. I'm still a bit confused about cards though and am probably missing something. It looks like the daughter boards are intended to upgrade older boards to newer so an MS2 daughter card is used to upgrade an MS1 to MS2 and MS3 daughter upgrades MS2. Either of these provides only 2 injector channel outputs and the jbperf boards require 4 inputs to drive 4 injectors independently. So it seems you need to modify/expand either MS board to get 4 output channels. Difference being that MS3X can drive high impedence injectors directly where modded MS2 requires a jbperf board. I might be missing something though.

@jd74914 you and @Frank S anticipated my next question. I knew talking about spraying into open valves was likely incorrect but was using it as shorthand for getting the most complete atomized charge into the combustion chamber. I think/hope the logic for my calculations still applies even though they ignore the timing issue. Last night I read about crank angle settings in the MS3 manual. They said you usually want the injectors to stop spraying just as the valve begins to open and suggested 360 degrees as a starting point. They then say further tuning was needed with dyno or road testing and my obvious question was, how? The link to the motec article answers that perfectly! beerchug.gif

Now after reading the MS3 manual about individual cylinder tuning, I'm looking at my Bursch collector piping and thinking, hey, I could mount 4 lambda sensors in there ... av-943.gif
VaccaRabite
I worry that the larger picture of building an EFI for a 4 banger aircooled motor is getting lost in the weeds here.

These aren't exactly smooth running motors to begin with, and the engine bay is somewhat open to the elements. I'd reconsider using the Microquirt and adding some simplicity back into your equation. The operational differences between wasted spark and individually controlled plugs isn't going to be noticeable when you are driving the car, but will make it more challenging to build and tune - increasing the time before you are back to driving the car again.

Zach
bbrock
QUOTE(VaccaRabite @ Oct 12 2021, 10:22 AM) *

I worry that the larger picture of building an EFI for a 4 banger aircooled motor is getting lost in the weeds here.


Was it that obvious? av-943.gif

You've probably given the best advice of all here, but I'm already down the rabbit hole. Here's the view from my window today so I'm already done driving until the middle of June '22. I need something to occupy myself on these long winter nights biggrin.gif

Click to view attachment
JamesM
QUOTE(VaccaRabite @ Oct 12 2021, 08:22 AM) *

I worry that the larger picture of building an EFI for a 4 banger aircooled motor is getting lost in the weeds here.

These aren't exactly smooth running motors to begin with, and the engine bay is somewhat open to the elements. I'd reconsider using the Microquirt and adding some simplicity back into your equation. The operational differences between wasted spark and individually controlled plugs isn't going to be noticeable when you are driving the car, but will make it more challenging to build and tune - increasing the time before you are back to driving the car again.

Zach


I already tried to sell him on the easy button (well slightly easier button anyways). biggrin.gif

As someone who enjoys this stuff though I full support the effort to go completely overkill!

beerchug.gif

I'm actually sort of curious about what will be needed to properly dial in the injector timing and what sort(if any) difference it will wind up making. The tuning process with batch injection is fairly straight forward for the home mechanic with a wideband sensor but to fully dial in sequential injector timing... I suspect that's going to take some serious time and heavy equipment.

Lots of stacked variables to deal with.
JamesM
Im just going to leave this here to give you more to consider about your injector selection. poke.gif

This is the datasheet I got with those A280H injectors

The 26lb rating was at 3bar pressure however they provide the flow specs and voltage corrections for running them all the way up to 7 bar!!!

I would imagine the 36lb sets come with similar abilities.

Have read that higher pressures improve atomization, and if you wanted to look at shortening your injector pulse even further, its something to consider.

Click to view attachment

bbrock
QUOTE(JamesM @ Oct 12 2021, 12:04 PM) *

QUOTE(VaccaRabite @ Oct 12 2021, 08:22 AM) *

I worry that the larger picture of building an EFI for a 4 banger aircooled motor is getting lost in the weeds here.

These aren't exactly smooth running motors to begin with, and the engine bay is somewhat open to the elements. I'd reconsider using the Microquirt and adding some simplicity back into your equation. The operational differences between wasted spark and individually controlled plugs isn't going to be noticeable when you are driving the car, but will make it more challenging to build and tune - increasing the time before you are back to driving the car again.

