High-precision stiff-spring 3Dmice

If you have questions or comments concerning any non-support related 3Dconnexion topic, please use this forum.

Moderator: Moderators

Post Reply
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

High-precision stiff-spring 3Dmice

Post by Hypersonic »

I had this idea about a year ago, emailed it half a year ago, and it seems to have subsequently found its way into the circular file.
After having used 3dmice on and off for 15years or so, I feel that these are some important hardware enhancements.
3Dmice are currently great for 3D movement. With these enhancements, however, I believe they can go from great to fantastic!
Although this might not appeal to some, dismissing the idea of using anything more than subtle force on an input device, there may be many out there who might embrace the idea for its added benefits.

Here's the write-up detailing the proposed hardware enhancements http://sites.google.com/site/forceinput/
jwick
Moderator
Moderator
Posts: 3331
Joined: Wed Dec 20, 2006 2:25 pm
Location: USA
Contact:

Post by jwick »

Excellent analysis Hypersonic.
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

Post by Hypersonic »

Thanks for the feedback jwick!

I admit I'm a bit speculative as to how many more input levels can be gained from accessing a greater force/torque range.
But I think it's something worth researching!
It might end up being analogous to rod/cone density lessening as the distance from the fovea increases.
In this case as the force increases the number of force levels per newton that one's fingers can reliably keep steady might start decreasing.
But still, it might end up adding quite a bit more input levels one's fingers could provide to a force input device.
Similar to impedance matching, perhaps studies could be done to determine how much finger force people can exert for long periods of time on stiff springs.
Stiff springs that hardly move so very little work is performed from the force, so at least in theory fingers shouldn't tire quickly.

For those who prefer a light touch should have no fear of stiff springs, as the drivers could be set so that a light touch maxes out input,
and still get a good amount of input levels out of it (same as today's 3dmice.)
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

Test Data (crude)

Post by Hypersonic »

I've done some crude testing to get a rough ball-park sense of how much stiff springs might benefit the user. https://sites.google.com/site/3dmouse6d ... /test-data
I sense that human fingers can only provide hundreds rather than thousands of steady force levels. However, they can definitely provide a lot more than the loose spring Space Navigator can receive.

It would be easy to manufacture two different versions, where the only difference internally is the spring stiffness (the stiff spring version wouldn't even require a heavy base.) The only difference externally would be a desk-clamp and a cap with finger-grips (particularly for Rx) for the stiff spring version.

Loose spring:
-have to move far to achieve target force (time consuming)
-snaps back slowly (time consuming)
-not much counter force feedback due to low forces (how much am i inputing?)

Stiff spring:
-don't have to move far to achieve target force (not time consuming)
-snaps back quickly (not time consuming)
-excellent counter force feedback due to higher forces. (I can feel how much I'm inputting.)
Even if the user only uses 3.4 Newtons on stiff springs they still get the benefit of saving time getting to their target forces.

The loose spring version could be for users who don't want to deal with clamping to the desk (not that clamping is a big deal really) and don't mind taking more time to get to target forces.
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

Test Data on a Space Navigator

Post by Hypersonic »

I've created a test program to measure human finger steadiness using a Space Navigator. Data is % dwell time, each session is the result of at least 10,000 samples.

First I placed an object on the device (perfectly steady) to detect unsteadiness due to sensor noise:
2.48 , 15.53 , 41.97 , 38.99 , 0.86
Reported input teetering between 2 values for the most part, only ~17% of the time not in the two teetering values

Then I made 3 attempts to keep the controller finger steady (I removed values < 1%, analysis rounds the values)

attempt 1
1.52 , 3.48 , 6.38 , 8.48 , 11.88 , 16.42 , 15.95 , 14.00 , 11.04 , 6.14 , 2.58
8 values were dwelled on for at least 06% of the time
5 values were dwelled on for at least 09% of the time
4 values were dwelled on for at least 12% of the time

attempt 2
1.13 , 2.44 , 4.17 , 7.01 , 9.16 , 10.39 , 13.41 , 12.47 , 11.82 , 10.49 , 7.32 , 5.14 , 3.06 , 1.17
8 values were dwelled on for at least 06% of the time
6 values were dwelled on for at least 09% of the time
3 values were dwelled on for at least 12% of the time

attempt 3
1.00 , 1.52 , 1.80 , 2.47 , 4.14 , 6.56 , 8.19 , 11.61 , 12.63 , 12.61 , 11.63 , 9.54 , 6.16 , 3.85 , 2.43 , 1.58
8 values were dwelled on for at least 06% of the time
5 values were dwelled on for at least 09% of the time
4 values were dwelled on for at least 12% of the time

