Time to do some tuning

Message
Author
rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#41 Post by rodstares » Wed Jun 03, 2020 9:17 pm

Alright, we may be getting somewhere! I had originally planned to buy and fit a lamda probe and can still do so but I'll try this manual mapping plan first. I'll get back to you with the results a bit later. Once again I have to thank you for your time and effort helping with this thing, if I ever get to Yorkshire I owe you a pint.

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#42 Post by rodstares » Wed Jun 03, 2020 11:17 pm

Hi Simon,
I have sent screenshots to your email at "cheapmotoring"

Cheers,
Rod

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#43 Post by LPGC » Thu Jun 04, 2020 6:39 pm

Thanks Rod, I've received your email with screenshots.

I meant to say you could post screenshots on forum, save your map to your computer, email me the map for me to modify and I could email you the modified map back for you to upload to your ECU.. I should have been clearer but it's no problem, I could still make a modified map for you to upload from scratch (based on your screenshots) and email it to you.

To save / load a copy of your LPG map to your computer just click on save configuration / load configuration. On my Windows10 setup the Emer software saves map files to the folder ThisPC>Documents>Multipointinj>ConfigLPG. Emer map files have the extension .ec8. When I've emailed you the map you'll have to move or copy it to the folder to be able to load it in the Emer software.

However there is still a problem - I expected that you'd be able to reduce the figure in your calibration table by a lot more without seeing manifold pressure rise to the extent it did. More on this below but I haven't emailed a map file yet because I think there's more to ask/discuss first...

To keep other readers up to date, Rod sent me screenshots showing the engine running with various 'flat map' figures (all numbers in the map table the same), I've listed the figures below, the first row of figures is running on petrol hence some figures are irrelevant.

map figure / manifold pressure reading / gas pressure / pinj / ginj
PET / 0.32 / PET / 3.49 / PET
255 / 0.34 / 1.16 / 3.70 / 8.57
229 / 0.35 / 1.16 / 4.02 / 8.55
206 / 0.42 / 1.20 / 4.90 / 9.04
185 / 0.47 / 1.13 / 5.27 / 9.02

Since (in the above figures) calibration figures are maxed out where we see minimum manifold pressure (map) idling on LPG we don't know if manifold pressure would continue to fall to be the same as or lower than it is on petrol (same or lower than 0.32) if the LPG calibration figure could be increased further than 255.

As said in a previous post I'm surprised manifold pressure (map) on this engine at idle is only 0.32 bar, I'd expect it to be a bit higher... The figures pose some questions, such as...
1.Is this engine drawing more vacuum at idle with correct mixture than I'd expect (which would mean we should aim for the same idle map on LPG, which also implies the LPG injectors flow much worse than I'd expect and decent results would be very unlikely).
2. Is the engine idling rich on petrol (which would mean we should aim for a slightly higher idle map on LPG, say 0.35/0.37 bar, which still implies LPG injectors flow poorly but at least we'd have a bit more chance of workable results).
3. Are the figures skewed by one or more poorly performing injectors (cylinders running a bit rich lower map, cylinders running lean or way too rich raise map).. We can check for poorly performing injectors by switching individual some cylinders to run on LPG and some cylinders to run on petrol in the 'diagnosis' screen (more on this below).
4. Are the figures skewed by a patching/routing issue (gas injector pulse to each engine cylinder should be based on the petrol injector pulse to each respective engine cylinder, e.g cyl 1 petrol injector pulse should control cylinder 1 gas injector pulse). This can be confirmed in the 'diagnosis' screen at the same time as (3).

To check above points 3 and 4, leave figures in the calibration table set at 255 then in the 'diagnosis' screen under 'gas injectors cut out' you can switch individual cylinders to turn off the gas injection on that cylinder and turn petrol injection on that cylinder back on. In turn (one cylinder at a time) click 'off' beside each cylinder and note changes to the manifold vacuum (map) reading. Do one cylinder at a time and switch it back to LPG (click on) before moving onto the next cylinder. Does switching any cylinder(s) back to petrol cause a misfire? Does switching and cylinder(s) back to petrol cause a different change to manifold pressure reading than switching other cylinder(s) back to petrol?