Zach


I already tried to sell him on the easy button (well slightly easier button anyways). biggrin.gif

As someone who enjoys this stuff though I full support the effort to go completely overkill!

beerchug.gif


Enabler! laugh.gif

QUOTE
I'm actually sort of curious about what will be needed to properly dial in the injector timing and what sort(if any) difference it will wind up making. The tuning process with batch injection is fairly straight forward for the home mechanic with a wideband sensor but to fully dial in sequential injector timing... I suspect that's going to take some serious time and heavy equipment.

Lots of stacked variables to deal with.

You and me both. This is for sure going to be the biggest challenge. I may eat my words, but tuning for sequential spark doesn't look much different from wasted spark other than throwing some extra $ at hardware and a few configuration settings to tell the ECU the timing rules get applied across 4 channels instead of two. I'm sure there will be some head scratching in there but doesn't seem quite as daunting as sequential injection. Worst case if I get in too far over my head is that batch injection will still be an option.
bbrock
QUOTE(JamesM @ Oct 12 2021, 12:33 PM) *

Im just going to leave this here to give you more to consider about your injector selection. poke.gif

This is the datasheet I got with those A280H injectors

The 26lb rating was at 3bar pressure however they provide the flow specs and voltage corrections for running them all the way up to 7 bar!!!

I would imagine the 36lb sets come with similar abilities.

Have read that higher pressures improve atomization, and if you wanted to look at shortening your injector pulse even further, its something to consider.

Click to view attachment


I was actually wondering about that. What's a safe pressure a Bosch 69133 could run? Need to find specs.
KELTY360
Obviously, while the snow is on the ground you need to build a dyno in your garage.

You'll never be able to test your results til you do and you'll never be satisfied til you can test the results of all variables. poke.gif
Superhawk996
QUOTE(bbrock @ Oct 12 2021, 01:10 PM) *

Here's the view from my window today so I'm already done driving until the middle of June '22.


chair.gif Don't be such a cynic. With all the climate change it'll be 60F by late next week and then you'll be back to mud season!
bbrock
QUOTE(KELTY360 @ Oct 12 2021, 01:36 PM) *

Obviously, while the snow is on the ground you need to build a dyno in your garage.

You'll never be able to test your results til you do and you'll never be satisfied til you can test the results of all variables. poke.gif


Ha! You talk like you have met me... oh wait... you HAVE biggrin.gif

The testing will be a bit of a bummer. If I was a good person, I'd tune this thing with wasted spark and batch injection, drive it long enough to collect some data, and THEN switch to sequential. That seems too much like work though.

BTW, I assume MS can calculate and log fuel consumption as mpg - correct?
KELTY360
QUOTE(bbrock @ Oct 12 2021, 12:57 PM) *

QUOTE(KELTY360 @ Oct 12 2021, 01:36 PM) *

Obviously, while the snow is on the ground you need to build a dyno in your garage.

You'll never be able to test your results til you do and you'll never be satisfied til you can test the results of all variables. poke.gif


Ha! You talk like you have met me... oh wait... you HAVE biggrin.gif

The testing will be a bit of a bummer. If I was a good person, I'd tune this thing with wasted spark and batch injection, drive it long enough to collect some data, and THEN switch to sequential. That seems too much like work though.

BTW, I assume MS can calculate and log fuel consumption as mpg - correct?


You're asking me? confused24.gif I still think a log is something you burn in the fireplace.
bbrock
QUOTE(Superhawk996 @ Oct 12 2021, 01:54 PM) *

chair.gif Don't be such a cynic. With all the climate change it'll be 60F by late next week and then you'll be back to mud season!

It's actually supposed to be 60 this weekend. Whose the cynic now? confused24.gif

QUOTE(KELTY360 @ Oct 12 2021, 01:36 PM) *

You're asking me? confused24.gif I still think a log is something you burn in the fireplace.


That question was for the collective. I expected nothing more from you biggrin.gif poke.gif

BTW... idea.gif


https://www.youtube.com/watch?v=XUX_YWsECWs
Superhawk996
QUOTE(bbrock @ Oct 12 2021, 04:31 PM) *



BTW... idea.gif



see PM
ClayPerrine
If you want distributorless ignition, there is an easier way.

GM built DIS (waste spark) ignition systems on a LOT of their cars in the 80s/90s. Some of the 4 cylinder models used an external sender for the trigger. Years ago, I fitted waste spark to the Type -IV engine in my 914 using the GM system. I took the flywheel off the motor, and machined in the correct grooves to trigger the sender, and mounted the sender in the hole in the back of the case used for flex plate bolts in VWs. I made the mount eccentric so I could fine tune the timing.

