Hey Guys, I've been lurking the forum for a while and have finally registered to ask you for professional help (Not that kind of Professional help, although I'm sure I need that too).

Long time bass player here, on the never ending quest for the perfect distortion. I know. However my question is not related to music, equipment, stompboxes or effects. It's automotive. If forum mods don't appreciate the unrelated topic to stomp boxes and want to delete I understand. But, out of anyone I know on the internet you guys would know how to build such a circuit.

I am looking to build a kind of (dare I say) voltage regulator that would adjust it's output voltage in the name of maintaining a specific input voltage. As input voltage increases it too would need to increase output voltage that would result in the input voltage going back down. Opposite for lower input voltage, taking away output voltage to bring input voltage back up to specified volts.

To get specific, I need to maintain an input of 2.5Vdc and output would be variable between a range of 0v-5v. The output would be ever adjusting as engine RPM, Load and driving conditions are always changing. Maintaining the 2.5v input could mean 1Vdc output at idle, 1.3v-3v during low throttle cruise and 3.5v-4.8v at higher RPM...all in the name of maintaining a 2.5Vdc input.

Is this possible? I'd like to see if it's even possible before altering input voltage for different AFR's. Can such a creature be created?



Automotive terminology used in previous graph:
AFR = Air Fuel Ratio
MAF = Mass Air Flow Sensor
DME = manufacturer specific acronym referring the the vehicles on-board fuel injection computer.
Innovate MTX-L is a brand of Air:fuel monitoring gauge with a sensor mounted in the vehicles exhaust stream that gives live data of the current air to fuel ratio being consumed by the engine. The gauge has a voltage output that can be used for datalogging or even "standalone fuel controllers" essentially what we're trying to build above.

Appreciate any and all help! Cheers.

You'll probably need to use some sort of PID control. That may not be fast enough for an engine.

Isn't this the sort of thing that the ECU should do? The fuel map is typically (incrementally) altered over time due to inputs from the o2 sensor. If I understand you correctly, you want the "voltage regulator" to change the maf input in response to the o2 output in order to maintain stoichiometric air/fuel ratios. That's a very tall order. I hate to say it, but if that were possible, ecu tuning wouldn't exist.

Welcome!

IMO, you'll burn a piston. If it runs and pulls a load, that is.

Feeding AFR to a MAF input seems quite wrong. The ECU needs MAF for other things. The AFR should not be 14.7 all the time; at the least it should run lean at light cruise and rich at full power.

Are you assembling a rig that is missing major parts (like MAF sensor)? Or trying to trick the ECU into a non-stock action?

TAKE CARE! I REALLY DON'T KNOW WHAT I AM TALKING ABOUT!

The problem is, you will burn the most of components you try. Trust me, i passed a similar problem before. But maybe you can use a dimmer to regulate the tension. Can be a nice idea, no? You can find several circuits on google, and you should to replace the potentiometer for a trimmer only for safe question, but it will be not automatic, you will need to adjust when want. At least, of course you use two dimmers and a couple of relays, like an PC power stabilizer (you see it?). Wait the others answer too for you see if it will works... But i think will. I used one on a bike battery to regulate the luminosity of the headlight lamps.

I hope help you! Be welcome!

Glad to see some automotive gearheads here wanting to help.

The application we're talking about here is an '95 BMW 540i with a 4.0L V8 and 6spd manual that is set up for hot street.





The OBD1 Bosch Motronic 3.3 computer system is very basic. With only just a few tables and parameters controlling the engine. This computer is kept in open loop with the narrowband O2's deleted.

Installed in the computer is a chip emulator that I can download, modify/rewrite the tune and upload to the emulator. The MAF table has the single most say so in instant fueling control, looking at the MAF table in tuning software shows that it's actually a Load vs Voltage graph and the table is filled with injector milliseconds.






The only effects on the Load axis is inputs from TPS/throttle angle and vehicle speed/MPH.

This is why I wanted to tie the aftermarket wideband O2 and gauge into the MAF control of the vehicles computer.

It would remove the MAF sensor so that I can build my own induction and intake manifold with different throttlebody sizes and designs. Having the wideband O2 controlling vehicle fueling would mean that it's instantly compensating for engine changes, modifications, camshaft timing, altitude, load...everything.

