Wanting to use a particle beam in an alignment tool. Will any commonly available radioisotopes give me a strong count if I made my slit very narrow? As I move my detector from left to right across the slit, I need to be able to pick up a very narrow Gaussian particle count distribution from a couple inches away. I need to be able to resolve the curve and tell where I am on it at increments of 0.03" or less, so I may need to make my collimator slit very narrow, perhaps 0.01" wide. Are there any commonly available radioisotopes that could give me strong beta particle counts of at least tens per second at that distance? Or are they all too weak? 
Edit- Follow up question: What sensor should I use if I plan to hook it up to a microcontroller?
 A: "Commercially available" radioactive sources... It really depends on whether you have a license (which depends on the country). Assuming you are in the US and have no NRC license, then you can legally have a sealed 10 µCi source (these exist in many fire alarms, for example). That will give you 37,000 decays per second, emitted into random directions. The question becomes - what solid angle of this radiation can you use?
Am-241 has an emission at about 59.5 keV, which is amenable to collimation with a few mm of tungsten. If you can afford slit collimation (only need to align in one direction at a time) and you want to be able to resolve 0.03" increments from 2" distance (or about 1° angular resolution) then you probably need a slit with that aspect ratio (that is, you first need to collimate the source down to a certain size, then have a slit to limit the beam) - because a "point source" of radiation is rarely a true point.
This diagram gives the idea:

Now the efficiency with which you are getting radiation to a detector (with a width S) depends on the other dimensions in this diagram. It will depend critically on the size of your source - if it's too big you need to have the first collimator present, and that will really hurt. Otherwise, the width and distance of the second slit will determine the solid angle - and thus the number of events that reach your detector.
A linear slit with 1° opening, and that extends ± 20° in the perpendicular direction, will have a geometric efficiency that is roughly given by the area of such a slit projected onto a sphere divided by the area of the sphere:
$$G = \frac{1\cdot \frac{\pi}{180} \times 20\cdot\frac{\pi}{180}}{4\pi} \approx 0.05\%$$
For your 10 µCi source, if it's small enough, that means 18 counts per second. If your source is bigger than the resolution you want to achieve, you will have to use the second slit - with a further reduction in counts.
In medicine, there's a modality called SPECT imaging (single photon emission computed tomography); it's also known as nuclear medicine (because it's used to image molecules labeled with radioactive isotopes, injected into a patient). The problem of collimation (of the detector) in order to image the origin of the radiation is one of the big problems for this modality. Either you get lots of counts but don't know (very well) where they came from, or you have just a few, with good precision. The above calculation illustrates that in a nutshell...
As for your follow-up question: there are various detectors available for this type of radiation. Since you will be subject to background, you will want it well shielded, and perhaps capable of energy discrimination (so you only detect radiation of the right energy and reject everything else). For compactness, CdZnTe detectors might be hard to beat, although modern scintillator+silicon photomultiplier devices are becoming more attractive. 
Are you sure that using radioactive sources is the way to go? I wonder if we're solving an XY problem here...