Does the exhaust smell like it's idling rich when it's on petrol or on LPG?

Does the engine seem to idle smoothly on petrol and on LPG? As smoothly on LPG as on petrol?

With the engine idling if you do a 'tip in' (quickly blip the throttle to around halfway) does the engine seem as crisp to accelerate on LPG as it does on petrol?

You mentioned you intended on fitting a lambda sensor, that would be a great move at this point.

I suspect it may be necessary (or a very good move) to change injectors to a higher flowing type but we're not at the point of making that call yet. If you can do the tests and answer the questions above it'll help provide more info to judge that. If above tests point to injectors etc being OK it'll confirm what we already thought about the reducer being under-spec for the engine... If you're happy to be able to just run on LPG during cruising type conditions you wouldn't have to change it but if you want the ability to run on LPG when you put your foot down a bit you'd probably need to upgrade it.

Simon
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#44 Post by rodstares » Thu Jun 04, 2020 11:58 pm

Hi Simon,
I ran the tests you describe above this morning. Warmed the engine up and had MAP at .33/.34 bar at idle on petrol. I switched to LPG then switched each cylinder to petrol and back to LPG checking the MAP at each change. I found no variation in MAP pressure, it remained at .33/.34 bar each time. There was no misfiring or rough running. The engine runs more smoothly at idle on petrol than gas but saying that it is not offensively rough on gas. The exhaust doesn't seem overly rich smelling to me. When I give it a kick in the guts (tip in) it revs up fine on petrol or gas, you might say better on petrol but it's not bad on gas, nothing that would cause me any concern, seems responsive enough.
If all this indicates the injectors are ok I'd probably go for petrol addition at high RPM until I can get around to installing a lamda probe and higher capacity reducer. I am off to work again Monday so it will be a couple of weeks before i get back to the old girl. It's also our Queens Birthday long weekend here now and I'll be away for the next couple of days, trout season closes this weekend and snow season starts!
Thanks for your perseverance, things are much harder to diagnose without the vehicle right in front of you...

Regards,
Rod

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#45 Post by LPGC » Fri Jun 05, 2020 1:53 pm

Hi Rod,
I emailed you 3 modified maps.
(A) R12C12M255fallingslowlyfrom6ms.ec8
(B) R17C17M191fallingslowlyfrom6ms.ec8
(C) R17C17M191fallingslowlyfrom6PA10msfrom10ms2500rpm.ec8

None of the 3 maps I sent have the usual 'shape' I'd expect because the usual shape only really applies to the usual range of ginj's but your range of ginj's begins at a ginj that would usually equate to medium engine load... Usually with fairly linear injectors maps start with a low/medium multiplier, the multiplier peaks at around 4ms, falls away quite quickly toward 6ms then falls at close to the same rate towards and beyond the peak length of petrol injector duration (V34's a bit different because I expect the peak to extent a bit beyond the usual 4ms to maybe 6ms). The reason for the usual map shape is that at short ginj's (gas injector pulse lengths) gas injectors usually over-flow (so multiplier needs to be lower to compensate below around 4ms), just as the gas injector is becoming linear at around 4ms there's the full effect of gas injector delay (slower opening compared to petrol injectors) combined with the still short ginj means the multiplier has to be comparatively high. As ginj increases further the same delay is still present but becomes a decreasing percentage of ginj. It's a bit more complicated than that really because the pressures involved etc also play a part. But in the case of your setup even at idle ginj is already high enough to be past the initial over-flow response range of the gas injectors, so there's no 'lead in' area to the maps I sent you (the multiplier doesn't fall away below 4ms).

(A) Is your map with map table numbers gradually falling from 6ms.
(B) Is to be used if you can increase reduce pressure to 1.7 bar. The map table numbers in this map are lower than the table numbers in (A) because I've adjusted the figures to hopefully account for the higher working pressure.
(C) Is the same as (B) but with 10ms of petrol addition anytime pinj is above 10ms and rpm is above 2500. When the 10ms of PA is active the petrol injectors will be pulsed for 10ms. If for example pinj was 12ms the 10ms of PA will be subtracted from the 12ms leaving a remainder of 2ms pinj, ginj will then be based on that remaining 2ms pinj, according to the modified map table the 2ms pinj would become 1.91 * 2ms so petrol injectors would pulsed for 10ms and gas injectors would pulsed for 3.82ms (both at the same time). Another example pinj 16ms, petrol injectors would pulse for 10ms, gas injectors would pulse for 1.76 * 10ms = 17.6ms.

Try (A) first.
Before trying (B) increase pressure at the reducer (use the reducer's pressure adjustment screw) while still using map (A).
Then try (B). If this works OK it will still switch back to petrol when you put your foot down.
If (B) works OK try (C).

Others may like to see the maps etc but I haven't time to post them at the moment.

Simon
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#46 Post by rodstares » Fri Jun 05, 2020 11:23 pm

Thanks Simon, that's awesome, I can't wait to get home and give these a try. Increasing the gas pressure is a pain but I can do it. There is no adjustment screw on this reducer so I have to shim it as per earlier advice from you. I may not get to it for a couple of weeks, I'm about to head off to work again. I'll update as soon as I can.

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#47 Post by rodstares » Mon Jun 08, 2020 12:46 am

Morning Simon,
I've had a bit of a reprieve on the work front so have a couple more days to play around with this thing. This morning I loaded map A and the results are pretty bloody good. I could only get it to switch back to petrol on the very steepest of hills at the hardest acceleration, far better than it's been. Its idling nicely and very responsive to throttle input. In short, running really quite well. From a standing start on the flat it goes all that way round to 5000 rpm, flat to the boards no issues. Its going up any hill without a problem unless it is very steep and I plant the foot from about 1500 rpm, when it hits about 3500 rpm it will switch back to petrol but I am talking wide open throttle second gear and pulling very hard.
So the question is, do I buggerise around shimming the reducer to increase the gas pressure for what may be a marginal improvement or just dial in the petrol based on map A?

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#48 Post by LPGC » Mon Jun 08, 2020 2:38 pm

Hi Rod,

I'd forgotten it was you who I advised to shim the reducer. Since you're getting decent results it's probably better not to try adding any more shims to increase pressure further, I remember last time shims were added there was a fine line between 1.2 and 3 bar... Just leave the reducer as it is.

We still don't know if mixture on LPG is correct and it would be best to check that using a lambda sensor but without a lambda sensor there are still some checks you can do... If the map table I sent causes lean running at any engine load (manifold pressure and rpm combination) the engine will produce a bit less power compared to running on petrol at that load point. To test you could find a bit of a hill to drive up at part throttle, while driving on LPG hold a constant throttle and switch back to petrol, without moving your foot on the throttle does engine power seem to increase? if it's an automatic gearbox does the box change up gear at the same moment on LPG as it does on petrol? Run the same checks with the ball on the map table at various load points.

If at any point during driving on LPG the yellow light on the switch flashes it's because the system has seen full duty cycle of LPG injectors so switched momentarily back to petrol, you'd also notice this in software (ginj figures go to zero and reducer pressure comes back up to around what it is at idle so around 1.2 bar with your setup).

Simon
Last edited by LPGC on Tue Jun 09, 2020 9:13 am, edited 1 time in total.
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#49 Post by rodstares » Tue Jun 09, 2020 3:48 am

Right, I can't fault it, really no noticeable difference between petrol and lpg. I am going to order the parts and fit a lamda probe. Narrowband you say? Was thinking I'd get 4 wire so it's the heated type and do a clamp mount as I'll soon be replacing the exhaust, do a weld mount to the new exhaust.

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#50 Post by rodstares » Tue Jun 09, 2020 5:49 am

For now I have map A running and have set petrol addition to 10ms petrol addition above 10ms pinj and 3000rpm. Just took it for a run up the hills and it's going well, no alarms, plenty of grunt. That'll do for now until my lamda stuff arrives.

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#51 Post by LPGC » Tue Jun 09, 2020 9:47 am

Good! Pleased to read it's getting there.

A narrow band probe probably makes most sense, you couldn't use a wide band probe anyway unless you also buy a wide band probe controller. Because wide band's need a controller it will be far cheaper to use a narrow band probe and you can wire the narrow band to the Emer ECU so you can see lambda voltage on screen if/when you adjust calibration.

Yes, a 4 wire (white and white non-polarised heater wires, grey signal earth, black signal) 0>1v probe is a good bet, you can wire the probe's heater circuit via a fuse through a relay that's switched with the ignition.

Connect the signal earth to either the battery negative pole or to one of the LPG system's temperature sensor earth wires (grey probe wire to an LPG system narrow black wire that runs in a sheath with orange or orange with black stripe). It can be better to connect to a narrow LPG system earth wire rather than the battery earth wire because that way means the lambda probe earth reference voltage and LPG system earth reference voltage are the same so you don't read a biased voltage on screen. Narrow LPG earth wire rather than wider (solenoids etc) LPG earth wire because solenoid etc earth wires have their own earth bias (earth voltage pulled up due to solenoid load). Or if you don't connect to an LPG system earth at least use battery earth rather than engine earth, when the alternator is charging heavily engine earth will be at a higher voltage than battery earth.

Simon
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#52 Post by rodstares » Wed Jun 10, 2020 8:40 am

I got my lamda probe and installed it today. Tomorrow the wiring and calibrating. Do I just get set it up and hit autocalibrate?

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#53 Post by LPGC » Wed Jun 10, 2020 10:15 am

No, you should never use autocal on this vehicle, autocal only works when the vehicle's fuel system is closed loop.

Fitting the lambda probe won't change the vehicle's fuel system from open loop to closed loop, it will only allow you to see (for sure) if the mixture is lean/rich. It's a far more accurate way of checking mixture than the methods we've used so far.

If the mixture is rich at any point on the calibration table you'd then lower the number at that point in the table and vice/versa.

The lambda sensor is narrow band, so it will give a 0v reading when mixture is just a little lean or extremely lean and a 1v reading when the mixture is just a little rich or extremely rich. A closed loop fuel system (which again your vehicle doesn't feature) would keep adjusting the mixture by only a very small amount to keep the lambda reading 'flicking' between 0v and 1v but on your open loop system you're unlikely to see much flicking and there'll be a fine line (when you adjust the numbers in the table) between getting a 0v reading and a 1v reading but you should err on the side of rich (1v) rather than on the side of lean (0v), especially at higher engine loads.

Simon
Last edited by LPGC on Wed Jun 10, 2020 8:54 pm, edited 1 time in total.
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#54 Post by rodstares » Wed Jun 10, 2020 11:18 am

Ah ha, that makes sense

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#55 Post by rodstares » Thu Jun 11, 2020 3:01 am

OK, I have the lambda up and running, sits at about 0.80 volts idling on petrol then goes to between 0.02 and 0.04 idling on gas. If I gently increase the rpm, keeping the green ball in the same ms range it doesn't change, if I give it a quick tip in it goes up to around 0.80 volts when the ball drops. This would indicate lean running when not under load but richer when load is increased?
Lambda petrol at idle.PNG
Lambda petrol at idle.PNG (138.79 KiB) Viewed 1275 times
Lambda gas at idle.PNG
Lambda gas at idle.PNG (133.85 KiB) Viewed 1275 times
Lambda gas at 2800 rpm.PNG
Lambda gas at 2800 rpm.PNG (137.24 KiB) Viewed 1275 times

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#56 Post by rodstares » Thu Jun 11, 2020 3:27 am

This below is a picture with the engine under load using the foot on brake while applying throttle technique. This shows the mixture becoming richer while under load?
Lambda Gas under load.png
Lambda Gas under load.png (163.85 KiB) Viewed 1275 times

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#57 Post by LPGC » Thu Jun 11, 2020 12:07 pm

"OK, I have the lambda up and running, sits at about 0.80 volts idling on petrol then goes to between 0.02 and 0.04 idling on gas. If I gently increase the rpm, keeping the green ball in the same ms range it doesn't change, if I give it a quick tip in it goes up to around 0.80 volts when the ball drops. This would indicate lean running when not under load but richer when load is increased?"