This has to be machined into the engine side of the flywheel, properly indexed to the hole.

Click to view attachment

The ignition module has a base timing curve that it will follow if there is no ECU input. The curve is really close to the stock 914 advance curve. I had carbs on the engine at the time, and it ran way better with the distributorless ignition than it did with the stock distributor.

There is lots of documentation out there for connecting megasquirt to GM waste spark ignition.

The only reason I pulled it off the car was that I had a starter eat the ring gear on the flywheel. I was in a hurry, and had to get the car running to get to work. So I put the stock distributor back in. I still have all the parts in my attic.

jd74914
QUOTE(bbrock @ Oct 12 2021, 03:31 PM) *

That question was for the collective. I expected nothing more from you biggrin.gif poke.gif

BTW... idea.gif

Well...since you asked...

You can get a wireless strain gauge amplifier; strain gauge up your axles, and have live torque measurement...all of the time!

http://www.izzeracing.com/ewExternalFiles/...2_Datasheet.pdf

Same thing could be done with homemade hardware or from other vendors...
Frank S
Had a chance to check some of my data.
With the 21lb/hr injectors I run with 5ms PW at idle and down to 2.5ms in some areas in overrun, which is the smalest PW I have.
Linear range of my injectors ends at 1.5ms and if you add deadtime this is getting close to the none linear region.
This is meassured data, not calculated...
Frank S
And here how it looks like if you integrate the MS3 into the D-Jet housing:

Click to view attachment

Just the Injector driver board for Full Sequential is not in jet...
bbrock
QUOTE(Frank S @ Oct 14 2021, 10:03 AM) *

And here how it looks like if you integrate the MS3 into the D-Jet housing:

Just the Injector driver board for Full Sequential is not in jet...


I like it! Is that a bluetooth module at the bottom? And are you running inputs/outputs to the original ECU harness header?

Thanks for posting those pulse widths. That is VERY helpful.
bbrock
QUOTE(Frank S @ Oct 14 2021, 09:58 AM) *

Had a chance to check some of my data.
With the 21lb/hr injectors I run with 5ms PW at idle and down to 2.5ms in some areas in overrun, which is the smalest PW I have.
Linear range of my injectors ends at 1.5ms and if you add deadtime this is getting close to the none linear region.
This is meassured data, not calculated...


I plugged 21 lbs/hr into my spreadsheet and played with the VE at idle until I got glose to 5ms PW. It hit at about 29%. I used 25% as a lower VE for a margin of safety and recalculated the VE curve. With that adjustment, calculated PW for 36 lb injectors at 850 rpm is 2.5ms (3.5ms for 26 lb.).
Frank S
QUOTE(bbrock @ Oct 15 2021, 01:36 AM) *

QUOTE(Frank S @ Oct 14 2021, 10:03 AM) *

And here how it looks like if you integrate the MS3 into the D-Jet housing:

Just the Injector driver board for Full Sequential is not in jet...


I like it! Is that a bluetooth module at the bottom? And are you running inputs/outputs to the original ECU harness header?

Thanks for posting those pulse widths. That is VERY helpful.


Yes, this is a bluetooth module at the bottom and the real time clock module at the right and yes I'm running inputs and outputs through the original ECU harness header but for the actual install I had to feed also some extra cables as I have added a second EGO sensor and an external knock module.
JamesM
QUOTE(bbrock @ Oct 14 2021, 03:49 PM) *

QUOTE(Frank S @ Oct 14 2021, 09:58 AM) *

Had a chance to check some of my data.
With the 21lb/hr injectors I run with 5ms PW at idle and down to 2.5ms in some areas in overrun, which is the smalest PW I have.
Linear range of my injectors ends at 1.5ms and if you add deadtime this is getting close to the none linear region.
This is meassured data, not calculated...


I plugged 21 lbs/hr into my spreadsheet and played with the VE at idle until I got glose to 5ms PW. It hit at about 29%. I used 25% as a lower VE for a margin of safety and recalculated the VE curve. With that adjustment, calculated PW for 36 lb injectors at 850 rpm is 2.5ms (3.5ms for 26 lb.).



Well your math is good as that is roughly what i was seeing when running the stock 36lb injectors.

