Devising a test to catch instability

[dpatrick911]

I understand the reasoning for wanting the "entire" track to move. I wonder for this test if allowing the CM to move isn't complicating things. See the attached link. Actually it is a fascinating video. https://www.youtube.com/watch?v=DY3LYQv22qY

If the test cycle we are talking about is a short cycle. The presence of movement should be easily observed at the front end of the car (I am thinking lever/moment arm affect). The CM becomes the fulcrum. If my understanding of this is wrong I am okay with learning. For simplicity, a fixed/stationary rail seems like it would be much easier. A dovetail could be milled across the center rail that the pin could be installed in. this would give you lateral movement of the pin and CM if needed. I am still thinking that the front end of the car would be presenting movement well before that force acted upon the CM to move it and that you would have conclusive evidence that the car was not stable without needing to allow the CM to move laterally. Again, maybe I am oversimplifying this. But if my understanding is correct the CM will want to follow a smooth parabolic curve regardless of what the front or rear of the car is doing. Maybe I am thinking on too big of a scale and not considering that the CM could/would move along much smaller parabolic curves. If that is the case my question would be, would it be visible given the max width of the car and the width of the rail? It is good to note that if the front of the car, moment/lever arm, is moving about the fulcrum that the movement on the back of the car, effort arm, would also be present.

Wayne Schmidt does have a nice "tread mill" he devised shown in a link from the OP a few replies back. I wonder if he still has this device and if he would be willing to run this experiment and record it for us? The other thought I had, and I don't know if it is useful or not, is with a "fixed" rail you could introduce some torsional load (not sure if this is the right term) by turning the rail slightly to be angled relative to the direction of travel. Not really sure if this is reality or if there is any useful data that would come from doing that.

[Stan Pope]

There are two instability modes of potential interest:
(1) rotation/oscillation about vertical axis. This happens typically when the DFW encounters a rail defect and is knocked sideways.
(2) rotation/oscillation about longitudinal axis. I don't know cause, but the back end seems to rotate within the limits of play between the axles & bores.

I think that the first is pretty well understood and effectively controlled by toe-in. In any case it could be investigated with either fixed or mobile CM.

l don't understand the cause of the latter well enough to know if limiting the freedom of the CM would change the process.

One of the values of running on a drum rather than a real track is that track anomalies can be introduced and the effect repeated once (or n times) per revolution.

Regarding the radius computation ... I repeated the derivation and am more confident in the approx 5' diameter. I'd still like someone else to sanity check the result. If the wheelbase requirement were reduced to 4-3/8", the diameter reduces to about 4'. If the full 3/8" - 1/4" = 0.125" difference were used instead of 0.100", the diameter reduces to about 4'2". And if both were applied, the diameter reduces to 39" (3'3").
Rather than risk dragging the belly on the rail, I'd aim for 0.120" (diameter = 3'4" for a 4-3/8" wheelbase or 4'4" for a 5" wheelbase.)

If you want the OpenOffice spreadsheet and text files for those computations, email me ... with my main computer down, updates to online stuff are difficult.

[Vitamin K]

There are two instability modes of potential interest:
(1) rotation/oscillation about vertical axis. This happens typically when the DFW encounters a rail defect and is knocked sideways.
(2) rotation/oscillation about longitudinal axis. I don't know cause, but the back end seems to rotate within the limits of play between the axles & bores.

I think that the first is pretty well understood and effectively controlled by toe-in. In any case it could be investigated with either fixed or mobile CM.

l don't understand the cause of the latter well enough to know if limiting the freedom of the CM would change the process.

One of the values of running on a drum rather than a real track is that track anomalies can be introduced and the effect repeated once (or n times) per revolution.

Regarding the radius computation ... I repeated the derivation and am more confident in the approx 5' diameter. I'd still like someone else to sanity check the result. If the wheelbase requirement were reduced to 4-3/8", the diameter reduces to about 4'. If the full 3/8" - 1/4" = 0.125" difference were used instead of 0.100", the diameter reduces to about 4'2". And if both were applied, the diameter reduces to 39" (3'3").
Rather than risk dragging the belly on the rail, I'd aim for 0.120" (diameter = 3'4" for a 4-3/8" wheelbase or 4'4" for a 5" wheelbase.)

If you want the OpenOffice spreadsheet and text files for those computations, email me ... with my main computer down, updates to online stuff are difficult.

Is calculating the required radius different from determining the height of a given arc for the circle?

Using the calculator here: http://www.mathopenref.com/sagitta.html, it says that a chord with a length of 5" on a circle of an 18" diameter will produce an arc of ~.35" in height. This is a little more palatable than having to construct a 60" drum, but I think the construction challenge is still significant.

[Stan Pope]

Is calculating the required radius different from determining the height of a given arc for the circle?

Using the calculator here: http://www.mathopenref.com/sagitta.html, it says that a chord with a length of 5" on a circle of an 18" diameter will produce an arc of ~.35" in height. This is a little more palatable than having to construct a 60" drum, but I think the construction challenge is still significant.

Same problem! Good find!

Keep in mind that the rail starts 0.25" above the running surface, meaning with very large radius there is only 0.125" clearance! So, work the height at 0.100" to 0.120" height!

[Stan Pope]

Is calculating the required radius different from determining the height of a given arc for the circle?

