[Gubi]

As just about everyone here knows, the Spider has an issue with the low fuel light blinking during vigorous handling. This is mostly because Alfa didn't put any electronic damper into the circuit. I've been searching around for potential improvements to this and may have found something that'll work.

Apparently Moto Guzzi motorcycles have a similar problem, and an owner engineered a solution: it's an additional circuit that goes inline between the lamp and switch to add a delay so the light only comes on when the switch is closed for a while. Some details and the circuit design at the links below:

Motorcycle schematics and utilities.

http://www.worwig.comule.com/schematics/Lightdly.PDF

I contacted the guy and it looks like the circuit on the bike is the same as in the Spider: there's +12v through the lamp running back to the switch at the tank. When the fuel is low the switch closes to ground and completes the circuit. He suspects it'll work fine on the Spider (the wattage of the bulb may make a slight difference in performance).

Some questions:

-Anyone out there with more circuit design experience than me have any opinions on this?

-Can someone double check that I'm reading the crappy Alfa wiring diagram correctly and have the right understanding of how the original circuit works?

-Does the low fuel light switch in the sender go directly to ground or does it go through a resistor? The wiring diagram is little confusing on this point. Anyone have one out where they could measure the resistance?

I'm tempted to try building this but it won't be until I get some more free time.

 

[Tifosi]

The papajam diagram I'm looking at shows the light to be just a switch activated thing.
Fuel level goes low enough, switch activates and sends + current to light, light comes on and grounds via fuel gauge body.

Wow, looking at that lightdly.pdf I gotta say, I don't have the first clue about what is actually being shown other than some squiggly lines LOL
Perhaps someone brighter than me can convert things into words like 'resistor' and 'diode'?

That aside, such a rig would prolly work just fine.
I recall a sorta simular thread from waybackwhen in the GTV subsection that worked in a resistor to stop the light flickering, but I'm not sure if it ever actually worked or was considered a viable way to do it amongst the participants at that time.
Basically it came up, got talked about, then fell off the world.

Do consider that someone is sure to trip by and say 'but the light makes a good G~meter' which at first kinda sorta makes sense, right up to the point where one realizes that if you actually need a light to tell you when side load is increasing then mabe that person shouldn't be trying to drive that hard because obviously thier cranially mounted stereoscopic fluid damped internal inertia detection devices aren't functioning up to par and they prolly fall over a lot just walking across a room

 

[86spider]

the 5.6k resistor keeps the lamp from illuminating when the tank switch closes

the 100k resistor and the 470uf capacitor form a time constant for charging the capacitor

when the tank switch closes the capacitor will start to charge. If the switch is not closed long enough the capacitor will not reach the necessary voltage. once the capacitor reaches sufficient charge the left transistor mps2907 will turn on, this will then turn on the right transistor, mps2907. this transistor is your new low fuel switch. when this turns on the low fuel indicator will illuminate. The diodes, 1n4001 control the direction of current flow, but I have been away too long and don't know what they are controlling there. Maybe they let the circuit turn off once the input switch is open.

changing the 100k resistor will change the time constant for the charging rate of the capacitor. It should work just fine.

 

[ghnl]

And all this is instead of adding fuel to the tank?

 

[Gubi]

Well, by "all this" you're referring to ten bucks in components and some soldering. It's really not all that much effort. Besides, I'm driven by curiosity and this is an interesting problem to potentially solve.

Anyway, if it just happened when the fuel were truly low it wouldn't bug me, but it happens any time I've got less than half a tank. It's a poor design that I wouldn't mind fixing. I've read enough complaints/questions about the issue on the BB that I'm sure I can't be the only one.

 

[Tifosi]

Try it and see if it works decently and I'll take a prefab off you for cost of materials + some extra for time, effort and shipping.

If there's druthers I'd prefer it sit right nearby the sender unit so it can get tucked into one of the nearby corners or perhaps even under the foam hose guard thingy.

Point of note: I would accept a small amount of flashing if it meant the light could come on at the top end of its range as opposed to not coming on solid until it 3 seconds away from too late. (passenger to driver: hey, what's that light. Driver to passenger: that's the indication that we'll be pushing in 1/2 a mile)

My light is already well calibrated and comes on solid with between 1 3/4 ~ 2 gallons of fuel left, but the flashies from half a tank down when bending the corners can be kind of annoying when doing such at night.