The 14.7:1 was the safe point to start with for testing to see if this could be possible & driveable. Really I'm only concerned about two AFR's for this application. 14.0-14.7:1 for idle, cruise, light throttle low load. And 12.5:1 for acceleration, high load and high rev. I can worry about lean cruise later, if ever.

BTW, I've modified the rear door hinges so the doors pull off completely. My Fender Pro 8x10" cab slides into the backseat, rehang the doors. My Mesa Boogie Titan v12 head and Fender American Jazz fit nicely in the trunk :-)

Any significant road trips & tours we use the truck & equipment trailer. But this works for me to bounce around from studio to studio.

i once had a maf die on me, didnt have the 160$ to replace it right away. i used a few diodes to get @ 2.6v and it ran, not great but ok for a few days till i could get the maf... i think i ran the diode btween +5v and the output pin...  my mechanic friend still loves to tell that story! 

What you want to do is theoretically possible using an error amplifier this does exactly what you want your "Automatic voltage adjuster" to do. I've got no idea whether you could actually build one that would work in practice, I doubt it's something that someone on a random internet forum could knock up for you.

The fundamental problem is that the computer needs the O2 sensor output. The MAF, together with some measure of engine loading, gives a preemptive injector pulse duration. The O2 sensor is "post-emptive". Another way to say it is that the MAF is feed-forward while the O2 is feedback in relation to the fueling table.

Think of it this way. If you wanted the O2 sensor to change the MAF input on the computer, the "voltage regulator" would also need the engine loading information. You would basically have to make a second engine computer just to process the wideband O2 sensor output! The MAF, in conjunction with the engine load/rpm, is the only way for the computer to decide how much air is flowing into the engine. Without the MAF, or a similar feed-forward sensor(s), the computer has no idea how much fuel to inject. The O2 only works after combustion has taken place, and it only takes one serious knock to damage a piston.

Theoretically, an error amplifier or a PID control scheme could work, but those systems have certain time constants associated with them that are still longer than the time it takes for a serious lean-condition to cause damage.

It doesn't matter what the scheme is, injector duration has to be controlled from the air entering the engine and is only corrected by the oxygen left in the exhaust. If engine computers could work without feed-forward air sensing, no one would use those sensors because your theoretical O2-only based system could be used for any engine (which is not the case). It may be time to look at megasquirt.

Appreciate all of the input. All valid points. I understand that, obviously, the wideband O2 is post combustion process. The Bosch Motronic does have a "accelerator shot" (like on a carb, for lack of a better term) when TPS is opened it increases injector pulsewidth briefly to compensate for throttle openings.

Even at a 750rpm idle a four stroke eight cylinder has a combustion process 50 times a second. I would imagine that would be enough time for the exhaust oxygen levels to be changed, monitored and compensated for.

Some of the fastest street cars I've ever worked on/ driven were mafless Speed Density. I was hoping to take it one step further with AFR tracking.

I'll start looking into the possibilities and limitations  of using an Error Amplifier circuit. Thank you.


Cheers.

I'm investigating the use of the TL431 as an error amplifier or even a TL081 and what I would need to do to implement for my idea and application.

Anyone have any ideas for a circuit? Surely it can't be as simple as:



Maybe with a small Cap on the cathode to vehicle ecm to buffer the voltage switching?

No, wait....the TL431 as an Error Amplifier is really a Shunt Regulator... the circuit would need to look more like:

Last point first: put a Holley 4bbl on and learn the jet kit. A Holley won't make the MOST power an engine is capable of, but it can comes so close that if you still need "more", you really want a bigger engine. You can bend pipe before and after the carb any way you like.

> The Bosch Motronic does have a "accelerator shot" ...when TPS is opened

Bah. 1988 GM TBI had that and it even works. I bet the early VW Type 3 EFI did, and it didn't have a computer. Perhaps the 1960s Corvette no-CPU did, the drawing is too much for me. The mechanical Hilborn "did", not-really, in that fuel was proportional to throttle lever, didn't actually track air-flow at all (only good for flat-out racing). "Accelerator pump" action is THE reason for TPI. (Secondary reasons are idle-to-driver control transition, and maybe full-pedal enrichment.)

OK...

> wideband O2 controlling vehicle fueling would mean that it's instantly compensating

What is "instantly"?

