The Fusion Brain carries a tapering voltage from one port to the next around in a circle. The attached picture has a 5V applied on input 6, a jumper on input 0 and nothing connected anywhere else. A large value input followed by a low value variable input is disaster. On/off inputs can be programmed around. Example: Check engine on 6 with 5V input "on", creates a 3.8V on my port 7 security indicator when off. So I've set it 0-3.899V=security light off, 3.9-5V=on.
I've decoupled inputs with 0.1 uF capacitors for an improvement, but precise results require a jumper on the preceding port grounding the input. Any other suggestions that doesn't require using 2 ports for each precise variable input?
Fusion Brain Version 6 Released!
1.9in x 2.9in -- 47mm x 73mm
30 Digital Outputs -- Directly drive a relay
15 Analogue Inputs -- Read sensors like temperature, light, distance, acceleration, and more
Buy now in the MP3Car.com Store
To expand on Toaster, if you have a analog input you treat as digital, it is easy to end up with the zero state floating. If you use a switch to toggle 5 V to the input, then consider a pull down resister to keep the input from floating when not connected to 5 V.
My inputs must be floating then. My tach for instance that is 0.4V at idle would pop up to 0.7V when my check engine light came on. A large change on my gauge. Everything has settled down nicely with a jumper in front of each of my variable inputs. I'll just need third FB to do everything I want.
I've seen the same issue. I use jumpers on any unused input, but I didn't notice any cross-talk when using adjacent inputs (very informal & limited testing). You might try some shielding on the wires/connectors. A little aluminum foil & some tape can go a long way, ground the foil. You might also seperate 0-5V sensors with either an unused input, or a all-or-nothing digital-type input, just reduce the V lower & set your trigger to like 2V, that will reduce any EMI a little.
FWIW, most signal-level wires in an automobile have some type of shielding for the same reasons, it's not just a FB issue. They also usually have conditioning circuitry in the electronics.......a car is a very noisy environment.
Another track you can use to reduce noise. Keep the big wires away from the little wires. Which means keep the wires that power things like fans, ignition, etc. away from wires that are used to cary signals to your fusion brain. Distance and shielding.
Part of this has to do with design compromises needed to keep the FB cheap. A comparable DEFI gauge setup would run you over $1200, which opens up component choices.
Part of the issue may be that the F2V converter cannot sink current. So, the sample and hold capacitor gets charged by the previous input, then that charge has nowhere to go - the ADC has very high impedance, and if the F2V converter can't sink it, it can't go anywhere. Try a pulldown resistor between the signal wire and ground on the F2V converters. Off the top of my head, maybe a 5kohm?
No argument on the cost of competitive products. They lack the versatility as well. Can't blame me for trying to work through the issues.
If the ground jumpers in between corrected the inputs, I don't think its wire interference. I tried several combinations down to different single sensors on the FB. All exhibit the added voltage, not just the F2V. My Tach is happy with this arrangement with only minor fluctuation. Wish I could smooth it out a little more without the sluggish delay a 10-15 average adds. The fluctuation matches what I hear, so I think it accurate.
Thanks for the pull down resistor suggestion. I'll try that next on the speedometer. Up and down my driveway, if I stop fast it will usually zero. Long slow stops leave a 0.15-0.25V. My other thought is to pick up the signal straight from the VSS instead of the "cleaned" sine wave from the PCM.
If you set the I/O timer to a lower value (say, 10mS) you can get away with 10 point averaging and now much sluggishness.