legacycentral bbs

OK, so it's known that I have no ability to hook up a mechanical speedometer to my new transmission.

Vikash came up with a good, simple way to fabricate a VSS signal, so the car won't be TOO fucked up.

But I want a speedometer. A postit note would be too inconvenient, espically since I intend to drive on the highway, and such for somewhat extended trips. I'd like to know how fast I am going.

So I thought. The speedometer uses a cable to spin. The tachometer doesn't. Newer cars with the electronic VSS don't use cables to spin the speedo either.

So how different are the tach drivers and the speedo driver?

Could a tach be suited to move based on a voltage? What kind of voltage range would there need to be in order for the tach to sweep it's range? Could a very simple circuit be built based on the VSS signal setup I'll have that would be able to convert that into a number the tach signal would be able to use?

How could I get it to work?

CN: Using a tach driver/motor/needle, but have it read in MPH.

Basically, I need it to read driveshaft RPM, but adjusted for wheel diameter and such to achieve actual speed.

Can it be done?

Tachometers count the positive ignition pulses...and since RPM and speed aren't really related, I doubt you could use it in place of VSS...unless you could somehow make it figure out what gear you are in, and then program it to know what speed is obtained at a certain RPM in each gear

Modern (non-cable-operated) speedometers are tachometers; they just have different scales.

So we'd just need to multiply or divide the frequency of the VSS signal, and level-shift it (from 0v/5v to 0v/12v).

So let's think... I'm guessing we'll want 100 RPM to correspond to 1 mph?

If I remember right, your VSS setup is now going to give you 4 pulses per revolution of the driveshaft. And you're running 225/45R17 tires? And a 3.900 final drive ratio.

So what frequency does it give you per mph?

[ (2* 225mm *45% / 25.4 mm/in + 17 inches ) * Pi ] / 3.900 driveshaft revs / 4 pulses/driveshaft rev = about 5 inches per pulse

63360 inches/mile / 5 inches per pulse / 60 min/hour = about 210 pulse/min per mph

Huh.

Huh!

Does our ECU's tach signal pulse once per revolution or once per ignition event? If it's the latter, you'd already be pretty close! And if it's the former, you'd just need to divide the frequency by two.

Ooh. Or you could just add another VSS setup. Put another magnetic switch further along the driveshaft, and put just two magnets on the shaft there. Connect one side of the switch to ground, and connect the other side directly to the tach and also to switched +12v through a 10K resistor.

You'd want to put a switch in there so you can toggle the tach between measuring RPM and measuring engine speed, I'd imagine.

Did I do that right?

BAC5.2 wrote:
Vikash came up with a good, simple way to fabricate a VSS signal, so the car won't be TOO fucked up.

I missed that one. I admit I haven't been around in awhile. My Subie has been in the driveway since August. I'll need to start using a car again soon though.

So what's the fix for the speedo? Does that mean the cruise control would work? I'd considered using a bicycle speedometer, but that would not help with cruise control. And, it would be difficult (unsafe) to try to read all the time.

V- Actually, I'd want to transfer the guts and gizmos of a tach, and use a speedometer face.

I could use another tach, as I have another one, and just remember that 6000 RPM on the tach is 60mph. That'd give me range up to 80 mph, which would be cool, but at the track, speeds will quickly exceed 80mph.

Could the tach signal be scaled further so that the needle would move to the marks on the speedo face? That'd be ideal.

The fix for a VSS (espically since it does not need to be exact), is to epoxy magnets to the driveshaft, and then use a magnet to pick up the magnetic field (creating a pulse). Wire that directly to the ECU, and I've got a signal that will substitute the VSS.

I'll be using both 205/55/16 and 225/45/17, both of which have an almost identical circumfrence. Close enough.

So the tach guts can be mounted where the speedo guts used to be?

Yeah, I think you should be able to scale it to match the frequency. I actually think I like the idea of just making a second VSS replacement to drive the tach... Can you figure out what the mapping would have to be between mph and RPM?

The tach picks up one pulse per revoloution of the engine, right? So if you had it pick up 1 pulse for every 4 revoloutions of the driveshaft, it would be very close to having it pick up one revoloution of the wheel itself, right?

As far as the guts go, I'm sure it could be made to work.

I don't know if it's one pulse per revolution or one pulse per ignition event (which would be two pulses per revolution).

You don't care how fast the wheels are spinning in RPM though, do you? You want to know linear velocity.

So we need driveshaft RPM @ X MPH right?

This should be about right....

I came up with that chart by the following.

(((2(225*.45))/25.4)+17)*Pi = 78.453 inches

78.453/63360 = .001238 miles

X RPM*60 = RPH
(8500)*60 = 510000

.001238 x RPH = Z
(.001238)*510000 = 631.38

Z/FD = Y
(631.38)/3.90 = 161.892

For the table above, I rounded through.

You can also circumvent all of this by simply doing the following:

RPM*Tire Diameter/Final Drive*336.1
(8500)*(((2(225*.45))/25.4)+17)/3.900*336.1

336.1 =(miles in inches/the minutes from RPM to express RPM in RPH)/pi

336.1 = (63360/60)/Pi

That kind of cuts out the middle man.