Hi Rod,
More or less correct, tip-in is a slightly different situation but looking at your screen shots it's lean on LPG at low loads as expected and mixture is correct (just slightly rich) at 9.57ms 1663rpm as hoped for and expected. Not bad for guestimating a calibration map at the other side of the world based on just anticipated injector response and manifold pressure info ;-)

We should check your lambda readings are correct (that there isn't a bias voltage). There are various ways of doing this, probably the easiest way is if you temporarily change reference (software) pressure from 1.2 bar to 1.6 bar... The Emer ECU will then see you have an under-pressure situation (reference pressure at 1.6bar, actual pressure 1.2 bar) and increase ginj (gas injector pulse duration) by around 33%. If the lambda sensor voltage isn't biased I expect this will see it reading near 0.95v for pinj of 9.57ms. If you do see near 0.95v in this test move on to the next steps below. If you don't see near 0.95v I'll need to know what voltage you do see before advising a probably different next set of steps for you to follow.

Your map is maxed out (255 figures) at the low load range but it's still lean at low loads.
We usually expect to see the numbers in the calibration table increase for increasing rpm (on any row, numbers increase as we move to the right across columns. As we go across the table from left to right the numbers shouldn't change as much as they do going from top to bottom but there should still be a change. We should aim conservatively for a 5% increase in the figures in the table from left to right between 500rpm and 5000rpm... but a 5% increase in your map from 500rpm to 5000rpm would mean even if mixture was correct with 255 figure in a 500 rpm we'd expect to need 268 in the 5000rpm column and we can only enter a maximum 255 figure.
We can use lower numbers in the map if we increase the reference pressure (again like we did in the above paragraph when checking lambda response) because (again) the ECU will see the under pressure situation and add a further percentage to fuelling to compensate for the under pressure situation....

I was going to post instructions here on how to modify your map to compensate for the higher reference pressure and dial in some rpm compensation but it's easier just to modify the map myself and email it to you, so that's what I've done.

We'll be lucky if the lambda readings are the same now that I've made the modifications because the pressure compensation isn't spot on but this new settings file should be a better starting point for your further tuning, it allows headroom to increase fuelling so should allow you to make mixture richer at low loads. The rpm compensation I've set is only a guestimation but I wouldn't bother at this point trying to adjust calibration further for different rpm's, instead concentrate on adjusting the map for load (adjust a full row at a time, don't try to adjust columns or individual boxes). If you hold the ball steady on a row (keeping the throttle and rpm steady), if lambda reading is 0v increase figures on the whole row by (say) 5%, or if the lambda reading is very near 1v decrease all the figures on that row by 5%. Aim for a reading of 0.5v, as you get closer to correct mixture you might want to adjust each row by less than 5% at a time. The exception is for higher load conditions (manifold pressure above around 0.8 bar) where you want the lambda reading to be closer to 0.9v

Post a pic of your finished map? You'll probably find that a setting you make under one set of conditions (weather, engine temp, etc) will seem to give different results (lambda) under different conditions, that's kind of to be expected for an open loop system and is one of the reasons it's better to err on the side of slightly rich than on the side of lean. A very slightly lean mixture can give best economy and lowest HC and CO emissions but sometimes at the expense of increased Nox emissions. A bit leaner than that and economy drops drastically and there's the risk of accelerated engine valve wear etc. Correct mixture lambda of 0.5 is the best all round compromise. A slightly rich mixture won't be much different from a slightly lean or correct mixture in terms of economy but HC and CO will start to rise a bit. With a very rich mixture economy will fall again. Between very rich and slightly rich is where you want to be for high load (high manifold pressures) because it makes the most power at the expense of wasting a bit of fuel but this is the best compromise for high loads. Remember that the lambda probe has to be hot to work properly, the engine will have to have been running a minute or so before it will give a proper reading.