Using the calculator here: http://www.mathopenref.com/sagitta.html, it says that a chord with a length of 5" on a circle of an 18" diameter will produce an arc of ~.35" in height. This is a little more palatable than having to construct a 60" drum, but I think the construction challenge is still significant.

Same problem! Good find!

Keep in mind that the rail starts 0.25" above the running surface, meaning with very large radius there is only 0.125" clearance! So, work the height at 0.100" to 0.120" height!

Filling my numbers in, the sagitta web page gave the same result! Thanks for the sanity check!

Terry (Teeeman) reminded me that this is an approximation since the wheelbase (axle to axle distance) is measured between axles and the wheels contact the drum less than 5" (wheelbase) apart. That is because the axle to track distance is measured along radials of the drum! I thihk the difference is very small, but need to confirm that assumption.

[Stan Pope]

I have another approach that I think might be even better to analyze "rear end action" because constructing even a 3' diameter barrel of sufficient exactitude would push my ability (well beyond my limits.) That approach is to construct a 1' to 2' circumference barrel on which only the rear wheels ride. The front is held in place by it's DFW axle. That axle is subject to synchronized disturbances, e.g. 1/32" shift at x degrees with y acceleration. An LED flash marks the introduction of the disturbance. A camera mounted behind the car records the rear-end action and LED flashes. The barrel can also be adjusted to turn at specific linear speeds, e.g. 9 mph, 10 mph, 11 mph, 12 mph ...

Since the role of the guide rail is to influence the DFW, there is no need to include a guide rail!

I think that is simpler to build and simpler to analyze since the DFW disturbance can be introduced with a period that is unrelated to the drum rotation speed an circumference. The camera recording can pack many repetitions into a minute or two of identical disturbances.

I think that I could "turn" a barrel like this with negligible run-out! Might not even need to have the rear wheels (or the drum) turning to do maningful analysis!

"DFW Disturbance" controlled by a solenoid with varying stroke length and acceleration modified by a spring/weight linkage.

[Vitamin K]

I have another approach that I think might be even better to analyze "rear end action" because constructing even a 3' diameter barrel of sufficient exactitude would push my ability (well beyond my limits.) That approach is to construct a 1' to 2' circumference barrel on which only the rear wheels ride. The front is held in place by it's DFW axle. That axle is subject to synchronized disturbances, e.g. 1/32" shift at x degrees with y acceleration. An LED flash marks the introduction of the disturbance. A camera mounted behind the car records the rear-end action and LED flashes. The barrel can also be adjusted to turn at specific linear speeds, e.g. 9 mph, 10 mph, 11 mph, 12 mph ...

Since the role of the guide rail is to influence the DFW, there is no need to include a guide rail!

I think that is simpler to build and simpler to analyze since the DFW disturbance can be introduced with a period that is unrelated to the drum rotation speed an circumference. The camera recording can pack many repetitions into a minute or two of identical disturbances.

I think that I could "turn" a barrel like this with negligible run-out! Might not even need to have the rear wheels (or the drum) turning to do maningful analysis!

"DFW Disturbance" controlled by a solenoid with varying stroke length and acceleration modified by a spring/weight linkage.

So...this is interesting, and might have potentially useful applications. However, I wonder if you're not testing for something slightly different than my original aim was when I proposed the test. My goal was to be able to take a fully assembled car and be able to test it for waggle-inclination (assuming that we didn't have an actual track to test on), and then adjust the steer-in until we had found the "sweet spot" where waggle was controlled, but we weren't oversteering.

Your test, with the DFW removed, and the wheels possible stationary, seems more like to test the inherent stability of a given design...which could have some value (especially if we examine if things like ballast height affect stability, or perhaps alternative weight placement like Puma-style weights), but (correct me if I'm wrong), it doesn't really offer us tuning advice for overcoming potential waggle, does it?

[Stan Pope]

Yes, you are right! This latter scheme is oriented at a more fundamental understanding of factors involved in rear-end "activity."

For zeroing in on "best" DFW toe-in it would not do anything.

Also, there are two rear-end issues with which I am concerned. The first is clearly affected by DFW toe-in adjustment ... that is the car "coming loose" from its smoothe rolling mode.

The other is the tendency of some cars to develop "roll" ... oscillation about its longitudinal axis. I see this from time to time, too, and it can be an "energy eater" even if the car is holding the rail nicely.

[Stan Pope]

I don't think that toe-in setting would be a good application for anything other than a real track. The reason is that the setting is highly dependent on the details of the track with some tracks needing more toe-in than others.

However, there may be some car issues that affect the amount of toe-in needed. For instance, you may know that your build should run well on a smooth track with 2" in 8' of drift. Perhaps a track analog (e.g. treadmill, drum, etc) can show that a specific build is not stable (will not hold the rail) with that drift, and that other causes should be sought.

I had originally excluded toe-in setting from my thinking because of its track dependency, but now I see the relevance.

On the subject of "roll stability" ... since the rear axles are intended to ride in the bottoms of the rear wheel bores, it is possible for the rear to rock side to side by binding first the tops of the right side of the bores and then by binding the tops of the left side of the bores. Since neither is a stable state, the car will return, and possibly overshoot, to the bottoms of the bores. If overshoot, then there is a distinct "roll" motion. I can see a lateral impulse to the DFW initiating such a roll. I can also see unequal loading of the rears contributing to that kind of roll.