AdrianH
|
| posted on 1/5/09 at 08:59 PM |
|
|
How to measure twist in the propshaft whilst turning?
This is just an idea I am playing with some electronics.
I want to measure the twist accurately in the prop shaft whilst the car is moving and subject to acceleration, breaking etc.
The movement of the gearbox and therefore one end of the shaft is getting me stumped. Under acceleration the shaft will be defected upwards or is it
down? so any optical sensor placed on the tube will have to deal with that movement.
If I have an optical sensor on the output of the gearbox UJ and another on the differential UJ then I do not know how any play in the joints will
effect the readings.
Any programmers or electro mechanical engineers on here?
Stumped at the moment
Adrian
[Edited on 1-5-09 by AdrianH]
Why do I have to make the tools to finish the job? More time then money.
|
|
|
|
|
nasty-bob
|
| posted on 1/5/09 at 09:11 PM |
|
|
could you not do it with strain gauges?
rob
|
|
|
AdrianH
|
| posted on 1/5/09 at 09:39 PM |
|
|
Not sure how strain gauges work, assumed they are like weighing scales as you make a piece of crystal deflect, it gives a bit of a voltage. If I used
a strain gauge then I would have to deal with rotating wires as the prop rotates. So trying to do it without being connected to the prop.
If I can figure it out then I could, possibly use it as a type of dynomometer built into the car.
With a fairly simple processor I should be able to work out timing changes between two sensors at either end of the prop shaft. From this measure the
twist in the shaft for any rotational speed.
The twist in the tube gives force rotational speed with the force will give power. Or summat like, I hope.
If clever, the unit could self calibrate at low revolutions with the car free wheeling and therefore no tension/twist in the shaft.
Thats the theory, I do some work with micro controllers but it would be easier to do with a PC I bet.
Adrian
[Edited on 1-5-09 by AdrianH]
Why do I have to make the tools to finish the job? More time then money.
|
|
|
tegwin
|
| posted on 1/5/09 at 09:49 PM |
|
|
Hmm.. would be very hard to get an accurate figure on a car.. there are far too many variables...
Would be easier on a test rig.... apply the same torque that the engine can produce on a dyno to the shaft and measure the axial displacement when it
is under duress..... Not sure how you would impart that much torque on it without some big gear....
Failing that.... there is some maths that would tell you theoretially what the twist would be for a given cross section etc..
------------------------------------------------------------------------------------------------------------------------
Would the last person who leaves the country please switch off the lights and close the door!
www.verticalhorizonsmedia.tv
|
|
|
MautoK
|
| posted on 1/5/09 at 09:57 PM |
|
|
Adrian,
Are you specifically looking for dynamic response or do you just need to know the torsional stiffness of the propshaft?
If the former, good luck! Sounds like an interesting project.
If the latter then you can calculate it, given metal spec, ID, OD and length; or build a static torque rig and measure it.
John.
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
AdrianH
|
| posted on 1/5/09 at 10:22 PM |
|
|
I accept it would be difficult to get very accurate figures on a car, but I assume it would be OK to get comparisons after tweaks and flat out runs up
to 60 or 70 MPH.
John, to get a dynamic response will be beyond me and my programming skills, but if I can record data, and feed at a later stage into a PC or if I
hone C programming do it on a PC.
Knowing what the prop is made from I can work out torsional stiffness. I would like to think that North West Propshafts could tell me the spec on the
tube they use.
What I was thinking was this, open to all for ideas or putting into practice!
Two solid disks mounted on the shaft as close to each end as possible, each disk has one slot in it to allow light through from a LED to its detector.
So as the shaft turns each detector gets one pulse, off to on and off again per revolution.
The slots do not have to be in line, in fact mounted approx 120 degrees apart with the front leading is probably better.
Just by using the pulses, I think you can determine, using maths a few things.
1) shaft revs and therefore engine revs for any given gear.
2) If calibrated whilst free running, you can work out the angle between the two pulses, the shaft will be under no load, this will be a function of
starting a timer from the first pulse and stopping the timer on the second pulse, and a function of revs. I think it is a linear formula.
Once the base line is set, then any lengthening of the time between the pulses is a measure of the twist in the shaft.
The processor should very easily count up one every 1/1000000 second
It will take me some time to get the programming correct and by that time I should have my car in a state to try it. Just need to work out how to fit
the sensors to cope with the prop shaft moving up down or sideways in the tunnel.
I wonder about reflective tape on the shaft and the opto sensors mounted to the side ?
Adrian
[Edited on 1-5-09 by AdrianH]
Why do I have to make the tools to finish the job? More time then money.
|
|
|
MautoK
|
| posted on 1/5/09 at 10:44 PM |
|
|
Adrian,
I would have thought that a propshaft would only twist a degree or so under the torque available from a typical engine.
Say 150 ft-lb from the engine and 4:1 first gear ratio gives worst case 600 ft-lb torque.
I recall a formula with (od^4 -id^4) and length and modulus from Mech Eng lectures at University (but that was over 35 years ago 
It's worth doing some ballpark calcs to see what kind of resolution you'd be looking at for any instrumentation setup.