Simon
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#58 Post by rodstares » Thu Jun 11, 2020 11:48 pm

"Not bad for guestimating a calibration map at the other side of the world based on just anticipated injector response and manifold pressure info ;-)"....indeed this is quite amazing, what a world we live in. As a bit of an old sci fi nut, the time lag in comms is very reminiscent of some books I've read. Extremely remote diagnostics.
So here are the lambda test results.
@1.2 bar 8.6ms .04v
@1.6 bar 9.5ms .84v
This is on gas at idle, call it 800rpm
A 33% increase in ginj would be 11.43ms? I'm really not sure what any of this means apart from we get a response from the lambda probe. I am going to load the new map and see what I see.

rodstares
New member
Posts: 118
Joined: Mon Nov 11, 2019 6:16 am

Re: Time to do some tuning

#59 Post by rodstares » Fri Jun 12, 2020 12:32 am

I loaded up the new map and no matter what percentage I changed the rows by, I couldn't get the lambda to read over .04, I think I may be at the limit of my understanding. I put the engine under load at 500, 1000, 1500 and 2000 rpm and observed the lambda reading, it remained steady at .02 until I got to 2000 rpm when it went to .04v. I thought I'd start at 1000 rpm and holding it steady there in gear with my foot on the brake I increased the row settings by 5% increments. I didn't see any change in the lambda voltage, it stayed at .02 to .04.
I'd love to be able to slip this thing into your workshop for a day or so to sort it out properly but right now I may have to stick with it the way it is and be satisfied, it's a long drive to the UK from here. I have my 400km work trip coming up again shortly and it will be interesting to see how it goes just on the petrol addition from 3000 rpm map set up. If my mileage is ok and it runs pretty right, that might be it I think.

LPGC
Installer
Posts: 4072
Joined: Wed Sep 29, 2010 6:01 pm
Location: Yorkshire
Contact:

Re: Time to do some tuning

#60 Post by LPGC » Fri Jun 12, 2020 3:19 pm

To understand some of the recent things discussed you need to know how/why the ECU displays pressures and compensates for pressures...

The reducer has a vacuum connection to the manifold and is designed to supply gas at a constant pressure relative to manifold pressure (gas pressure output rises and falls along with manifold vacuum).

With this type of LPG ECU the software displays absolute manifold pressure (0bar reading would be absolute vacuum, 1bar reading would be atmospheric pressure, if this were a turbo'd engine and there were 1bar of boost above atmospheric pressure the manifold pressure reading would show 2bar).. But it doesn't display absolute gas pressure, instead it shows gas pressure relative to manifold pressure. E.g. If an engine is idling with a screen map pressure reading of 0.3bar (0.7bar below atmospheric pressure) and a screen reducer pressure reading of 1.2bar, if we were to connect a mechanical pressure gauge to the gas pipes that supply the gas injectors it would only read 0.5bar because the 1.2 bar software screen reading is relative to the 0.3bar manifold pressure reading which is 0.7bar below atmospheric pressure. But if we then opened the throttle the manifold pressure would come up to atmospheric pressure, so the map reading would come up to 1bar, the reducer would supply gas at 1.2bar above atmospheric and the mechanical pressure gauge would read 1.2bar.

In simple terms the amount of gas a jet (injector nozzle) will flow depends on the size of the jet, how long the injector is pulsed for (ginj) and the difference in pressure between the gas supply at one side of the jet and the vacuum (or atmospheric pressure at full throttle) at the other side of the jet. It's the difference in pressures that's important, not the actual pressures. A gas injector pulsed for a given length of time (a certain length ginj) would deliver very close to the same amount of gas in any of the following situations (list below) because in all of the following situations the difference in gas pressure and manifold pressure is the same (1.2 bar)...

map relative to atmosphere (absolute map) / gas relative to atmosphere / absolute gas pressure / difference between gas pressure an manifold pressure
0.2 / 0.4 / 1.4 / 1.2
0.3 / 0.5 / 1.5 / 1.2
0.6 / 0.8 / 1.8 / 1.2
0.9 / 1.1 / 2.1 / 1.2
1.0 / 1.2 / 2.2 / 1.2
1.5 / 1.7 / 2.7 / 1.2 (this would be a turbo/supercharged engine running with 0.5 bar of boost).

So that's why the software shows gas pressure above manifold pressure and doesn't show gas pressure relative to atmospheric pressure.. We (and the software) don't really need to know absolute gas pressure or gas pressure relative to atmospheric pressure. Early pressure sensors for your type of ECU didn't even show manifold pressure, they just showed the difference in pressure between gas pressure and manifold pressure... But we've seen how knowing manifold pressure can be useful (when we made the map for your open loop system).

