How would you measure pedal kickback?

You have to measure at the crank as opposed to at the cassette because we aren't sure how much is just the chain flapping about and not PK at all.

what's the SI unit for measuring kickback? 

Just plug it into Linkage like a lazy person lol. Cute graph and everything.

instead of measuring forces, why not displacement at the pedal spindle (rotational/angular distance)? that's essentially the riders contact point with the system. 

Here's an angle nobody has mentioned yet: isn't pedal kickback a little bit like free acceleration? 

Suspension compresses = chain "tightens" = assuming no movement from the rider's foot = cassette turns forward. 

I’m not sure pedal kickback force is the right metric for quantifying the rider experience.

I think frequency or duration might be a better metric to correlate to what the rider feels on trail.

High speed video would be my first choice to see what’s going on. Find a section of trail or a set of stairs or something repeatable and see what’s happening. 

Why not ask the manufacturers for their pedal-kickback values? Or do you actually want to measure it via instrumentation?

Would need a standardized model. What style of suspension, how much travel, length of crank arm, gearing...

Also PBK is typically communicated in degrees so easiest to do this in the stand is: remove shock, lock the brake up, level(ish) the crank arm and preload it by hanging a weight off it, attach an inclinometer to the crankarm and record the difference between starting and ending angles. 

PBK is dependent on the exact instance that the chain speed exceeds the wheel/FH speed (meaning full FH engagement) and does not include chain forces caused by bumps. Also note that the above PBK method doesn't exact take into account the full range of degrees if the values drop mid-travel cycling. You'd want to observe the live readout to see if there's a peak value before the end.

the main items to standardize would be ring and cog size used. Crank arm length would always yield the same degree value HOWEVER shorter cranks will travel a smaller distance vertically but a longer distance horizontally. I'm not sure how important that horizontal measurement is in terms of feel, but that vertical distance sure is. If someone is running a 155mm vs 175mm crank arm on a bike with 25 degrees of PBK, that's about 9mm difference vertically.

Have you considered reaching out to Matt at Cass Labs? Last I saw him he was running an ochain and said he was hoping to do a follow-up to that video you posted. Pretty approachable guy. 

In my mind I'm not sure if your 'you can only choose one' directive could work, I don't see how you could isolate the cause to be pedal kickback rather than some other input without multiple measurements. 

My 'other' suggestion would be weight sensors on the pedals, looking for an abrupt increase in weight on the leading foot and a decrease in weight on the trailing foot. This is what I think my experience of PK is, and is the thing I'd be hoping to reduce with one of these devices. 

I wonder if the Garmin pedals would give you the readouts you are looking for.

The simplest thing I can think of would be to just put a powermeter on the cranks.  Pedal kickback is essentially just torque on the cranks, which is exactly what a powermeter measures.  

I imagine a crank-based strain gauge or power meter would be your best bet, they normally use something like an IMU to detect if the crank is rotating so you could separate out pedalling related torque from pedal-kickback. Most power meter products transmit data at a fairly low sample rate so you might need something a bit more custom which would be easier to combine with braking and suspension data at a higher logging rate too. Guys like BrakeAce (their technology is basically a powermeter) or syn.bike might be best to supply something. 

The challenge with that would be filtering the noise. 

Angular speed vs shaft speed, but I watched that video too! You'd want at least a button, or some kind of event for your "human" feel of pedal kickback to correlate to see what you see in graphs. 

If I understand correctly what the wish is here, I'd say somehow measure the rotation of the cassette and compare it to the rotation of the chainring st the same time (acquire rotational data for each and just do a subtraction later). Maybe add rear wheel and crank rotation as well. 

How to measure it? Maybe in a similar way to how abs sensors work, maybe something with a Tacho laser counting how many teeth passed, though this would probably not be the best solution as the resolution would be too low. 

A rotary encoder with a good enough resolution with a 3D printed tooth negative mounted on the frame interfacing with the cassette and the chainring where the chain is not jn contact with it? 

You’re not gonna be able to simulate pedal kick under heavy braking on linkage

I think, more to the point, the challenge would be bypassing the filtering that the power meters already do, because they're designed to report specifically the power you're putting into the crankshaft.  PK isn't a power input and I expect they also do something to minimise extreme transients in the signal.

I think you'll need to take measurements of multiple parts of the drivetrain and then develop a theory that links them all together.

Please in "Fries needed to refuel the energyloss"

It’s easy to use a shock position sensor and wheel speed sensor to determine if an impact could have potentially cause kickback if you know what gear you are in. I’d just pick a gear and be done with it to avoid having to also log cassette position. Actually measuring kick back would certainly have a lot more noise to sort through. A strain gauge on the chain ring would tell you if the chain is tugging the chain ring. With ochain you’d need another strain gauge downstream of that to see what torque is getting passed through to your feet. These would both need to be calibrated, which could be done easy enough by hanging weights from the pedal or putting a torque wrench on the crank spindle. Strain gauge on the crank arm itself would introduce a lot of noise related to how you let your legs handle an impact. 

Wheel speed sensor and shock position sensor are not enough as you don't cover the chain slapping around. I really think the rotations should be covered. 

That’s just to determine if a kick back event was possible. Kickback and chain slap are different. You won’t want to use a position sensor on the cranks because it’s possible to have zero crank movement and feel pedal kick back. 

Could you do an accelerometer on an ochain? Knowing the mass and radius of the chainring, you’d be able to calculate the peak force being applied from the chain with the peak acceleration of the chainring. You’d be able to convert that to force at the feet with crank length too. 

call me silly but, the only time I've ever felt kickback was when my chain was too short (I used to set my chains as short as, for weight) an hit a little drop in the large cog. My own fault for bad set up an bad gear selection.

Learned my lesson, I run a couple extra links an never felt kick back since.

So to measure the symptoms not the cause (the debate of kickback vs chainslap can carry on in the background), you want to measure the force the front foot is experiencing from 'kickback' whilst removing the other forces from the equation (pedalling and normal force).

You'd need to isolate the force at the pedal (the front foot pedal), from the power through the cranks AND the normal force from both pedals supporting the riders body weight.

So theoretically:

A set of Garmin pedals (you'd want to see the raw data from the force transducer, which I assume they use?)

A crank based power meter

An an accelerometer near the BB

Should give you everything you need! Haven't got enough time to do the maths, but that's my first thoughts

And closely followed by second thoughts. @TEAMROBOT your idea of the strain gauge on top length of chain is great. I can't picture achieving that practically, but if you can it negates the need for the Garmin pedals and the accelerator.

But I think you'd still want a crankbased powermeter to try to remove pedalling forces from the equation, and even if your setup stops the rider pedalling for the experiment entirely you may as well have the data to compare with tension in the chain

if you really wanna get good benchmark results, you need to isolate as many variables as possible.  if you have the means, you might want to make a test apparatus to generate lab data, then compare it against the real world data.