John.
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
AdrianH
|
| posted on 1/5/09 at 11:03 PM |
|
|
Good Idea John, It was several years ago for me also as in 30 plus, but I need to keep the grey stuff working.
I will do some maths over the next day.
Engine torque 147 Nm at 3000 rpm. what the hell does a Newton Metre relate to!
A mere 108 ftlbs
Adrian
Why do I have to make the tools to finish the job? More time then money.
|
|
|
craig1410
|
| posted on 1/5/09 at 11:56 PM |
|
|
Hi,
Can't you just place optical discs between the flanges of the propshaft and gearbox/diff and attach the actual sensors to the gearbox/diff
casings. There should then be no relative movement between sensor and disc at each end regardless of gearbox movement.
Another option might be to fire a solid state laser from gearbox to diff with the optical sensor on the diff. You could then place discs anywhere
along the propshaft which would cut the beam. If you make the discs with an appropriate relative phasing then you should be able to identify the
individual discs and use the rising (or falling) edge of each to work out relative propshaft twist.
I hope this helps,
Craig.
|
|
|
richardlee237
|
| posted on 2/5/09 at 05:52 AM |
|
|
It may be possible to do it by mounting a toothed disc on each end of the prop shaft. Use a probe at each end of the shaft to detect the teeth.
Use an oscilloscope to align the 2 pulse streams at idle and then monitor the phase shift under acceleration.
Quote Lord Kelvin
“Large increases in cost with questionable increases in performance can be tolerated only in race horses and women.”
Quote Richard Lee
"and cars"
|
|
|
britishtrident
|
| posted on 2/5/09 at 06:46 AM |
|
|
A pair of toothed wheels is a very accurate system widely used in ships but almost impossible to use on hookes jointed shaft. Straingauges with
connections via slip rings are likewise accurate widely used but the same problems arise ---- using some form of radio tag for data transmision as
used on some tyre pressure warning systems might work.
A much easier method of torque/power measurement is simply to measure the torque reactions at the engine mounts and signal process the output.
[Edited on 2/5/09 by britishtrident]
[I] “ What use our work, Bennet, if we cannot care for those we love? .”
― From BBC TV/Amazon's Ripper Street.
[/I]
|
|
|
sickbag
|
| posted on 2/5/09 at 07:48 AM |
|
|
Howabout a disk at each end with a single thin slot in each disk, use this to allow a LED to shine through to a light sensor. Align them up so the are
perfectly in phase. Then measure the time difference between each pulse to work out the amount of twist.
Finally back on the job!
|
|
|
Dangle_kt
|
| posted on 2/5/09 at 08:09 AM |
|
|
you hold one end, your mate holds the other and then try and give it a chinese burn.
Results will only be in thumb widths as you won't have any spare hands - but it might be a start.
I love reading what all you clever people can come up with, it reminds me that I'm just lightly dabbling in this car building game.
|
|
|
craig1410
|
| posted on 2/5/09 at 11:15 AM |
|
|
quote:
Originally posted by britishtrident
A much easier method of torque/power measurement is simply to measure the torque reactions at the engine mounts and signal process the output.
[Edited on 2/5/09 by britishtrident]
I was thinking this too although I think it might be easier to design a cradle for the diff to fit inside which could be designed with pivot points
placed in positions which would make direct correlation between force and torque. You could also design the cradle with the correct amount of
stiffness to give some deflection without causing the driveshafts any articulation problems. My issue with engine mounts is that the engine is
connected to the chassis loosely by 101 different cables and pipes which will all affect the readings to some extent. The diff is connected only by a
few direct mountings.
Another option is to read strain in the trailing arms with strain gauges. Since these are usually parallel and equal length with a panhard rod which
is usually laterally located the geometry of the forces should be easy to work with. The advantage of this is that it would allow you to measure
braking forces and lateral forces too.
I'm sure the F1 guys already do all of this sort of stuff on their cars.
Cheers,
Craig.
|
|
|
AdrianH
|
| posted on 2/5/09 at 05:03 PM |
|
|
Haynes Roadster so diff fixed to chassis, I rang NW Propshafts this morning and was asked to try again for technical stuff on Tuesday. The outer dia
is 51mm but the inside diameter is not known.
I guess if I want dyno type readings then I could just try and count revolutions over time, weigh the car before so I have a good idea of mass and
plot curves like with the GPS and pocket book software.
I will just have nto work out some maths and see if I can fabricate the sensor wheels and try it, nothing to loose but time.
Adrian
Why do I have to make the tools to finish the job? More time then money.
|
|
|
MautoK
|
| posted on 2/5/09 at 05:40 PM |
|
|
Adrian
I just did some sums using the standard equation
theta = TL/JG
where
T = Torque (in-lbs)
L = Length of propshaft (in inches)
G = Modulus of Rigidity
J = Polar Moment of Inertia
for a hollow shaft, J = (pi/32)(OD^4-ID^4)
So, let's say your prop length L is 24 inches long, 2.2 inch OD and 2 inch ID (0.1 inch wall)
For non-exotic steel, G = 12 X 10^6 lb /in^2 (approx)
Suppose the engine generates 150 ft-lbs of torque and first gear is 4:1. Therefore, T = 150 * 4 * 12 lb-ins = 7200 lb-ins worst case..