In simple terms none of the above figures in the above list are of much important except for the 1.2. If we pulsed a gas injector for (say) 10ms with that 1.2 bar difference pressure it will flow a very similar amount of gas regardless of whether manifold pressure is 0.2bar or 1.5bar. But the ECU does need to compensate fuelling if the difference in pressure changes from 1.2. For example if the ECU reference pressure is set to 1.2bar but the pressure reading is only 0.9bar we have an under-pressure situation, we only have 75% of the pressure the system expects (software reference pressure). In this situation the ECU would increase ginj by around 33% (1/0.75) from 10ms to 13.33ms so that the injector would flow the same amount of gas in 13.33ms with 0.9bar pressure as it would flow in 10ms at 1.2bar pressure. In practice reducers pressure relative to manifold pressure decreases when the reducer is working harder (flowing more gas), so the ECU will usually need to compensate for an under-pressure situation at high engine loads on a decent sized engine.

When I set reference pressure to 1.6bar in software I knew that actual pressure from the reducer (and the pressure reading) would stay at 1.2 bar and I knew that the ECU would compensate for what it saw as under-pressure situation by increasing ginj (by about 33% for the 33% under pressure situation). I used the pressure compensation to increase fuelling.

On most setups for a ginj of 10ms pinj would be around 9ms, in which case we'd have 111 in the row for 9ms. The number 111 in a box in the 9ms row would tell the ECU to multiply 9ms by 111% which equals 9.99ms (approx 10ms). If the ECU read an over pressure situation the 9.99/10ms ginj would be reduced by the ECU, if the ECU read an under pressure situation it would increase the 9.99/10ms. With your setup we've had to enter far higher figures than 111's in the table boxes because we needed a higher multiplier than 111 to provide rich enough fuelling. But instead of increasing the numbers in the table we could just increase the reference pressure so that (behind the scenes) the LPG ECU would (effectively) increase the numbers to compensate for an under-pressure situation. It's normally best practice to set reference pressure to equal actual pressure but in your case we couldn't get rich enough fuelling at idle because numbers in table boxes were maxed out at 255, so I've gone for a compromise which should allow mixture to be rich enough at idle with lower numbers in boxes than 255, the compromise involves setting reference pressure a bit higher than actual pressure.

The last map pic we've seen is the one you posted before I emailed the latest map for you to upload. Can you post pics of the map I last emailed and that map now you've increased figures (with the engine running at various loads)? The old map was capable of giving lambda reading of 0.76v at 9.57ms at 1663rpm with a multiplier of around 236% (the ECU interpolates between numbers in boxes), the new map uses reference pressure of 1.6bar instead of 1.2bar so all other things being the same it should provide around 33% richer mixture than the old map at 9.57ms at 1663rpm with the same 235/236 number in the table box... So if it now runs leaner with the same numbers in the map something else must have changed or there is some aspect we haven't considered (*more on this below). I did lower the numbers in the map table because I tried to account for the extra 33% fuelling that the under pressure situation should give but I did say that I didn't quite expect lambda readings to be exactly the same as they were before because the pressure compensation isn't quite correct... But if you increase the numbers in the map back to what they were before it should definitely be richer now than it was before.

* Aspects that might have changed or what we might not have considered include: If the engine is misfiring due to being over-rich it will cause the lambda to read lean mixture. Obviously if there's a problem with the lambda sensor itself or if it hasn't warmed up to operating temperature it won't give a proper reading. If the reducer and/or gas temperature readings were different (most likely if they were lower) when you did latest tests the ECU will compensate for lower temperature readings by decreasing ginj (at any given pressure cooler gas is more dense, the ECU not only compensates ginj for pressure it also compensates ginj for temperature readings).

Simon
Full time LPG installer
Servicing / Diagnostics / Repairs to all systems / DIY conversion kits supplied with thorough tech support
Mid Yorkshire
2 miles A1, 8 miles M62,
http://www.Lpgc.co.uk
Twitter https://twitter.com/AutogasSimon
07816237240

Post Reply