Speed reading mystery

[Phoenix]

Hello All,

The speedometer in my 73.5 911t has been reading 10-15% slower than actual speed according to several iPhone speed apps. After sending to Palo Alto Speedo and their running several bench tests as well as looking for obvious issues inside, they see no problems. Bench tests show an accuracy of w/in 1mph across the full range. I did have NHS rebuild this a few years ago and PAS confirms it looks well rebuilt.

The mystery:
if the Speedo is accurate on the bench, why is the reading off vs several speed apps? (Apps used: GPS Speed, Speedometer Simple, Speedometer One Speed Tracker, GPS Digital Speed Tracker and Speedometer)

Anyone have any thoughts?


A few facts:
- Original tire/ wheel configuration: 185 70 R15 (Vredestein Sprint Classics)
- Rims 5.5j x 15
- Rebuilt original 915 tranny; now with a LSD
- App speed rating: (gauge reading = app reading)
30 = 33
35 = 40
40 = 45
50 = 57
60 = 70
70 = 81

[Magyar Kiwi]

Hi
Could you please clarify; what are your current tire and rim sizes ?
Many thanks

[Walt Fricke]

Well, either you have a taller ring and pinion, or taller tires. This mechanical speedo reads pinion shaft revolutions, mechanically the same as later electronic speedos measure differential revolutions.
The '73 came with the 7:31 915. Later 915s had 8:31. Changing a ring and pinion is an involved (and thus expensive) proposition. After a rough calculation, that would about cover the discrepancy unless I have it backward.

You might write down what your rpms are at 50 and 60. I assume those were in 5th, or you can make them be. That would make for an easier check of ring and pinion.

Also useful would be what Vredestein thinks the diameter or circumference of its tires are. However, for what it is worth, perusing old PCRs suggests that the tires which came on the 5.5" x 15 rims in that era were 165s.

[tsuter]

Have you had transmission work where someone mucked with the reed switch?

[David Inc.]

10-15% would be a hell of a tire difference.

[dtxscott]

Does your car have an impulse sender from the trans to speedo? Perhaps it is dirty and not sending full signal to the speedo?

[dannobee]

31/7 = 4.4286 31/8 = 3.875

3.875/4.4286 = 0.875

(30+35+40+50+60+70) / (33+40+45+57+70+81) = 0.874

I think Walt nailed it.

[Phoenix]

Thank you for all the great input!

The car currently has 185 70 R15 which, as near as I can tell is the original size. (original owner's manual lists 165 and 185 as options. I suspect the 165s came with the steel wheels and the 185s with the Fuchs which the COA list mine as coming with.) The speedo is mechanical with no sensors that I'm aware of.

Palo Alto Speedometer indicated the speedo test was done using a "speedometer ration of W=800" (i have no idea what that means.)

I did have the transmission rebuilt many years ago and added a LSD though nothing else was to change. The speedo seemed fine after that.

If it is a ring and pinion issue, is there anything I can do short of rebuilding the tranny back to original? (e.g. Can the Speedo internal gears be changed to adjust?) Or is this a situation where I need to live with it reading slower than actual speed?

[thetorch]

I believe W=800 means 800 revolutions per mile.

Sounds like you have an 8:31 R&P with a 7:31 speedo per Walt.

Call Hartmut. To recalibrate my same setup he had me perform a rolling test of a measured 1/100 mile counting the exact cable revolutions at the speedo. This takes into account all the variables: tire and wheel size, R&P, speedo gearing in the nose cone.

I believe my count was slightly over 7.0 or W=700.

Then he replaced the gear in the speedo. Reads perfectly now.

Edited: 1/100 not 1/10 mistake fixed.

[Phoenix]

@thetorch - palo alto had me do 3 rotation tests prior to sending them the unit. (# of cable revolutions over 52' 9.5"). Your explanation clarifies what this test measures - thank you.
Results of each test on my car were 1) 7.75, 2) 7.75, 3) 8.0. This seems to match their W=800 bench test calibration. If I understand this correctly, the Speedo was accurate on the bench w/ W=800 and should then be reading accurately on my car which is also W=800. Is this correct?
If so, I'm starting to question the accuracy of the Speed Apps.

