Zemco Fuel Computers

[bkahler]

Frank, the 24' motorhomes (Argosy or Classic) are almost as scarce as the 20' Argosy motorhomes. There aren't many of the 24s and from my perspective they are for the most part an ideal size. Big enough to not feel cramped inside yet small enough to maneuver in traffic and parking lots.

Oddly enough I have a 24' Argosy I recently towed out of the mountains and it's currently sitting by my shop. Even stranger it's vin number is within 3 digits of Tony's 24' Argosy.

Your Fitech sounds like an interesting system to work on. I've found the GM throttle body rather easy to work on. I mounted the PCM in the center of the dash behind the radio. If I'm not mistaken the wiring harness construction from the injectors back to the PCM features twisted pairs of wires. At least that's how the harness for the TR6 was manufactured.

I guess I've just never been of fan of intrusive type connectors. Should I decide to make the Zemco permanent I'll do a proper job of fabricating a new pigtail adapter. Actually I'll need to do it twice as I will likely add a Zemco to my TR6 as well since it's also fuel injected using the GM system.

Brad

[Magnet18]

So while thinking about Brad's question I simulated about half the circuit in LTSpice (for no real reason), and I have a nit to scratch. Yall can ignore this unless you want to hear an EE jargon ramble.

I don't know why they tied two op-amp outputs together, as that seems like a no-no since opamp outputs can both sink and source current, and by design when one is sinking the other is sourcing, so I feel this risks eating your own signal

However, for the specific part used the source and sink characteristics are unbalanced, so it works

I would have tied them through a resistor divider and sampled in the middle, you're guaranteed to lose exactly half the signal, but you know it will be there.

That said the circuit works fine in simulation and hopefully in reality, and I won't pretend to have a lot of analog design experience. Just something for me to keep in mind if I decide to make changes.

Also the 12V to 5V power conversion via zener diode seems cheap, I'd just plop down a proper linear regulator, but it works and I suppose one could argue anything more is wasteful.

Just my musings
I'll probably simulate the full circuit just for kicks
I've half a mind to build an Arduino version of the Zemco
We'll see, one step at a time, first this

[FrankGRUN]

Brad, My current sating vehicle is a 1982 Born Free Class C built on a Ford E350 chassis with a 460 engine. It is 24 ft and 11,000 pounds GVW and just the perfect size. Its also a perfect hobbyist beast to play with just beyond the tender reach of the DMV/EPA. The size is ideal for city, highway and camping. However, the 24ft. 240 was significantly bigger inside. I thought the 20' Argosy was a bit cramped, though I only saw one briefly on a cross country tour in Nebraska.

After years of frustration tuning carburetors while working within the 1.75" clearance from top of carburetor to the dash brace in the E350, I find the Fitech a fresh opportunity. I looked at the throttle body injection systems, but the factory ECU systems were locked to closed loop and a narrow band O2 sensor. Going to aftermarket ECU systems like MegaSquirt courted with potentially obsolete systems with no hardware support. Software issues were tenuous. But the Fitech system complete is the same cost as the MegaSquirt control system, which leaves purchase of EFi throttle body and ignition trigger systems left to fund. Besides, all the GM TBI throttle bodies I have encountered are 2 BBL which fits poorly on my 4 BBL manifold!

In any case, my focus on the Fitech system is the ability to run at user controlled stoichiometries which can be modified from the self-tuned parameters if one wishes to do so. Think cruise at 15.52:1AFR!

Also, twisted pair is good! I only have seen simple straight to the injectors.

[FrankGRUN]

Rob, I was always frightened of op-amps! But I thought Zeners in the 80's were to EEs like Amoxicillin was to pediatricians!

BTW, I'd certainly vote to take the final circuit to 2020's electronics! Your surface mount approach is already a reliability enhancement, especially compared to my wire-wrap suggestion. But Arduino ......

How could that compete with the clean, understated 1980's design of this Zemco ZT-3?

[Keyair]

Good job guys.
FWIW, look what fell out of box yesterday into my hands.
TBI from my old 1987 454 Suburban.

[FrankGRUN]

Yep, sure looks familiar!

[Magnet18]

Circuit boards are ordered! Should be here in about two weeks. Between now and then I'll finish the BOM and get the components on order.

This is my rendering of what they should look like, QR code functional, just playing around

Strain relief will be left as an exercise to the user for this iteration since I ran out of room and it's midnight. Usually I like to put a couple drilled holes in front of off board contacts to weave the wire through for strain relief. Or maybe I'll go to a standard connector header?

Now I need to get out the other CAD and make some housings to 3D print!

[Magnet18]

Originally Posted by FrankGRUN

Rob, I was always frightened of op-amps! But I thought Zeners in the 80's were to EEs like Amoxicillin was to pediatricians!

BTW, I'd certainly vote to take the final circuit to 2020's electronics! Your surface mount approach is already a reliability enhancement, especially compared to my wire-wrap suggestion. But Arduino ......

How could that compete with the clean, understated 1980's design of this Zemco ZT-3?

Op-amps were never my favorite either! It looks like the part was specially chosen to be designed to interface with logic circuits, so the output impedance seems to be pretty high. When I swapped a "normal" op-amp into the system, it broke, but I can add a couple diodes and make a (in my mind) simpler system. But the original parts seem to still be available, and if it ain't broke I probably ain't gonna fix it.

2020s electronics... throw a micro controller on, bluetooth to the phone and log all data... like I said, one thing at a time



How many injectors are in your fitech system? This whole premise revolves around the zemco not knowing the difference between this and the fuel flow sensor and using the calibration to tell it fuel per pulse. I wonder what the bounds to that calibration are, as in did they design for 2 injector systems, and would a hypothetical 8 injector direct injection system then be so far out of whack it can't calibrate?

I'm guessing it will be fine, but if it turns out there is a sweet spot the computer likes it would be pretty easy to put some jumpers in here to tell it how many injectors there are and divide the pulses less or more. Just a think

[FrankGRUN]

Four injectors on the Fitech system. Higher Hp rated Fitech systems have 8 injectors. At this point, I don't know if they are all firing simultaneously or whether there is a pattern response involved. I'm sampling one injector.

Getting close! Exciting!

[bkahler]

Originally Posted by Keyair

Wow, its been strange... the Zemco godfather must be smiling at me!

Picked up a used ZT4 a while back for $20, and it was just the head unit and fuel flow sensor... Figured it might work for an idea I had to watch the fuel consumption of the onboard genset... or dissection or something...

Anyway, with Magnet18 working on duplicating the EFI Module, I figured I would mail it to him to mess with, test or something... but there was no wiring harness or speed sensor, etc. He said cool, he could figure it out.

Then another popped up... used but looked complete.
Made the guy a cheap offer which he accepted.
Attachment 318879

That is heading your way next week Magnet18.

The surprise was hiding in the box...
An opened, but seemingly unused "P/N 4792 Fuel Return Kit".
Probably means nothing to 99.999% of people... but to me, having seen it mentioned in the install manual, I always wondered what it was, how it worked, and if I needed it...

Drum roll.....

Attachment 318878

Steve,

I jut looked through the five (or six?) boxes of Zemco's that I have and two of them had the return kit. Kinda surprised me when I saw them

In my collection I have either two or three ZT-4 units, one ZT-3 unit and a ZT-41. Of the group the ZT-4 is the nicest and smallest.

Rob,

Your comment about using an Arduino got me wondering if you could remove the guts of the display unit and use it as part of a conversion to Arduino or Pi based system.

ZT-41


ZT-3 and ZT-4


And, for whatever reason one of the boxes had this chain included


Brad

[Magnet18]

Originally Posted by bkahler

Steve,

I jut looked through the five (or six?) boxes of Zemco's that I have and two of them had the return kit. Kinda surprised me when I saw them

In my collection I have either two or three ZT-4 units, one ZT-3 unit and a ZT-41. Of the group the ZT-4 is the nicest and smallest.

Rob,

Your comment about using an Arduino got me wondering if you could remove the guts of the display unit and use it as part of a conversion to Arduino or Pi based system.

ZT-41
Attachment 337278

ZT-3 and ZT-4
Attachment 337279

And, for whatever reason one of the boxes had this chain included
Attachment 337280

Brad

Yes, that is where my mind was going. Just drop in an Arduino or pi to do all the computing and drive the original displays and read the original buttons. With that, it might be possible to take reading from both the genset and engine and use it for calculations involving both. The tiny ZT4 might actually be a challenge for space, but i suspect it's still doable with surface mount components.

Other than being functional with the genset, I'm not sure what the advantages really are though?

[FrankGRUN]

Very nice idea - maintaining the Zemco mileage computer look. As just a side suggestion as you are contemplating approaches to a further generation, I believe there are chips available which include imbedded GPS and return actual road speed. That would really reduce the install hassle to 12V, ground and the EFi signal tap!

Just fantasizing!

BTW, the FiTech injectors are bank firing.

[FrankGRUN]

Unfortunately, as best as I can determine, GM TBI systems sequentially fire their injectors at stoichiometric and lean AFR levels. For rich AFR's they revert to simultaneous or bank firing. Hopefully, the calibration approach will resolve this. Otherwise, a summing game may be required.

[Magnet18]

Originally Posted by FrankGRUN

Unfortunately, as best as I can determine, GM TBI systems sequentially fire their injectors at stoichiometric and lean AFR levels. For rich AFR's they revert to simultaneous or bank firing. Hopefully, the calibration approach will resolve this. Otherwise, a summing game may be required.

Hmm, that's interesting. I'm not sure how the calibration would be able to know the difference, but as I think about it real time while writing this stream of consciousness down, I'm not sure it really needs to.

If you think about it, each pulse from one injector still represents two pulses, wether or not the pulse from the second injector happens simultaneously or sequentially should be irrelevant, right?

[Magnet18]

Changing gears for a second, can those with EFI setups get me more information on the pulses? Mostly I need to know roughly what the width of the pulse is, and roughly the frequency range. Preferably oscilloscope captures, but datasheet information from the manufacturer would be great too, trying to figure out what to expect.

I tried simulating the rest of the circuit, and the second opamp pair is confusing me. As-is, it doesn't want to work... At all, not in simulation at least. I'm also not sure what they do.

Suffice to say, I intend to bench test both the original unit, and the new one when it arrives, and compare the performance. In order to bench test, I need to know what the pulse train looks like. For simulations I've assumed a 30mS pulse every .125 seconds, but i would like some verification. Should be easy enough to emulate the output via Arduino

[Magnet18]

Originally Posted by FrankGRUN

Very nice idea - maintaining the Zemco mileage computer look. As just a side suggestion as you are contemplating approaches to a further generation, I believe there are chips available which include imbedded GPS and return actual road speed. That would really reduce the install hassle to 12V, ground and the EFi signal tap!

Just fantasizing!

BTW, the FiTech injectors are bank firing.

Yes, when in college a friend of mine interfaced to one of those chips for a project, shouldn't be too terribly hard. Fwiw, it would still need either a clear view or antenna on the dash, but yea, in general I think gps speedometers are a fantastic idea!

[FrankGRUN]

Rob, I'll get you some oscilloscope traces tomorrow. But, one thing I need to note here is my experiments with the flow sensor packaged with the Zemco carburetor units. As you probably know, the sensor is essentially an optical isolation switch with a rotating interrupter. The lower part of the optical stage of the sensor is just a continually on 12v GE161 incandescent light. This light is focussed with a lens to a channel that bisects the raceway channel with the opaque circulating ball driven by the gasoline flow. On the other side of the light channel, there is another lens focussing the light on a photocell. The output of the photocell is the input to the Zemco unit. I found no voltage output at the photocell terminals. Instead, the resistance with the light on and an uninterrupted channel was 0-2.2 ohms. Interrupt, the resistance was greater than 10,000 ohms. Running, you see a on-off switch (nice square wave) the frequency of which varies with the flow rate. Hope this helps, more data tomorrow!

[Magnet18]

Thanks Frank!

[bkahler]

Speed sensor is now attached. I can't install the magnets until I raise the rear wheels off the ground so I can rotate the drive shaft. Maybe later this week.

[Keyair]

Looking good Brad, altho the sensor might need to be closer to the magnets.