J = pi * (2.2 ^ 4 - 2.0 ^ 4) / 32 = 0.729 in^4
Everything is in units of pounds and inches or combinations thereof, so:
theta = TL/JG gives
theta = 7200 * 24 / ( 0.729 * 12 * 10^6 )
theta = 0.01975 radians or 1.13 degrees.
If you fit a pair of trigger-type wheels to the prop, say 4 inch diameter, the relative movement at the periphery of the wheels would be 0.0395 inch,
or almost exactly 1 millimetre.
Theta is directly proportional to torque.
In top gear (1:1) the propshaft will twist by 0.28 degree and the relative displacement at the rim of the trigger wheels would be ten thou or 1/4
mm.
HTH
John.
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
MautoK
|
| posted on 2/5/09 at 05:55 PM |
|
|
More...
Let's say engine speed is 3000 rpm, so in first the prop does 750 rpm.
The peripheral speed of the trigger wheel ( 4 inch dia ) would be 750 * pi * 4 / 60 inches per second or 157 inches per second
The 0.0395 inch twist displacement therefore takes
(0.0395 / 157) *1000 milliseconds or 0.25 ms to pass the detector.
So you'd need resolution in the order of 10 microseconds for 4% steps.
HTHA,
John.
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
MautoK
|
| posted on 2/5/09 at 06:25 PM |
|
|
...and in top gear, again 3000 rpm engine, but now 3000 rpm propshaft, the torque is less (from above) giving the relative displacement of about 0.01
inch.
Trigger wheel peripheral speed is 3000 * pi * 4 / 60 inches/sec or 628 inches/sec
In this case the 0.01 inch passes your detector in around 16 microseconds. You'd need sub-microsecond resolution....
Can I ask what is the primary aim of your experiment, by the way?
John.
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
richard thomas
|
| posted on 2/5/09 at 06:55 PM |
|
|
Sounds like a lot of work....have a can of lager instead!
|
|
|
MautoK
|
| posted on 2/5/09 at 07:03 PM |
|
|
Make mine a large glass of Merlot - oh look, there's one right here! Cheers! 
He's whittling on a piece of wood. I got a feeling that when he stops whittling, something's gonna happen. (OUATITW/Cheyenne)
|
|
|
Liam
|
| posted on 2/5/09 at 09:17 PM |
|
|
Just get one of those performance meter thingies that use accelerometers. Gonna be as accurate as anything else you can come up with.
Or the engine mount deflection idea is a good one since that's the largest torque-dependant displacement you can measure. Calibrate displcement
vs torque with a dyno run and bob's your uncle.
Oh and if any of your hoses/cabling are significantly resisting the movement of your engine, you've got problems! And it wouldn't effect
your your readings anyway if you were to calibrate displacement vs torque.
Liam
|
|
|
AdrianH
|
| posted on 3/5/09 at 08:02 AM |
|
|
quote:
Originally posted by MautoK
...and in top gear, again 3000 rpm engine, but now 3000 rpm propshaft, the torque is less (from above) giving the relative displacement of about 0.01
inch.
Trigger wheel peripheral speed is 3000 * pi * 4 / 60 inches/sec or 628 inches/sec
In this case the 0.01 inch passes your detector in around 16 microseconds. You'd need sub-microsecond resolution....
Can I ask what is the primary aim of your experiment, by the way?
John.
Just got back to this thread this morning and John many thanks for the maths. You have shown there is next to no time to take measurements and there
is not really enough accuracy to be of much use.
I did some basic calcs myself assuming it could deflect 1 degree at 5000 rpm on the prop and found that 1 degree would be 0.33 microsecond in duration
so even running a 10 meg sample rate I would not have enough pulses to have any accuracy. So your calcs have proved the same point.
As to why! I have a box of electronics I used on another project and to learn more I have to find tasks for it and me.
Atmega 128 web-server from tuxgraphics, has several analogue and digital inputs and I can write stuff to a simple LCD screen.
And building the car I want to know performance changes.
Yes I can buy bits but, would be good with diy solutions. Its an old age thing!
I did not like the idea of the engine mountings initially until I thought of using a torsion bar rather then rubber mountings, but still reckon the
pot holes and bumps in the road will deflect more then any engine torque.
Cheers
Adrian
Why do I have to make the tools to finish the job? More time then money.
|
|
|
craig1410
|
| posted on 3/5/09 at 09:27 AM |
|
|
just get yourself an iPhone. It comes fitted with three axis accelerometers, GPS and already has a number of in-car dyno applications written for it
which cost next to nothing and have been verified on the drag strip.
The one I use is PocketDyno+ which you can see at www.pocketdyno.com. Only costs $12.99 or equiv in GBP and is very good.
Cheers,
Craig.
|
|
|