I'm also wondering if there could be an issue with the cable itself. Is it possible that it works fine at very low speeds (as during the rotation test) though have issues at higher driving speeds? If it's easy to replace, I may just try it. I'm already in the hole ca $150 for testing the speedo (shipping and test) and would hate to walk away still having an issue.

[thetorch]

@Phoenix sorry I posted 1/10 instead of 1/100th...fixed it and the rotation number for my test. So you did the same test and it sounds like your speedo is well calibrated.

The apps are generally accurate +/- 1-2mph when held constant speed, with a slight delay in my experience. Hard to see how a cable would cause this.

I am stumped. I think a call or email to Hartmut is still a good idea.

Hard to believe it's not the speedo, but I am willing to learn something new here. Keep us posted.

[dannobee]

Replace the two speedo drive gears with those out of an 8:31 tranny.

See if California Motorsports or G-Box has a set.

[Walt Fricke]

It might be a good idea to see just what differential you have.

The first thing to do is to see what the transmission serial number tells you. Various sources provide this information. So do you have the transmission the car came with from Porsche. If the number corresponds with an 8:31 model, problem solved.

If not, one way to check this is to jack up the rear,, and measure the rotation of a tire against rotation of the engine. Since you have an LSD, both tires should move the same amount, which simplifies this. Here are how many degrees of wheel rotation versus exactly one crank rotation:

7:31 3d ratio 1.26 64.5*
4th 0.96 84.5*
1st 3.18 25.5*
8:31 3d ratio 1.26 73.7*
4th 1.00 92.9*
1st 3.18 29.2*

You should be able to figure out a way to measure degrees. An inexpensive digital angle measurer taped or bungeed to a rim should work. Two people useful for this - one to turn the crank, and one to hold back pressure on the tire to remove spline/tooth slack before starting, and to keep that pressure on until the one rev mark comes up again on the crank pulley. Dealing with this slop is critical to accuracy (which should be within one or two degrees.

When I was using this method to see if race cars had the right transmission I had spread sheets for all this. If you use 1st gear as your test, there will be no ambiguity as to what the gear ratio might be, since all 1sts for the 915 are the same (absent aftermarket gears used for racing). But 1st has a much smaller degree difference.

You could use a degree wheel and a pointer on a jack or other stand. You can download and print a degree wheel used for cams.

[7783911]

did you contact Palo Alto, they could adjust the speedo for you as they did for me by providing indicated versus actual speed, tire size etc. They were very responsive in my case and my speedo is 100% accurate now

[Phoenix]

Thank you for your continued support. I'm getting more and more confused.

Do the cable rotations I measured not indicate an 8:31 model/ R&P and the W=800 bench test performed is for 8:31?

@Walt - the car is garaged for winter and I can't lift nor drive it until April.

@the torch- who is Hartmuth? I would very much appreciate talking this through with someone whose been down this road.

If the Speedo and tranny are mismatched, on which do I need to change gears? If on the tranny, does this require its removal? (Not sure the ROI will be worth this.)

@7783911 - I'm not sure what you're getting at here. The speedo is on its way back from PA as, according to their tests, there is nothing wrong with it. Are you suggesting the reposition the needle to make up for the difference? If so, would the the difference not need to be linear for this to work? Also, would the odometer not still be off?

[Walt Fricke]

Oliver
Let's review a few facts:
1) The car is a 1973T.
2) It came with a transmission with a 7:31 ring and pinion.
3) The transmission has a mechanical speedometer sender.
4) Pere the parts book, that transmission came with senser gears, both the output gear and the gear attached to the pinion shaft, for the 7:31.
5) It appears that the 915s with cable speedo drives came with speedo drive gears for both the 7:31 and the 8:31.
6) You can't change the tranny speedo drive gears without removing the transmission and taking off the nose piece. This is within the competence of a DIY guy - remove the transmission from the car, unbolt the nose piece and slide it off, change the speedo gear pair, and put everything back together. No finicky adjusting of anything. The R&R would consume the most time of doing this. Still, not trivial like setting ignition timing or changing brake pads.
7) You have used multiple methods to show that the speedometer reads slow despite the fact that the speedo itself is working fine - it is reading what it is supposed to read with a 7:31 R&P.
8) Assuming the rolling circumference of your rear tires is what Porsche set the speedometer for originally, the most likely cause of the mismatch is that your car has an 8:31 ring and pinion. You don't know how that could have happened.
9) You can't check the transmission numbers stamped on its underside until the spring, nor can you do one of several tests to confirm what the R&P ratio really is until then.

