LM-1 WOT closed loop trick with freq. valve CIS
 

I had posted a little info about a CIS trick you can do with the LM-1 wideband controller from Innovate Motorsports in an EFI thread. Rather than derail that thread I wanted to provide a little info here for those that were interested.

First, a little background.

the K-basic CIS used in 78-79 US and all euro SCs (as well as all CIS 2.7s and the 73.5 T) is completely mechanical, bar some electrical heaters like for the warm up regulator.

The K-jetronic with lambda system on the 80-83 SC adds a "frequency valve" which can modify the amount of fuel delivered based on its input. The lambda "computer" takes as input from the 02 sensor a voltage reading, and modifies the frequency valve output to deliver more or less fuel.

The narrowband 02 sensors, as used in K-jetronic with lambda, have a very steep voltage curve around stoichiometric (14.7:1 air/fuel ratio) so that they essentially output 1 volt if the ratio is measured on one side of stoich, and 0 volts on the other side. This tells the lambda computer which direction to change its output for the frequency valve. A wide band 02 sensor (as used with the LM-1) has a more linear response to a wider range of A/F ratios.

This lambda add-on allows the car to run at 14.7:1 during cruise (this is called "closed loop" mode), which can be noticeably beneficial to fuel economy vs K-basic.

For power and engine safety reasons it is desirable to run richer than 14.7:1 at wide open throttle (WOT). K-jetronic (CIS) with lambda achieves this by ignoring the oxygen sensor, which has the effect of richening the mixture. The CIS uses a switch tripped at WOT to know when to ignore the O2 sensor. this is called "open loop"

The inventor of the LM-1 is named Klaus and he goes by klatinn on this board. He has a 80-83 US SC motor (or at least CIS) in his VW bus, which he uses for testing sometimes (all the time?).

He designed the LM-1 with some really neat features, including the ability to simulate a narrowband sensor on one of its outputs, which could be used in the case where the wideband sensor replaces the narrowband in the normal narrowband location, but you still need a narrowband output for the stock engine management.

Here's the trick: set up the CIS to run "closed loop" WOT, except at a max power air/fuel ratio (say 12.8:1) instead of 14.7:1. The way this is done is to program one output of the LM-1 for a normal narrowband simulated output (14.7:1), and the other output of the LM-1 to a narrowband simulated output centered around 12.8:1 instead. Then, use the WOT switch to activate a relay to change the O2 input to the lambda computer from the 14.7:1 narrowband output to the 12.8:1 output. This is all the WOT switch now does, and the WOT input on the lambda computer now never gets a message.

Klaus also mentioned somewhere that it is possible to modify the "reaction time" of the simulated narrowband output, which I think may even improve throttle response at WOT.

Thus, "closed loop" at WOT with just the right air/fuel ratio!

If anything is unclear, please ask. Here's a link to Pelican's page with the LM-1 for sale:

http://www.pelicanparts.com/cgi-bin/ksearch/PEL_search.cgi?command=show_part_page&please_wait= N&make=POR&model=911M&section=FULcis&page=15&bookm ark=57&part_number=INN-LM1-KT

Mwahaahaha! "Hacking" the CIS system!

Anybody know what the duty cycle for the Bosch MFI fuel cutoff solenoid is . . . Mwahahaha!

(did anything ever come of using a linear actuator to control the rack?)

John, I was really hoping that something would come of that, but never heard anything. Here's the original "Electronic MFI" thread for reference:

http://forums.pelicanparts.com/showthread.php?s=&threadid=175818&perpage=20&highl ight=mfi%20rack%20stepper&pagenumber=1

Andy, so was I, maybe in some spare time (in my spare life!)

Solenoids are pretty robust and fast-acting, however, and the control rack is spring-loaded. I bet if you applied a pulsed DC voltage to the Bosch solenoid it would act just like any other voice coil. You might even be able to play music on it (no joke, if you hook a meter movement to a headphone jack you'll get sound out of it.)

Anyway, when I read the manual for the LM-1 I was similarly intrigued-- why couldn't you simulate narrowband output centered around best power mixture and crank the rack all the way rich.