Yes, you change manifold, the O2 sensor will guide the ECU.

But I believe there is a significant delay, tenths-second?, between change of O2 in the exhaust and change of O2 sensor voltage.

Indeed (if I am right) that is why they put all that other stuff on. MAF or MAP*RPM can be sensed fairly quickly. TPS almost gives fore-sight of what is starting to happen. O2 lags, and is used to refine the rough-guess (fuel map) after-the-fact. O2 lags first because the mix has to go through the engine (2 revolutions) and head-pipe, second because it is a thermo-chemical process. Yes, O2 "errors" are stored in RAM and used to inform future ECU actions. (As I discovered when we replaced my ECU-- drivability was bad and improved over time as the used ECU "learned" my worn engine.)

I admit I am not hep to these wide-band O2 sensors. Mine just give O2 yes/no. Yes is lean, no is rich, and the ECU wobbles the mix so the O2 comes out mostly-yes for cruise, mostly-no for go.

slacker> possible using an error amplifier

Yes.

Tablize your data!

O2:
5V = super-lean
0V = super-rich

Injector:
0V = short injection
5V = long injection

No engine can really use a lot more than 30:1 range of injector times. In fact if it is *time*, then we really only have to cover air pressure change (RPM change will take care of itself), a 3:1 range. So most of those "13"s and "4491" in the table are never needed. Assuming I am wrong here, I still doubt it needs more than a 1V to 4V range. (Hmmmm... if engine spins 7,200RPM, one revolution is 8 milliSeconds. If "4491" is 4.4mS, that's about an intake stroke. Yes, injection does not have to be in-sync with intake stroke; my GM jug does not even try. And *maybe* you "need" 5V signal. A voltmeter and a brisk run would tell you what you really have now; no mod short of NOx will allow a big-heap longer injection.)

We want perhaps 4V output swing from much-less than 1V input swing. The wbO2 output is 5V for 15AFR, so 1AFR accuracy is 0.3V of wbO2 swing. (Do not sneer at 1AFR accuracy if you don't face smog-checks or $4 gasoline. 1AFR rich is rarely wrong for power.) Oh, call it +/-0.1V wbO2 swing.

And be SURE you know which-way does-what!! "Error amplifier" is same-as a governor on a steam engine. And Murphy's Law says it will get connected backward: too-fast *opens* the throttle and it goes too-too-too-fast.

ASS-ume 2.5V O2 = 15 AFR. (I'm not going to look-up your parts for you.)

And if wbO2 voltage is high, we want a higher injector voltage.

So if wbO2 is 2.6V, we want injector voltage down toward 1V.

And if wbO2 is 2.4V, we want injector voltage up toward 4V.

Check me on those which-ways!!

If so, we want a non-inverting amp.

Reference input trimmed to 2.5V (whatever your wbO2 thingie gives at your target AFR).

And +/-0.1V (0.2V) input error causes at least 3V output swing, gain >15. (Note that an error amplifier with too much gain WILL oscillate, law-of-nature.)

I thot your bimmer had a Ferrari tag but I was wrong:

1932

"Last point first: put a Holley 4bbl on and learn the jet kit."

I considered it, using my Demon carb and mad max style carb scoop. I'd even found an intake manifold I could modify to work and bolt the carb down.





I've done an OBD1 Bosch Motronic to carb conversion before, on a SAAB 900SE 2.5L v6. Ignition timing remained computer controlled. Fueling was with two Redline Weber carbs on a home made plenum:











"I thot your bimmer had a Ferrari tag but I was wrong:"

I put the pegasus silhouette on most things I build. The mobil pegasus just happens to be the easiest to find in vinyl sticker. I put them on cars & guitars.  I have horses, I figured a horse with wings is a good insignia to emblematize my projects:



Here is another project vehicle, 2004 Silverado mashup:





"So if wbO2 is 2.6V, we want injector voltage down toward 1V.

And if wbO2 is 2.4V, we want injector voltage up toward 4V."

Flip your idea. 2.6v out of O2 gauge is leaner, increase voltage to MAF signal to add fuel to richen fuel mixture to compensate and O2 gauge voltage will go down.

2.4v out of O2 gauge is richer, decrease voltage to MAF signal to lean out fuel mixture and O2 gauge voltage will go up.