At this point, seems the only thing you could do toward fixing things is to send the speedo back to the shop and ask them to recalibrate as if for an 8:31. However, since you won't drive it until the spring, and an accurate speedo isn't essential to driving around, why not wait until you can do more checks?

However, this will mean waiting to get it back on the road in the spring with a speedo, as the shop recalibration isn't an overnight job what with shipping etc.

The one thing I could think to do now is to ask the shop if they can adjust a mechanical speedometer made to run with a 7:31 R&P to run accurately with an 8:31.

The only cable speedometer I have ever opened was on an old VW, where the cable spun a magnet, which caused drag on a steel cup which surrounded it and exerted drag against a spiral spring which retarded it. All you could adjust was the spring's tension. I suspect your speedo translates physical rotation into either an analog or a digital electrical signal, and some means (resistance or capacitance or inductance) allows adjustable components to do the changing.

Moving the attachment of the needle doesn't sound like a good idea. The guys who have written about working on speedometers all say this tricky and that it is easy to break the thin rod the needle attaches to.

Good luck. But if you ever get it fixed, report back.

[7783911]

It was not a repositioning of the needle, it was a recalibration by PA to match rolling diameter and recorded versus actual speed

[Walt Fricke]

Oops - I thought Oliver or someone wondered if just repositioning the needle would be a fix. I must have been thinking of someone else's issues and suggestions.

I calculated how much taller a tire would be needed to have the speedometer read 7 mph high at 50 mph in 4th with a 7:31 R&P. 185/70-15 works out to a diameter of 25.197" To get to 57 mph with just a tire diameter difference would require a 29" tire. That seems unlikely to be what was on this car when the data were recorded.

To Oliver the mystery may be how the transmission ended up with an 8:31, if that indeed is the case. I don't recall that he gave the car's history, other than at some point it went into a shop to have an LSD installed. Maybe a prior owner had some transmission work done, or the transmission replaced because of some serious malfunction at some point? When did he get the car, and has it always had the speedometer read high like this?

[thetorch]

Hartmut is the son of the founder or PA Speedo and runs the business now. He's very helpful. His email is the one on the website. Or call him.

There is a lot of discussion of R&P, nose gears, wheels, and the like, but the none of that matters if the OP properly did the calibration test that PA Speedo requested -- the calibration on the ground at the wheels takes all of that into account. It simply relates distance (1/100th mile) to revolutions of the speedo cable at the back of the gauge.

It appears he did that.

The only question I see is why the speedo isn't translating that rotation count into the right speed. Which it seems like only Hartmut can answer, since PA Speedo did the calibration.

Or I am missing something fundamental, which certainly wouldn't be the first time.

[Phoenix]

Thank you Walt and torch. I will call Hartmuth.

@ Walt...the background... In terms of the tranny... many years ago I had the engine and tranny rebuilt adding LSD by the now infamous Motormeister. The engine story was not pretty (you can search for it here). In essence they not only "replaced" my 7R case with a later one, but the engine they rebuilt ended up catching fire (hence the Phoenix name). I never had a really cause to question the tranny work as it shifted smoothly, was quite and has had no issues. I did notice fuel economy drop though suspected that was due to the carbs. Speedo was always a few miles off though didn't think is was an issue. After my recent rebuild to 2.9 w/ EFI, I expected better fuel economy as we worked to fine tune the mapping to achieve this. Despite the mapping work, I am not really seeing improvement. This caused me to question the odometer which is when I started seriously cross-checking with speed apps. It may well now be coming to light that perhaps MM did not only "replace" my 7R case with something different, but the tranny as well. (I was to naive to capture the tranny number before sending it.)

All this still does not address the issue thetorch clarifies in that the cable rotation measured should be what PA measured against and is showing an accurate gauge. This is what I hope to clarify with Hartmuth.

the mystery continues... I'll post what I learn