I have had success with one of my horizontal seismometers. But, it was a surplus commercial variety and with very sturdy construction. It was a "garden-gate" type. