Fuel Rail Pressure Sensor Calibration?

[rbattelle]

Originally posted by KBennett

One more:



If you still have that onboard computer you could add a data acqusition card or serial converter that can convert voltage signals to something the computer could read. This value can be read by software and displayed on your screen in any format including analog.

Kbennett,



Well, the Simpson I got is apparently quite rugged and able to withstand shock and vibration. I already bought it, so we'll see what happens. Impedence is apparently 1k/volt, which is (hopefully) enough to keep the current down and not effect the ECM. Question is, how low is "low enough"? I'll test the meter when I get it. If it's too low, I won't use it.



I do still have the computer. I'd love to get a NI DAQ board and build a Labview application to read all the sensor voltages! But, that equipment is expensive and I'm not willing to invest in it just now. I'll need the DAQ, and a connector panel for it. I need to keep an eye on Ebay, I guess.

 

[JNutter]

I wonder what the impedance across the sensor is? I'd suggest putting an ohm meter across the ground and +5 terminals and measuring. The electrical part of the sensor is just a pot, with the wiper attached to the signal connections. At least that's how most pressure type sensors work. If the impedance of the gauge is significant at all (when compared to the sensor), it will drag down the voltage seen by the ECM. This will cause the ECM to increase the fuel pressure, possibly far higher than you'd like.



I'd look for an gauge with an input impedance of at least 100 times the impedance of the sensor. 1K/volt may not be enough. Higher impedance on the gauge is better. I'd think that the 1 meg ohm gauages would be a good idea, but I don't know what the impedance of the sensor is.

 

[rbattelle]

Originally posted by JNutter

The electrical part of the sensor is just a pot, with the wiper attached to the signal connections. At least that's how most pressure type sensors work.

Hmmm. . I've been assuming it's a diaphragm-type sensor with an on-board Wheatstone bridge. If that's the case, there is no penalty for connecting an additional resistance across the bridge arms. Since it's exposed to fuel at such pressure, the wetted parts (ie diaphragm) is most likely stainless. Besides, diaphragm-type means you don't need a spring.



Anyone know if this sensor is a diaphragm or a potentiometer? And if it is a diaphragm, is it using a standard Wheatstone bridge-type circuit?

 

[silver03]

it doesn't work that easily. it will have some resistance in the off position, but putting a meter across is in the "ohms" position will throw it off when it is running. putting a resistor across it will work, but it will effect idle just as much as it will full throttle. you would need to come up with a circuit with a transistor that turns on the resistor in parallel slowly with increase fuel pressure. it has been done onthe fords with no problems.

 

[rbattelle]

Originally posted by silver03

it doesn't work that easily. it will have some resistance in the off position, but putting a meter across is in the "ohms" position will throw it off when it is running. putting a resistor across it will work, but it will effect idle just as much as it will full throttle. you would need to come up with a circuit with a transistor that turns on the resistor in parallel slowly with increase fuel pressure. it has been done onthe fords with no problems.

I don't understand how this works. Exactly what type of sensor is it? Strain gauge? Potentiometer? Those are the only 2 types I know. If it's a strain gauge, a voltmeter across the bridge arm would work fine. If it's a potentiometer, it may be a bit more complex (like JNutter mentioned).



If you connect a meter to it in parallel, then both the ECM and the voltmeter will see the same voltage, by definition.



I must be missing something here...

 

[sag2]

Well the adding a resistor to the sense circuit sounded too good to be true, but I thought I would try it. I went to Radio Shack and purchased a potentiometer assortment, and gave it a try. I first cut the sense wire and added 10 ohms, but it did not make any difference in the sense voltage. Then I switched to a 250 ohm pot, and it only slightly changed the voltage. Then I went to 1000 ohms and as I raised the resistance, the signal voltage went up not down. So I thought to myself, self, what's with that?

