IR tachometer response time?

I'm using these for IR break beam setup to count the revolutions of a shaft. I have a disk on the shaft with 4 evenly spaced slots (about 5mm wide) to connect the beam. My problem is that the rpm spikes at times. The shaft at full speed is spinning about 5000 and the IR detector and emitter are roughly 5/8" from the centerline of the shaft. The emitter and receiver are roughly 5/8" apart.

The signal seems good then I see a spike to 20x the actual rpm then it comes right back. Even at slower speeds like 1000rpm or so it still does this.

Is it possible the IR cannot react fast enough at that speed? When I thought of using IR I never looked to see if it was capable of this kind of speed.

The device has some vibration which may cause some problems, but at under 1000 rpm the vibration isn't noticeable.

Using an UNO with the layout that's on this page. I'm not sending to an LCD, I have a windows program that reads the signal from pin 2.

Thanks for any help.

How far from center is the 5mm gap?, and for how long time does the receiver get light?
Rise time is 10us.

If you mean from center of the shaft, it's about 7/16" and goes all the way to the OD of the disk.

Haven't calculated that yet, but I will.

Where did you find the rise time? I couldn't find anything, but then again I'm electrically challenged.

Thanks.

Using a slot width of 4.3mm (I think this is closer to true size, but I don't have the disk in front of me so I can't be sure) and 5000rpm's, by my calculations it takes the slot .0006 seconds (600us if I'm correct) to rotate 20.5º.

The 20.5º is what I calculate the arc swing of the slot to be. This is guessing on receiver distance from center of shaft and actual slot width.

I'm sure there is some misalignment between the transmitter and receiver which would reduce the time that the receiver has to detect the IR beam even more.

Knowing this would I be better off making 2 opposing slots wider and blocking the other 2? Don't think I really need all 4 slots for my application as I don't need 1/4 of a revolution accuracy. I can't widen just 1 slot because it may cause more of an imbalance.

And then there is my vibration problem which could be moving the Rx and Tx because they are soldered to my breadboard and are roughly 10mm from the board and them skinny legs aren't very strong.

Is there a method to get alignment between the Tx and Rx better than eyeballing?

Thanks.

I found rise/fall times in the datasheet. 600us should be enough time to get a good reading.
If possible, paint the slots surfaces black (non reflective)