Suspension strain gauge loads vs. time

[DaveW]

I still think there is useful things to be learned on the original question via accelerometers on the uprights even without the rest of the standard array of gear.

If they are positioned carefully, calibrated accurately, coupled with potentiometers and processed sensibly, then I agree absolutely. Most of the caveats would apply equally to all measurements...

[Tim.Wright]

I still think there is useful things to be learned on the original question via accelerometers on the uprights even without the rest of the standard array of gear.

If they are positioned carefully, calibrated accurately, coupled with potentiometers and processed sensibly, then I agree absolutely. Most of the caveats would apply equally to all measurements...

OK, but if all you wanted was for example the push rod or shock absorber force, then all you need is damper displacement and a knowledge of the spring and damper curves.

However, if you want to somehow quantify the damper effects on the contact patch force, then I think hub accelerometers will help.

[gixxer_drew]

I still think there is useful things to be learned on the original question via accelerometers on the uprights even without the rest of the standard array of gear.

If they are positioned carefully, calibrated accurately, coupled with potentiometers and processed sensibly, then I agree absolutely. Most of the caveats would apply equally to all measurements...

OK, but if all you wanted was for example the push rod or shock absorber force, then all you need is damper displacement and a knowledge of the spring and damper curves.

However, if you want to somehow quantify the damper effects on the contact patch force, then I think hub accelerometers will help.

Experienced people are usually cheaper than sensors. Too bad I can never convince team managers about that.

[Jersey Tom]

Experienced people are usually cheaper than sensors. Too bad I can never convince team managers about that.

Experienced people are a double edged sword. Some are "experienced" but still full of ----.

[gixxer_drew]

Experienced people are usually cheaper than sensors. Too bad I can never convince team managers about that.

Experienced people are a double edged sword. Some are "experienced" but still full of ----.

haha good point, I should have phrased that "good people".

[riff_raff]

......OK, but if all you wanted was for example the push rod or shock absorber force, then all you need is damper displacement and a knowledge of the spring and damper curves.....However, if you want to somehow quantify the damper effects on the contact patch force, then I think hub accelerometers will help.

If you want to measure the forces acting on the pushrod, then you would strain gauge the pushrod itself. If you wanted to measure the forces acting on the spring/dampener, then you would strain gauge the lower attachment of the dampener body. This would take into account all of the linkage mechanical frictions, hysteresis effects, and dampener hydraulic flow losses, all of which can be quite variable at any given instant, and thus would be difficult to quantify accurately from a simple performance curve. As an example, you could have up to a total of 13 rotational joints in each suspension linkage (3 per A-arm, 2 per tie rod, 2 per pushrod, 2 per dampener, and 1 per rocker pivot), with each one producing an end-moment friction loss equivalent to 1/4 of the applied radial load. When you sum them all up, that can result in quite a bit of friction dampening between the upright and dampener attachment.

I would imagine that the best way to measure the instantaneous force acting upon each tire contact would be to use some type of non-contacting displacement sensor at each corner that records the relative distance to the track surface. This would give the data that is of most concern, which is the collective roll, bump and pitch motions produced at each tire contact.

[Tim.Wright]

If you want to measure the forces acting on the pushrod, then you would strain gauge the pushrod itself. If you wanted to measure the forces acting on the spring/dampener, then you would strain gauge the lower attachment of the dampener body.

Yes, this would be the most accurate way and I think what was asked for in the original post. I only mentioned the damper pot method after people started talking about accelerometers which I don't think is a good way to measure the either the vertical force or spring force (unless you put the car on a 4 post rig with an iterative algorithm like RPC).

Regarding the friction I agree somewhat. But even putting strain gauges on the spring will not be measuring the friction in the control arms. This is still an unknown as it would be when you use damper position to calculate your spring forces.

[Jersey Tom]

Depends on the resolution of data you need, and the practicality of it all.

If it were me, in most open wheel applications I'd probably just go about it by creating math channels based on damper travels without direct force measurements. That's the cheapest and easiest solution by far, and probably gets you most if not all of the answer you need.

If there's a little extra cash around, then I'd probably pick up some of these or these or something similar. Off the shelf components all the way, rather than trying to manually strain gauge components. Saves having to rate each spring, ARB, and damper curve.

Beyond that.. I don't know how much more resolution you need. At some point you start chasing minutiae and there are more value-added things you could be doing with the time.

[DaveW]

Cheating a little bit, and using the results obtained from a rig test of a real vehicle, chosen more or less at random, this is comparison of the estimated damper trajectories using position sensors (shown in red) and the difference between hub and body accelerometers (shown in green). Hopefully, you might be convinced that the two trajectories are fairly similar, and the effect of linkage friction is relatively small (It would be apparent mostly at low velocities).

To obtain the trajectories, I had to double integrate and scale the accelerometer difference measurements to obtain "wheel" displacements, scale the position measurements by the motion ratio (to convert "damper" to "wheel" displacements), and differentiate both displacements to obtain velocity. The load was obtained by subtracting sprung mass times hub acceleration from contact patch loads.

I had to do one further operation on the accelerometer difference measurements, because they can be (are) effected by the displacement of a notional spring in series with the damper. To take account of that, the measured displacements are modified by subtracting Load/Kis before the final differentiation (Kis is the estimate of installation stiffness). In this case, the value used for Kis (1919 N/mm) is shown in the legend. The effect usually cannot be neglected - here is the same plot before correction.

To complete the picture, here is the load/deflection trajectories fully corrected, and here is the same thing before correction.

To be fair, track results wouldn't be better than this, but I would feel fairly comfortable in using the idea as a basis for track measurements. However, the ubiquitous Peak Load vs Peak Velocity plots generated by damper dynoes would not represent an adequate model, and the effects of a spring in series with the damper would be essential, in my view.

[GSpeedR]

Dave, it appears these plots are one cycle at roughly 4.5Hz; I would imagine that the coherence of your accelerometers drops significantly below 2-3Hz(?) and might make this procedure less accurate at lower freq responses. You've shown how powerful such techniques are, though if a person requires accurate average (DC) loads in certain components then strain gauges are probably a more reliable option in my opinion.

That said, I wouldn't discount position based measurements; we should remember that though we consider load cells to measure 'force', fundamentally it is a strain measurement which is really a displacement. Mechanical sensors usually (always?) measure motion.

[Tommy Cookers]

That said, I wouldn't discount position based measurements; we should remember that though we consider load cells to measure 'force', fundamentally it is a strain measurement which is really a displacement. Mechanical sensors usually (always?) measure motion.

there is something wrong if the load cell compliance is not insignificant relative to the compliance of the system being studied

[DaveW]

... these plots are one cycle at roughly 4.5Hz; I would imagine that the coherence of your accelerometers drops significantly below 2-3Hz(?) and might make this procedure less accurate at lower freq responses.

Quite right. I was trying to suggest that all measurements are important when trying to understand how a vehicle responds.

... if a person requires accurate average (DC) loads in certain components then strain gauges are probably a more reliable option in my opinion.

I agree. I would add that PRL's are very useful measurement channels but, unless your interest is only PRL, they are just additional information. It is possible to use a multi-post rig to calibrate them as vertical load, but when multiple loads are applied or geometry is changed, etc., the calibration may not be valid.

... Mechanical sensors usually (always?) measure motion.

Often, but not always (e.g. "Force Balance" & certain "Coriolis Effect" transducers). It is usually helpful to understand exactly what they are attempting to measure.

[DaveW]

there is something wrong if the load cell compliance is not insignificant relative to the compliance of the system being studied

Heisenberg's Uncertainty Principle....

[Tim.Wright]

Hopefully, you might be convinced that the two trajectories are fairly similar, and the effect of linkage friction is relatively small (It would be apparent mostly at low velocities).

Certainly an eye opener. One point though;

The load was obtained by subtracting sprung mass times hub acceleration from contact patch loads.

So you need the contact patch load from the rig to calculate the spring/damper force? Did I understand that correct?

I assume also there is some sort of drift correction in the accelerometer data?

Tim

[DaveW]

So you need the contact patch load from the rig to calculate the spring/damper force? Did I understand that correct?

Yes.. but the point is that having at least a reasonable estimate of the spring and damper characteristics, I should then be able to predict force using only measurements of damper position & accelerometer signals.

I assume also there is some sort of drift correction in the accelerometer data?

Double integration requires two constants of integration to be estimated. For the example shown, the constants are assumed to be correct if two successive trajectories over-lay (more or less).