Well the circuit is a little more complex than I remember from training on three wire sensor circuits. I remembered that the PCM's have an internal voltage sense circuit that has a resistor built in. So I probed the ECM side of the sense circuit and yes it had 5v coming out. So by adding the resistance to the sense wire, you are impeding the sense from going to the ground circuit, and it results in raising the voltage at the ECM, not lowering it.

So then I tried putting the resistance in the 5v feed side, but at about 1000 ohms, it takes the 5v circuit out, and that is a shared circuit with all the other sensors. Truck died in a hurry.

So it is just a little harder than it seemed after all. I know it can be done, because I tried the Dr. Performance box on my truck that goes in line to the pressure sensor, and it works great. Not sure what is inside that little box, but it can't be too much. Hopefully I will figure it out and save the $600 for the Dr. Performance.

 

[KBennett]

rbattelle,



Check this out: http://www.dataq.com/194.htm



I don't know what the software is like but it may get you started.

 

[DLeno]

I think you guys might be over analizing the pressure transducer - at least for the purpose of adding a fuel pressure gauge. You shouldn't make any assumptions about the internal impedance of the device, which may or may not be measureable with a passive device. The best approach, in my opinion, is to buffer the transducer output with a simple op amp circuit yielding sufficiently high input impedance to guarentee that you won't disturb the information sent to the ECM. Forget all passive measuring devices. You already have voltage calibration information -- so set the op-amp gain to whatever is convenient to drive the desired gauge device to the correct scale. maybe find a 0-3 mA meter and voila you have 0-30,000 psi scale.



on edit: incidently, the pressure-to-voltage transfer function of the sensor output is pretty linear - slope of . 000153 and y-intercept of . 5. That means 26,000 psi is right at 4 volts and "full scale" (30,000 psi) is 5. 1 volts. So with an op amp with a gain of 1, you only need a 0-3 mA meter from the likes of . these guys

and a 1. 69K 1% resistor and you're set. full scale (5v sensor output) is now equal to 3 mA (with the 1. 69K resistor in series with the meter movement) and you don't have to do any changes to the meter face.





raising rail pressure to me is more complex than adding a single passive device. Like boonsur, I would question the "pressure fooling" approach which basically lies to the ECM in hopes that the ECM will compensate by raising pressure. Its certainly doable, I would imagine, by a simple op am circuit designed with whatever transfer function you want to insert in between the ECM and the pressure sensor (which could be as simple as a gain of less than one). the problem with this approach is that it introduces error in the feedback loop -- meaning that the ECM itself is still in control of fuel pressure, but buffered from reality, so to speak. I'm not sure that is a good approach.



I don't know how the better pressure boxes work, but suspect that they control the HPCR directly (in the same way the ECM does) and then send a bogus signal back to the ECM. This way, the ECM is not in direct control over pressure, so there is no loop error. The box itself decides what pressure to dial up, perhaps based on boost readings from the MAP sensor. just a couple of thoughts. I don't have any first hand knowlege of how the boxes actually work.



so if you're out to build a pressure box, you could do it all on the analog domain it seems to me. all you'd have to do is design your own mapping of manifold absolute pressure to desired fuel pressure, then build the analog circuitry that represents that transfer function. That way, as MAP increases, fuel pressure would increase. You could even make current fuel pressure and MAP both as inputs. At the same time, you would want to keep the ECM happy with both pressure readings and MAP readings -- both analog voltages

 

[rbattelle]

Originally posted by DLeno

I think you guys might be over analizing the pressure transducer - at least for the purpose of adding a fuel pressure gauge. You shouldn't make any assumptions about the internal impedance of the device, which may or may not be measureable with a passive device. The best approach, in my opinion, is to buffer the transducer output with a simple op amp circuit yielding sufficiently high input impedance to guarentee that you won't disturb the information sent to the ECM. Forget all passive measuring devices. You already have voltage calibration information -- so set the op-amp gain to whatever is convenient to drive the desired gauge device to the correct scale. maybe find a 0-3 mA meter and voila you have 0-30,000 psi scale.

Sag2 and Dleno, excellent posts! Dleno, I had considered adding a small op-amp circuit, then reconsidered and decided it would be simpler to just buy a 0-5v voltmeter and wire it directly to the pressure transducer. What is the advantage to converting the signal to current? You must know something about electronics that I'm missing. I gotta check the impedence on this panel meter I bought (when I get it). My data aquisition experience does involve pressure transducers. In that case, I was measuing very tiny voltage fluctuations from the transducer (on the order of 0. 02 mV), amplifying them as much as 1000 times, and sampling at around 200 kHz.



KBennett - you gotta wonder what a $25 data aquisition system is! On the other hand, if I bought it and it was junk I'd only be out $25.

 

[KBennett]

Yes, you do have to wonder about the quality of a system that costs $25. They are trying to gain market share by offering a low cost version of their product. I received a free version about two years ago but I ended up not using it on the project. I should fire it up just to see how well it works. I want to connect it to my Palm and write my own driver.

The cool part about the Palm is that it is an embedded system. I am looking for one of the newer models on Ebay with the trans-reflective color displays so I can read it in direct sunlight.



I wish you luck on building your own op-amp circuit. Watch out for ground loops, long leads, conflicting power supplies, etc.

 

[DLeno]

hey rbattelle,



If the guage you use is passive analog then you are already converting voltage to current, as all analog (passive) voltmeters are really measuring current and just calibrated in volts. Indeed, the 0-3 mA meter I suggested could just the same be calibrated in volts. it would be a 0-5 voltmeter with an impedance of 1690 ohms. The simpson meter, at 1K/volt input impedance, would have a meter movement of 1 mA full scale on the 5v scale and have an input impedance of 5,000 ohms. that worries me, as the general rule of thumb is that the transducer output impedance has to be less than 1/10th of this -- or 500 ohms. better yet, 1/100 th of this, which is 50 ohms, and that doesn't sound likely to me (a 50 ohm output impedance for the transducer)



The motivation behind my design is that you don't have to know anything about the transducer itself, and you can (with the op amp) present essentially infinite input impedance to the pressure transducer with very little part count. My design goals were:



1. passive, analog meter off the shelf with no calibration changes

2. very high input impedance (10s of megohms are quite doable)

3. low cost (the panel meter is ~$20)

4. treet the pressure transducer as a black box without understanding it.



If your 5v meter presents a 5K input impedance, then you have to start worying about the output impedance of the transducer itself and its effect on what the ECM sees. you dont' want small errors, you want zero errors. The thing about after market engineering is that you must absolutely guarentee zero effect on the stock circuit. For this reason I agree with Jnutter -- you need a very high imput impedance meter. I would choose not to load the transducer with 5K ohms, prefering instead at least a couple of megohms and (better yet) over 10 megohms. That approach removes all doubt, and frees you from having to research the design of the pressure transducer.

 

[silver03]

regardless of the meter impedence, all a DMM does is measure voltage. in the ohms position, it applies a small current and measures voltage. if the circuit is on (truck is running) current is all ready flowing, and will mess up the reading not to mention possibly effect the PCM's reading.





the sensor is an active type, so it does not fit in the strain or potentiometer category.

 

[DLeno]

silver03 -- I can't figure out what your point is or what issue you are addressing. A DMM is certainly an appropriate instrument to measure pressure transducer output while the truck is running. no one has advocated making a resistance measurement of an active circuit that is powered up, and yes its true that DMMs measure voltage.

 

[rbattelle]

Originally posted by silver03

regardless of the meter impedence, all a DMM does is measure voltage. in the ohms position, it applies a small current and measures voltage. if the circuit is on (truck is running) current is all ready flowing, and will mess up the reading not to mention possibly effect the PCM's reading.





the sensor is an active type, so it does not fit in the strain or potentiometer category.

I, too, am confused by this. What do you mean "active type"... I've never heard that term associated with pressure measurement devices.