Power up LM-1, heat the sensor. Start engine, running rich, You give it throttle and raise the revs. Ordinarily, the space cam would meter the correct amount of fuel, but you've got the pump set so rich that the space cam's out of regulating range. By now the sensor is heated up and measuring AFR's in the 11:1 range. (VERY possible with an MFI pump). You then flip a switch and connect the simulated narrowband output around 12.8:1 to the solenoid through a relay. The solenoid triggers, pushing the rack until the fuel cuts off enough to drop the afr to 12.8:1, after which point the simulated narrowband is showing zero, so the solenoid de-energizes and releases the rack, which follows the current space cam postion and starts running rich again. Whack! The solenoid fires off again.

Ideally this would happen many times a second, which is why I wondered what the duty cycle of the solenoid is.

I remember following that thread with bated breath, hoping a product would emerge.

One of the original advantages of MFI was the high pressure nozzles giving superior atomization and thereby better vaporization. Having electronic control (with feedback) of the MFI pump satisfies many requirements. I see it as dramatically extending the useful life of MFI and improving its performance. There is no substitute for a closed loop – something MFI didn’t originally have.

The same is true with the CIS feedback that Andy has done.

I understand that today’s high frequency piezoelectric nozzles used in some EFI systems achieve even better atomization. Andy has researched this far more extensively than I.

Andy is on the right path. Whether it is MFI, CIS or other; current technology better closing the loop is of great benefit.

Best,
Grady

John,

Take a solenoid and drive it with a powerful audio amplifier designed for sub-audio. See what its response curve is. How it differs from an audio speaker is the mass of the armature.

Interesting experiment.

Best,
Grady

I should clarify that Klaus thought up this trick, not me.

"The narrowband 02 sensors, as used in K-jetronic with lambda, have a very steep voltage curve around stoichiometric (14.7:1 air/fuel ratio) so that they essentially output 1 volt if the ratio is measured on one side of stoich, and 0 volts on the other side. "

1. The NB O2 sensor never really reaches 1.0 volts.
2. The K-Lambda system already HAS a fuel enrichment mode at WOT & at cold running.
3. The K-Lambda system can be easily modified to offset the AFR to the desired
value other than 14.7 and still be closed-loop w/o complex circuitry, i.e.
even at the WOT.

Bottom line: The WOT AFR setup on the stock K-Lambda system is most
likely fairly close to the ideal of AFR=12.6 any way.

Hi KobaltBlau,

The trick does work indeed very well. I'm using it on my daughters CIS Lambda VW Golf and on my CIS Bus. I noticed, that the CIS on the Porsche CIS has a lean spike when opening the throttle fast. A 3rd gear WOT run shows AFR not steady, but varying by over 1 AFR over the RPM range (11.9 to 13.1). With the closed loop trick the lean spike is nearly gone, leading to better throttle response, and AFRs stay within a 0.5 AFR range.

On the VW Golf I used an LC-1 ($199.- with sensor) to make a smaller (and cheaper) package.

One of our engineers is a retired aerospace guy and races old Ferraries. He did the same on a twin CIS 12 Cyl. with 2 LC-1's and two custom programmed LMA-3 instead of the Lambda Computer. With this he as programmable duty cycle map (16 x 16) depending on MAP, IAT and AFR.

We are actually thinking of making it a product if there is enough interrest. Basically you can "map" the CIS and program fine-regulated target AFR numbers via the frequency valve. In addition you can also program the frequency of the output and the PID parameters for a possible frequency valve control.

As to the solenoid control:
You will need a linear acting solenoid. Some have more of a snap action. The frequency has to be high enough so the inertia of the solenoid and attached controlled mass does not vibrate. With duty cycle controlled current the rack-control should be possible.


Regards,
Klaus

Klaus,

If you can design a product to improve CIS by giving us better fuel economy for highway cruising and better throttle response in town, you'd have a winner. If the price is right you will sell a bunch.

Why don't you take a rough guess at a price for the product and see how much interest it generates?

Jeff

Hi,

Price-wise I am thinking in the $200-300 area without wideband. It might be possible to throw in EDIS spark control there too (distributorless). Wideband would be ~$200 extra.

Re: fuel economy. The same trick as with WOT closed loop is also possible for better economy in cruise and idle. From the research papers I read (and confirmed myself on the engine dyno with a VW engine) you get lowest BSFC (brake specific fuel consumption) at about 15.2 to 15.4 AFR. That's where you burn the least amount of fuel per hp produced. Smog controlled cars run at 14.7 because at stoich the least amount of pollutants are produced (CO, HC and NOx). You can reprogram the Narrowband simulation for the LM-1 or LC-1 to force the Lambda computer to run at lowest BSFC in idle and cruise. Running leaner than stoich also has the additional benefit that CHTs and EGTs are lower. At stoich they typically max out, CHTs normally just a tad below stoich. Ign. needs to be advanced a little for lean operation because a lean mix burns slower (burn speed maxes out at 12-13 AFR for most cyl. head types).

