Sunbird Turbo Boost gauge interface analysis

[pmbrunelle]

For those who wish to interface with something besides the GM 2-bar MAP sensor that this gauge is supposed to work with, this post may be useful.

I traced out (visually) the portions of the gauge that I deemed relevant to someone wishing to interface with the boost gauge.

********************************************************************************

As for the tachometer, I'd expect that the tachometer could be calibrated in the usual Fiero manner, per the following Oliver Scholz page:
http://www.fieros.de/en/articles/tach.html

The white factory-trimmed resistor piece looks the same on the Sunbird gauge as it does on a Fiero gauge.

********************************************************************************

I may share my thoughts on the best way to interface this with a PWM output in a few days.

[The Dark Side of Will]

The boost gauge uses the same MAP sensor as the ECM? I would have thought even in the '80's GM would have a MAP output from the ECM to drive that guage.

I tried recalibrating The Mule's tach for an 8 cylinder and ended up lifting the trace[*] that leads to the resistor.... so I went BIG TACH instead. I might try again with a pot if someone wants to volunteer me a tach, but I have other things that are way higher priority in the WBS...

[*] And I'm not a buffoon with a soldering iron... except apparently this one time

[pmbrunelle]

From this write-up, it says that a GM 2-bar MAP sensor will work with the gauge:
http://www.fiero.com/forum/Archives/Arch ... 10302.html

Of course, what matters is the voltage(MAP) function.

I don't know how the gauge is presented with the correct voltage(MAP) function in the Sunbird.

The obvious solution is to split the 2-bar MAP signal wire in two; one wire towards the ECM, one wire towards the gauge.

It would be possible for the ECM to take the MAP sensor signal, buffer it, and send the buffered version out to the gauge, but I don't think there's much added value to doing it that way.

********************************************************************************

I appreciate the Fiero philosophy for gauges, with the dash instrumentation being divorced from the engine controls. Yet another reason why I like Fieros.

When I look at the dash gauges, I want to see the car's various parameters as they are (according to one version of the truth).

I don't want to see transformations and corruption caused by the ECU. If the ECU is misinterpreting a sensor and running the engine poorly as a result, I want to debug it with a scan tool. I don't want the erroneous value to be presented on the dash gauge. Two signal chains allow for me to see if both chains agree with each other.

[pmbrunelle]

Eric, can you find out what PWM frequencies you could choose from to drive this gauge?

I could check my own TunerStudio, but I don't know if you're running the same version 1.4.0 as I am.

[Emc209i]

Please don't bite off more than you can chew, darling.

Sub'd. I'm sure this is important to someone who can't afford a boost gauge. Don't screw it up.

[ericjon262]

Eric, can you find out what PWM frequencies you could choose from to drive this gauge?

I could check my own TunerStudio, but I don't know if you're running the same version 1.4.0 as I am.

I honestly haven't gotten quite that far into it yet, because I haven't applied power to the MS3 Pro, or microsquirt yet. I'm almost done wiring them but I'm waiting on the M12 connector for the CAN bus port so I can divorce the chassis harness from the engine harness more easily, and tunerstudio won't let me do much without it communicating with the MS3. I also need a new ampseal connector for the MS3 pro, one of the terminal locks got slightly off center and broke... if you didn't know, ampseal connectors are a royal PITA compared to most Dephi stuff.

Please don't bite off more than you can chew, darling.

Sub'd. I'm sure this is important to someone who can't afford a boost gauge. Don't screw it up.

I would like to do this because it should give me an acceptable reading, while maintaining a more or less stock appearance. to the untrained eye. the MS can be run in "Test mode" to test the individual outputs, and then the PWM signal adjusted to make the readout correct doing away with accuracy problems related to ground resistance and sending units.

That being said, I plan to investigate the possibility of developing an instrument cluster controller to control the whole cluster including some of the dummy lights instead of the Microsquirt, which doesn't have quite as much flexibility as I had originally thought due to a smaller number of I/O's available, but that's a little(long) while down the road.

[pmbrunelle]

I whipped up this drawing in LTspice. *.asc isn't an accepted filetype, so if anyone wants the native file, ask me for it.

*****Design priorities*****
Easy to build: few additional components, all located at the gauge
Easy to connect: single wire connection to PWM output
Response time: fast enough so the needle doesn't look slowed down
Accuracy: not an ultra-high priority (but still considered), since the gauge doesn't have ultra-precise graduations anyway.

*****A note on the schematic reference designators*****
Reference designators such as C6, D5, R11 represent footprints that already exist on the gauge PCB.
C6 and R11 in particular must be changed from their original values.

D5 remains stock.

Where there are three-digit reference designators (such as R101), these denote new parts to be added to the PCB. You are on your own to mount these parts in a suitable manner.

*****Schematic explanation*****

Over at the left we have an open-drain PWM output from some device. It is connected to the green wire.

D5 is the stock 5.1 V Zener diode. We will use its cathode to take some power for the pull-up resistor R101.