DLeno, many thanks for the electronics education, I understand where you're coming from, now, and I agree with you 100%. Your idea is very sane, and I think I'll try to cancel my order for that particular Simpson meter and build an op-amp. I've used these before:



Burr-Brown OP-Amp OPA2604



Which has an input impedence of 10 GIGA-Ohms!!!!! I think I'll use that one!!

 

[DLeno]

yikes, 10 Gigohms. that should do it! main thing is that you've used it before and can vouch for its stabillity, and that it works with single ended power supplies, and as a non-inverting amp configured as a buffer with a gain of 1, etc. this should be an interesting project. Ever since dan. poitras posted the calibrations I thought wow, what a worthwhile endeavor. I pasted that data into excel and run regression analysis on it, and found very linear behavior.



keep us posted on your result. I kind of liked the small horizontal panel meter that might fit in the ash tray . you just need to find a good power supply for the op amp -- even 12V would be fine, which would give you the bonus of not touching the ECM 5 volt power supply or corrupting it in any way. some coax or shielded twisted pair cable for the input, a small project box, and you're on your way. you could even limit the output swing to 6v or so with a cheep little zener diode -- protection for the panel meter if things go crazy.



another thought on the meter itself. I'd like to have a dual pressure meter that displayed both the high side and the low side pressure. check with Westach -- they can silk screen anything you want, and I would expect they could build up a dual movement where one side is 0-5 volts calibrated to 0-30,000 psi, and the other side is their own low pressure fuel gauge. reasonable prices too.

 

[KBennett]

silver03 has a valid point on the active circuit.

Definition of an active circuit: An electronic circuit which uses active devices such as transistors or integrated circuits for its operation and which requires a power source for operation.



In this scenario the pressure transducer may have an amplifier built into it that sends the signal to the ECM. If this is the case the impedance of the meter may not matter as long as the amplifier can handle the added load. This is what happened to sag2 when he added the resistor, the amplifier overcame the load and the ECM saw no difference.



rbattelle, you can put your meter right on the output of the pressure sensor if it can overcome the additional load. If I am calculating this correctly with a 1K/volt meter and a 0-5 volt signal the meter will draw 0-5mA. Very little load on most active devices.

 

[DLeno]

A good way to test this would be to hang a load onto the pressure transducer output and calculate the load impedance. If the ECM can be removed from the circuit in such a way as to prevent compensation for the reduced voltage output, then such as procedure would be quite valuable. then make sure the voltmeter is a minimum of 10 times this figure.



But without knowing, the only way you can guarentee no influence is to buffer the output. this also creates additional protection in case something failed in the meter circuit, a wire shorted, etc.



Even if the tranducer includes active components doesn't mean you want to load them down or touch a factory circuit in that manner. A voltage follower will prevent that, and is the simplest of circuits.

 

[silver03]

my point is a meter can not measure the resistance of the sensor while it is on. and for that matter, the resitance in the off position is probably invalid as well since it is not one. that point aside, the voltage follower circuit is the correct way to do this. the 5volts on the sensor is enough to power the circuit as well. all the boxes out there power themselves off this same circuit.

 

[DLeno]

Originally posted by silver03

my point is a meter can not measure the resistance of the sensor while it is on. .

not directly, no. In general, the source impedance of any active or powered device cannot be measured directly. But the source impedance the circuit can be calculated from voltage measurements made while the truck is running. Use the DMM to measure output w/o load, then with a load. use the delta to calculate source impedance.

 

[rbattelle]

Well, I called Dr. Performance to see if they'd sell me their splice connector for the sensor. Whomever I spoke with was not interested in any such transaction, and explained to me that their "mold cost $21,000" and unless I bought a large quantity then they couldn't sell me any. So let me get this straight... I can by your complete fueling box for a couple hundred, but you can't sell me just a connector from the kit for like $30?



Do the Edge, TST, and Banks boxes also have a splice connector for the fuel rail pressure sensor?