Regards,
Klaus

Klaus,

What sort of detonation/octane issues are there when running that lean (15.2-15.4:1)?

I’m greatly impressed you are transferring this to both dyno and road testing.

Best,
Grady

Re: LM-1 WOT closed loop trick with freq. valve CIS
 

Very cool. I ordered the LM-1 today and will be installing it in my '80SC.

Could you explain the relay wiring details? I'd like some additional info on attaching the 2 LM-1 outputs to a relay and the WOT switch.

Thanks for the help. I can't wait to get the LM-1 installed in my SC!!!

-Scott

Re: LM-1 WOT closed loop trick with freq. valve CIS
 

Klaus,

Could you explain the reaction time settings of the LM-1? I'm curious what settings would be best for throttle response in my '80SC.

Thanks!
-Scott
Soon to have a LM-1

:) Me again...

Do you have suggested ignition timing settings for 15.2 AFR for idle/part throttle and 12.8 AFR for WOT using 93 octane gas?

My '80SC has SSI's and an OE 2 in-1 out muffler.

Thanks for the help,
-Scott

Hi,

I advanced statically about 4 degrees for lean burn.
In that configuration I am running WOT a little richer (12.1 - 12.3) to compensate for the advanced ign.

Grady,
Detonation/Octane issues are non-existant on lean-burn operation. Actually if there were any, they are LESS when lean of stoich. At stoich the EGTs and CHTs max and therefore you have HIGHER knock probability. A leaner mix is harder to light, and therefore is less prone to self-ignition.

The problem when running lean is possible lean-surge. This comes from 2 facts:

1. Harder to light mixture. I am running a stock 6-wire CD and NGK plugs.

2. Uneven air or fuel distribution between cylinders. A lean cylinder might run too lean and gets a lean missfire. I matched all the injectors on the dyno engine.

Re: LM-1 speed setting
The LM-1 reacts as fast as the sensor can. The reaction speed is actually dictated by the gas-flow in the sensor, not by the controller's feedback loop as in other widebands. The LM-1 samples the exhaust gas and calculates Lambda at about 100 times/second. This is also reflected in the "Lambda bar" on the LM-1. For display on the LM-1 display it's averaged over 1/3 second. Datalogging is averaged over 1/12th second. The analog outs in fastest speed get also updated about 100 times/second.
With the CIS Lambda you can run fastest speed as it does internal filtering anyway to about 5-7Hz as I could tell.


Regards,
Klaus

This would be Awesome! I realize now that this should be possible with the frequency valve CIS, but it never occurred to me.

Between the large VW CIS market, the single CIS V8 ferraris, dual CIS V12 ferraris, CIS mercedes, and even dual CIS countaches, there might be a substantial market.

I assume if someone had K-Basic CIS they would need to switch to lambda controlled, what about KE-jetronic systems?

Cheers,

Hi Andy,

AFAIK, the KE-Jet is similarly controlled.
I have not experimented with a K-Basic, but it should be possible to bypass the WUR with a frequency valve (or even a just a high-flow injector used as frequency valve) to control enrichment. To allow lean control authority in this case requires that the basic control pressure is slightly raised. With some experimentation it should actually be possible to eliminate the WUR completely.

EDIT: BTW, the lean burn trick also works on 3.2 Motronic cars.

Regards,
Klaus

Also, if we are brainstorming old induction systems, as a food for thought:

It should also be possible to retain from the CIS only the fuel distributor, WUR (if needed), frequency valve and injectors of course. Then operate the control piston with a fast acting linear actuator controlled by a computer. You could even run individual throttle bodies and stacks with that. This would give CIS (if you can still call it that) a bit more power because that big restrictor plate in the air-stream would be gone. Also would allow to run radical cams if the controlling ECU can handle TPS controlled maos in low speed.

Regards,
Klaus

Klaus, you are an idea factory :) All the stuff I knew the background for but hadn't put together. I wonder if there would be any advantage to using the CIS fuel distributor system with individual throttle bodies, instead of a TWM setup with "regular old EFI"

Cheers,