Now stop playing with spreadsheets and build the damn thing poke.gif
bbrock
QUOTE(JamesM @ Oct 15 2021, 10:15 AM) *

QUOTE(bbrock @ Oct 14 2021, 03:49 PM) *

QUOTE(Frank S @ Oct 14 2021, 09:58 AM) *

Had a chance to check some of my data.
With the 21lb/hr injectors I run with 5ms PW at idle and down to 2.5ms in some areas in overrun, which is the smalest PW I have.
Linear range of my injectors ends at 1.5ms and if you add deadtime this is getting close to the none linear region.
This is meassured data, not calculated...


I plugged 21 lbs/hr into my spreadsheet and played with the VE at idle until I got glose to 5ms PW. It hit at about 29%. I used 25% as a lower VE for a margin of safety and recalculated the VE curve. With that adjustment, calculated PW for 36 lb injectors at 850 rpm is 2.5ms (3.5ms for 26 lb.).



Well your math is good as that is roughly what i was seeing when running the stock 36lb injectors.


Oh wow! There are several school teachers who would likely keel over in shock to here that if they are still alive.

It's a great relief to have real measurements that agree with my chicken scratch! Thank you.

QUOTE

Now stop playing with spreadsheets and build the damn thing poke.gif


Shit. I thought this was just a mental exercise. You mean I actually have to build this? biggrin.gif

Got a few smaller projects I need to finish before I can dedicate serious time, but I'm ready to pull the trigger on ordering the MS3 kit. Think I finally figured out how the daughter boards figure in. If I've got it right, the v3 board is just a base "non denominational" board and the daughter board determines which version of MS. That was a little muddy until I looked at the assembly instructions. Think I'm going to go with MS3Extra as it will be easier for only about an extra $50 overall.

I sent Mario an email to get an idea of when his goodies will be available for purchase again, and also when the v2 of his CAM SYNC might be ready. That would also greatly simplify things. He has this note on the description: "It will incorporate a small trigger wheel in stead of the flying magnet. For basic applications this can be used without a crank trigger. In performance applications both are still recommended."

Still not sure how much precision is being sacrificed using a gear driven distributor to send tach signal vs. a crank sensor.
Frank S
QUOTE(bbrock @ Oct 16 2021, 02:21 AM) *

Still not sure how much precision is being sacrificed using a gear driven distributor to send tach signal vs. a crank sensor.


In theorie you will sacrifice some precision, in paractice I don't experience a difference.
bbrock
I've been wondering if there is an advantage in adding a CHT ring sensor on #3 spark plug to the system beyond just generally being able to track head temps. Since we have to run these engines a bit rich to keep the head temps down, it seems like the most efficiency at cruising could be had with a feedback loop between AFR and CHT so the mixture could go as close to stoich as possible while maintaining safe head temps. The TunerStudio manual talks about CLT to monitor coolant temp for warm up (I'll be mounting mine in the stock CHT location), but I'm not seeing anything about integrating CHT with the AFR targets or VE table. Would this be a bad idea?
JamesM
QUOTE(bbrock @ Oct 15 2021, 04:21 PM) *

Still not sure how much precision is being sacrificed using a gear driven distributor to send tach signal vs. a crank sensor.


This is entirely observational but "some"

I went from picking up my tach signal from a pertronix on a locked dizzy to Marios 36 tooth crank wheel and the difference under a timing light was visibly noticeable. couldn't tell you in a number what the difference was and im sure its very minor, like less than 1 deg, but with the dizzy trigger there was a bit of jitter under the timing light, the timing mark had a slight haze to it. After moving to the 36 tooth crank wheel the mark under a timing light was rock solid, like i was looking at a laser pointer.
JamesM
QUOTE(bbrock @ Oct 16 2021, 08:42 PM) *

I've been wondering if there is an advantage in adding a CHT ring sensor on #3 spark plug to the system beyond just generally being able to track head temps. Since we have to run these engines a bit rich to keep the head temps down, it seems like the most efficiency at cruising could be had with a feedback loop between AFR and CHT so the mixture could go as close to stoich as possible while maintaining safe head temps. The TunerStudio manual talks about CLT to monitor coolant temp for warm up (I'll be mounting mine in the stock CHT location), but I'm not seeing anything about integrating CHT with the AFR targets or VE table. Would this be a bad idea?



I mean it would be cool if you could figure out a way to do it, but AFAIK its not an existing feature.

The other issue you need to look at is the spark plug ring sensors are are thermocouples where as Megasquirt uses thermistors for temp sensors.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2024 Invision Power Services, Inc.