That last equation just simplifies down the whole process. You could also do it as this:

RPM*Tire Circumfrence/FD*.002975

.002975 = (pi/mile in inches)*(the minutes from RPM to express RPM in RPH)

.002975 = (pi/63360)*60

This is a somewhat awkward method of doing it, since it's repetitive, but it comes out the same. I prefered the first one.

If you want to find X speed at Y RPM in Z Gear, you simply change the denomenator to reflect the gear ratio in decimal form.

So, if I wanted to know my speed at 5000RPM, on 225/45/17's with a 6:4 transmission gear ratio, I'd divide it to decimal form (4/6 = .6666667, to be expressed as .667) and plug it in.

(5000)(24.97244)/(3.90*.667*336.1) = 142.8144mph

RPM is, technically engine RPM, but if you have a 1.00 transmission ratio, then your driveshaft speed is the same as engine RPM. Direct drive.

Was that what was needed Vikash?

holy shit my head hurts.


any luck getting it to work?

Haven't tried yet. I'll be working on the car over spring break. Tranny isn't even built yet, lol.

I expect good things. If this can be figured out, then it'd be very easy to make it so the ECU saw actual vehicle speed too!

Well, from the looks of your chart, you came up with the same results as I did for the first part, which was good.

The second part makes sense to me but I haven't double-checked it.

But that's not what I was asking for... What we need, I think, is a mapping from RPM to mph on the gauge face. That is to say, if you put a speedo face on the tach guts, what RPM will correspond to what mph?

Ooooohhhhh

Well, zero RPM = 0 MPH

8000 RPM = 140 MPH

ALMOST but not quite 2:1

http://www.thawa.net/bac52/91SS/91SS%20009%20small.jpg

The first 500 RPM and the first 10mph are, essentially, omitted from both gauges.

It looks as though it's a ratio of ABOUT .0175. That is, any given RPM x .0175 = speed on the speedo face. It's pretty dead on.

Here's the chart:

Oh, huh, I hadn't realized that the needle sweeps were almost identical.

Does anybody know what the frequency of the tachometer signal is? Is it one pulse per revolution or one pulse per spark?

I'm going to guess it's one per revolution, actually, because I don't think people who do H6 swaps need to change their tachometers...

Okay. Let's work with that assumption, and say that we need to convert 1 mph into 1/0.0175 = about 57.14 RPM.

Tire circumference = [ (2* 225mm *45% / 25.4 mm/in + 17 inches ) * Pi ] = about 78.45 inches

1 mi/hr / 60 min/hr * 63360 in/mi / 78.45 in/wheel rev / 3.900 gear ratio = about 11.79 RPM at the driveshaft

57.14 / 11.79 = about 4.8

Hmmm... Do you think maybe you could get away with putting five magnets evenly spaced on the driveshaft? At 55 mph, it'd read 57 mph. That's pretty close, and easily within the variance between tire brands. I think that's closer than the factory speedometer.

I could do 5 magnets evenly spaced.

I bet that'd be fine. JBWeld is pretty strong stuff, and I know how to evenly space things around a circle well enough.

Should I do that for BOTH the VSS AND the Speedotachometer?

Use two sets of magnets and two pickups so both signals are independent of eachother?

If it reads slightly faster than I am traveling, I wouldn't mind. Keep my butt out of trouble.

I see no compelling reason for there to be two sets of magnets. I would use one set of five magnets, but two pickups.

Each pickup would have one side grounded.

The VSS pickup would have the other wire go to the ECU's VSS input.

The speedotachometer pickup would have the other wire split and go two places: 1) directly to the speedotachometer, and 2) through a 10K resistor to an ignition-switched +12v source.

What's the purpose of the 10k resistor?

Wow, maybe I'll actually be able to use cruise control!

It's called a "pull-up" resistor. When the switch is not shorting the signal to ground, the resistor pulls it up to 12 volts. That makes it so the wire switches between 0V and 12V rather than between 0V and floating.

It'd be interesting to see how cruise works. The cruise computer (like the ECU) will see a vehicle speed significantly lower than the actual vehicle speed.

lol, hmmm.... good point....

The ECU sees 0-5v for the VSS signal, right? How could THAT signal be modified so that the VSS sees correct vehicle speed?

Is there a "pull-up" resistor that would allow that wire to switch between 0 and 5v?

Didn't we determine that if you put 16 magnets on the driveshaft the VSS would be about right?

The ECU has a pull-up resistor built into it for the VSS, that pulls it up to an internal 5-volt rail.

So to get a correct VSS signal, having 16 magnets is the only possibility?

No, but it's probably the most straightforward.

What other way would there be?

It seems that the there would be 1/3 as many pulses as required. If you tripled the number of pulses, it should read pretty close to real speed?

How could you do that without tripling the number of magnets?

In principle you could rig up a signal conditioning circuit of some type to multiply the frequency. It's not very simple though.