R101 will force the green wire to assume a potential of 5.1 V when the PWM output switch is open. The waveform on the green wire will be PWM, with the duty dependent on the duty of the switch.

R102, C101, R11, C6 form a 2nd-order low-pass filter. Rolloff is at 7.2 Hz. With 250 Hz PWM, that gives 1.5 decades of separation. With a 2nd-order filter, that should give 60 dB of attenuation.

The objective is to let the average voltage on the green wire to pass through, while blocking the PWM frequency. I don't know how the gauge would react to seeing unfiltered PWM at its input; I wouldn't want to try on a rare gauge.

Naturally, a low-pass filter will impact the response time. On the blue curve, from an initial voltage of 5.1 V, we see that the filtered signal settles to mid-scale in about 200 ms.

I could not meet both my attenuation and response time targets with a 1st-order low-pass. That is the reason for the parts-count increase.

Questions?

Ground offsets will have an influence on the readings in the high vacuum range. 0.1 V offset could result in 5 kPa error. At full boost the ground offset would have no effect. I could propose something that would eliminate the effect of ground offsets, but at the cost of build complexity. As things are, the modifications could be reasonably done with dead bug / point-to-point methods, with no additional PCB.

[pmbrunelle]

I'll add that I can come up with a suitable Digi-Key BOM shortly if desired.

[ericjon262]

you're clearly much more electrically inclined than me! lol! When I mentioned ground resistance, I was referring more to the other analog gauges on the Fiero cluster with divorced sending units. running them PWM more or less eliminates ground resistance as an added variable, because the sensors receive power and ground via the PCM (or MS3 in my case) and aren't grounded via the engine block. if the ground is weak for the PCM, it then affects both the power to the sensor, and the ground from, so the overall result is still a more reliable signal compared to a stock setup IMO. Because the sunbird gauge uses the map sensor's output, its more or less the same scenario, so it should be a fairly stable signal within a wider range of ground resistance.

Am I crazy to thing all that?

[pmbrunelle]

Yes, if a voltage source and measurement are both grounded to the same "ground island", or domain, you'll have better results. The island can shift with respect to the vehicle, and items living on the same island won't see a difference.

If we assume the Sunbird setup is as follows:
Boost gauge: grounded via the gauge cluster to the unibody
MAP sensor: grounded to the ECM's ground plane

An accuracy problem could arise due to the offset between the ECM domain and the gauge cluster domain. Ideally, to reduce the offset between the two, the gauge cluster should be grounded directly to the engine block. Since the ECM is also grounded to the engine block, the two domains will remain close to each other.

You could ground the gauge cluster to the ECM, but then you would be injecting currents from the gauge cluster into the ECM's ground plane. Don't do that.

********************************************************************************

On my Fiero, I'm running both wires of the water level gauge (a 2-wire variable resistance output) back to the auxiliary gauges, so it can be grounded in "auxiliary gauge domain". That's an easy one.

My WBO2 output is grounded to the MS3, which makes for accurate computer readings. However, it is perhaps not ideal for an auxiliary AFR gauge which is located in "auxiliary gauge domain".

At some point, I will measure the voltage difference between MS3 ground and auxiliary gauge ground. If the difference would cause too much of a gauge reading error, I will use a differential amplifier in auxiliary gauge domain to generate an output voltage that is equal to the voltage difference between the AFR wire and the MS3 ground.

[ericjon262]

I think the percent error will be negligible, either way though. For the most part we're talking errors that are way outside of what we need the instruments for, when I drive I just look to see if it's about where it needs to be, if my coolant temp gauge is off by 5 degrees, it's not the end of the world as I'll tune it base on logs not observed reading on the gauge.

The ECU on the other hand, I want to have as accurate and repeatable of a reading as possible, because it's doing so much more than just observing data.

[The Dark Side of Will]

Right. If you take the car to the track, you need the gauge to tell you to pit because the car is overheating. Then you'll look at the ECM logs to determine how much, how long it took, what parts of the track had the most temp gain, etc.

[pmbrunelle]

Minimum requirements for C6, C101
Through-hole, axial-leaded
C6 has 0.021" lead diameter and fits in existing PCB holes
10V
X7R dielectric
1 µF, 20%

Part that meets or exceeds requirements:
Digi-Key 399-13735-1-ND

Minimum requirements for R11, R102
Through-hole 1/4 W
R11 has 0.023" lead diameter and fits in existing PCB holes
No particular temperature coefficient requirement (typical carbon film OK)
22 kΩ, 10%

Part that meets or exceeds requirements:
Digi-Key 22KQBK-ND

Minimum requirements for R101
Through-hole 1/4 W
No particular temperature coefficient requirement (typical carbon film OK)
1 kΩ, 10%

Part that meets or exceeds requirements:
Digi-Key 1.0KQBK-ND

********************************************************************************

This more or less sums up what I had to do with the gauge. I am ready to send this now.

If I ship this to your house and you happen to be at sea, is that OK?