Overall spread
8 to 8 values were dwelled on for at least 06% of the time
5 to 6 values were dwelled on for at least 09% of the time
3 to 4 values were dwelled on for at least 12% of the time

Conclusion
Measuring with 9.8 milli-newton precision seems to be overkill for Human fingers (which is the precision of the Space Navigator.) It seems that increasing the force range can be done without humans noticing any precision degradation in the 0 to 3.4 Newton force range (that of the Space Navigator) as trying to keep it finger steady seems to result in bouncing around a 50 milli-newton range anyhow. The Space Navigator can measure ~350 levels. Rather than use these levels for un-needed low-force precision, why not use them to increase force-range instead? Having higher-precision on the position sensors (hence more levels) would allow even less over-all distance traveled (throw) to move to the target forces. However, 350 levels appears to be fine for a 1.6 milli-meter throw (that of the Space Navigator.)

It would be nice to run this program on Space Navigators with various spring constants, as well as with many different people to determine a sweet spot for the spring constant.
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

Stiff springs are less tiring than loose springs

Post by Hypersonic »

Stiff springs less tiring? Are you crazy?! Understandable initial reaction. To understand why this is true consider a key factor of stiffer springs:

For any given target force, you move a stiff spring through a shorter distance than a loose spring to achieve the same target force.

Hence you use less energy to reach that target force due to work(energy) = force x distance. Why would anyone want to expend more energy to accomplish the same task? Just because a stiff spring 3DMouse is capable of detecting up to say 20 Newtons doesn't mean everyone is going to set the full 20 Newtons for max input in the settings.

Analogy:

Lifting packages:
Imagine lifting packages up onto a conveyor belt 1.6 meters higher when they can accomplish the same task lifting packages onto a conveyor belt just 0.274 meters higher? It takes more energy to lift them that extra 1.3 meters, the weight (resistive force) of the package is the same either way.

Moving a 3DMouse:
Why would someone want to push 3.4 Newtons through 1.6 milli-meters when they can accomplish the same task pushing 3.4 Newtons through 0.274 milli-meters? It takes more energy to push that extra 1.3 milli-meters! Sure only milli-meters, but these are mere fingers doing this potentially thousands of times per day, which adds up!

Note about the analogy:
The force is constant on the package lifting example. Although the force isn't constant in the 3Dmouse example (it ramps up) the average force is the same for both the loose and stiff springs for the same force range.

Nerve impulses cause muscles to contract. The stronger the impulse the stronger the contraction (force.) Force detection is the quickest method of collecting input (without somehow collecting input directly from specific nerve impulses.) Waiting for the force to move a spring through an extended distance is not only wasting time, but it is wasting energy as well!
Hypersonic
Posts: 265
Joined: Mon Jul 12, 2010 5:58 pm

More tiring even just keeping steady

Post by Hypersonic »

Loose springs are not just more tiring moving to new target forces, but loose springs are also more tiring when just trying to keep steady at one target force. This is due to the slight unsteadiness of human nerves/muscles causing slight jittery motions. Jittering the same force range on loose springs results in more expended energy than on stiff springs due to the fact that loose springs move more distance per newton of force. Although the impact of this might not be as dramatic as with moving to new target forces.

For anyone skeptical about loose springs expending more energy here's an example using 2 different methods to calculate energy expended. Both methods yield the same results:

Method 1: Potential energy for linear spring formula
http://en.wikipedia.org/wiki/Potential_ ... ear_spring

3.43 Newton Space Navigator
x = 0.0016 Meters (0 to 3.43 newtons)
k = 2143 Newtons/Meter = 3.43 Newtons / 0.0016 meters
u = 1/2 * 2143 * 0.0016^2 = 0.00274304 Joules = 2.7 milli-joules

20 Newton Space Navigator
x = 0.000274 Meters (0 to 3.43 newtons)
k = 12500 Newtons/Meter = 20 Newtons / 0.0016 meters
u = 1/2 * 12500 * 0.000274^2 = 0.000469225 Joules = 0.47 milli-joules (less than 1/5 of 2.7 milli-joules)

Method 2: Average force times distance formula
http://en.wikipedia.org/wiki/Work_(physics)

3.43 Newton Space Navigator
x = 0.0016 Meters (0 to 3.43 newtons)
force(avg) = 1.715 newtons
u = 0.0016 meters * 1.715 newtons = 0.002744 Joules = 2.7 milli-joules

20 Newton Space Navigator
x = 0.000274 Meters (0 to 3.43 newtons)
force(avg) = 1.715 newtons
u = 0.000274 meters * 1.715 newtons = 0.000469 Joules = 0.47 milli-joules (less than 1/5 of 2.7 milli-joules)
Post Reply