 

[Gubi]

the 5.6k resistor keeps the lamp from illuminating when the tank switch closes

the 100k resistor and the 470uf capacitor form a time constant for charging the capacitor

when the tank switch closes the capacitor will start to charge. If the switch is not closed long enough the capacitor will not reach the necessary voltage. once the capacitor reaches sufficient charge the left transistor mps2907 will turn on, this will then turn on the right transistor, mps2907. this transistor is your new low fuel switch. when this turns on the low fuel indicator will illuminate. The diodes, 1n4001 control the direction of current flow, but I have been away too long and don't know what they are controlling there. Maybe they let the circuit turn off once the input switch is open.

changing the 100k resistor will change the time constant for the charging rate of the capacitor. It should work just fine.

Thanks, that makes sense in looking at the circuit. I'm a Chem E so most of my electronics falls under the category of "stuff I've picked up while doing other things". I generally remember not to pick up the soldering iron by the hot end, though.

He mentions that the circuit is designed to illuminate the light for ~30sec at start up as a test. Perhaps that's the reason for the additional bits. This is important for a bike that has no gauge, not so much for a car. I may talk to some of my EE budies, see if they can come up with any improvements.

 

[Gubi]

Okay, so this is pretty cool: it's a free circuit simulator that's really easy to use. It's a browser Java applet but you can also download it and run it locally (which seems to work a lot better, plus you can import/export).

Circuit Simulator Applet

If anyone else is interested in this, the delay circuit as described in the original link is coded below and can be imported into this app. I think I'm going to play around with the RC values to change the timing as it's a bit too slow in current form.



$ 1 9.999999999999999E-5 40.34287934927352 47 12.0 43
v 640 368 640 112 0 0 40.0 13.6 0.0 0.0 0.5
t 384 160 464 160 0 -1 4.480367288427315 -0.6855316359104338 40.0
t 336 176 384 176 0 -1 3.8876749541845164 -0.5926923342427988 40.0
w 464 112 336 112 0
w 464 112 464 144 0
d 336 176 336 112 1 0.805904783
r 336 112 192 112 0 5600.0
r 336 176 192 176 0 100000.0
d 192 256 336 256 1 0.805904783
w 336 256 336 176 0
w 192 256 192 176 0
w 192 176 192 112 0
c 336 368 336 256 0 4.7E-4 -3.8876749541845164
w 384 192 464 192 0
w 464 176 464 192 0
w 464 192 464 368 0
w 336 368 464 368 0
w 464 368 640 368 0
s 48 112 192 112 0 1 false
w 48 368 48 112 0
w 48 368 336 368 0
181 640 112 464 112 0 1803.7879188393038 1.2 12.0 0.2 0.2

 

[Tifosi]

Is that a loop back/full circut ground I see rather than a remote ground?

I mean is it going to require a + wire all the way up to the light, then a - wire to come all the way back to the gizmo?
(if that were the case I suppose it could all be done up in the console as opposed to back at the fuel sender)

 

[alfaparticle]

I don't understand the function of the lower diode. The upper one enables the transistor to turn off by discharging the capacitor through the lamp after the switch opens. The lower one appears to bypass the charging resistor. I am an EE but I graduated in 1969 and I have not worked on circuits in a long time.

 

[Alfacliff]

the lower diode charges up the cap if the switch is open through the 5.6k resistor.
cliff

 

[Gubi]

You think you've got it bad, Ed. I'm a friggin' Chemical Engineer. I look at this crap and go "ook, ook" half the time.

Yeah, it looks like I can reduce the initial lamp on time by reducing the 5.6k resistor to something like 2k, and then speed up the response time by reducing the 100k to maybe like 20k. Or I can probably do the same thing just by using a smaller capacitor, I think. Any advantages to adjusting the response using one or the other, or is it just a balancing act?

 

[Gubi]

Is that a loop back/full circut ground I see rather than a remote ground?

Ground is ground, Tifosi. I just mocked it up as you see it in the picture but in reality everything will just ground to the chassis in the rear.

I envision this as a little box in the trunk. You unplug the wire from the tank switch and plug it into the box. Then the box has two outputs: one goes to the switch and one goes to the ground connection at the tank switch.

 

[Tifosi]

Ground is ground except when its part of a unit that loops back to itself where the unit controls the ground circut, not the things attached to it (like the ECU controlling the ground for the injectors for example) which is why I asked.

I don't have a clue what that drawing is other than squiggly lines as I said before.
To me it looked like the light ground had to come back to the diodes and whatnot to actually complete the circut.
If it was more like the attached, it wouldn't have even come up.

If it can be done with a little box plugged in as you describe, that would be cool.

 

[Gubi]

Here's a cleaned up and fixed circuit: changed some of the transistor values to be more correct. I may try to put a second circuit in the box to add some damping on the needle.

$ 1 9.999999999999999E-5 40.34287934927352 47 12.0 43
t 496 192 576 192 0 -1 2.191217811720318 -0.6889805644072853 100.0
t 448 208 496 208 0 -1 1.5950765489806666 -0.5961412627396514 100.0
w 576 144 448 144 0
w 576 144 576 176 0
d 448 208 448 144 1 0.805904783
r 448 144 304 144 0 5600.0
r 448 208 304 208 0 100000.0
d 304 288 448 288 1 0.805904783
w 448 288 448 208 0
w 304 288 304 208 0
w 304 208 304 144 0
c 448 400 448 288 0 4.7E-4 -1.5950765489806666
w 496 224 576 224 0
w 576 208 576 224 0
w 576 224 576 400 0
w 448 400 576 400 0
s 160 144 304 144 0 1 false
w 160 400 448 400 0
181 752 144 576 144 0 2488.9380587726982 1.2 12.0 0.2 0.2
g 160 144 128 144 0
g 160 400 128 400 0
R 752 144 800 144 0 0 40.0 13.6 0.0 0.0 0.5

 

[John Gallagher]

Looks like Pac Man to me.

 

[crstophr]

Been playing with that circuit simulator...

So if the in tank switch is closed for more than one second the light comes on and stays on until this ignition is shut off. Changing the 100k resistor upwards lengthens this time. i.e. 500k means the switch has to be closed for about 4.5 seconds before the light comes on. neat.

I'm relatively new to electronics. Do we need numbers for the transistors and diodes? I will build this for sure once the parts details are filled in.

If this works out maybe I'll finally get inspired to do my tiny arduino controlled idle air valve to replace my defective AAV project.

--Chris

 

[Gubi]

So if the in tank switch is closed for more than one second the light comes on and stays on until this ignition is shut off.

Er, no, that's not how it should perform. Light should come on at startup then dim over ~30sec. If switch closes, light will slowly get brighter until reaching full brightness. If the switch then opens, it should slowly dim again.

Components are in the original link.

Transistors: MP2907 (NTE159 is an equivalent)
Diodes: 1N4001 (NTE116 is an equivalent)

Everything else is generic.

 

[alfaroad]

lower diode it is actually a transient or surge protection for the base of the first transistor while smooths capacitor´s charging-re charging. The idea here is to create an active RC charging resistor to trigger -bias the base of the first darlington transistor so to make the final transistor close meaning finally to be connected to the ground. Lets say the 5.6k creates i.e 1V drop across the line, also depending of the exact type internal resistance , the later dictates the circuit main current. Normally, bulb will be still on in case nothing is connected to the first transistor´s base. However, in this circuit within lower diode at the very first time when tank switch is on, gives a first bias voltage to the transistor base so to keep the output transistor open. Just after and depending the RC time costant the bias voltage goes on from the charging capacitor. Output transistor will close so to be connected with the ground only if bias on the front will be greater than a point. If during this circle switch opens again then capacitor will discharged via the same lower diode and at this time the 100k resistor behave like a bleeder resistor

 

[alfaroad]

If when firing the circuit , the bulb lights on and then dims, yes, the idea is the same above just means that the choosen transistors are PNP type so, if nothing bias the base of the front transistor the output transistor close, practically near close since the internal resistance creates some voltage drop thus the not full bulb lighting when firing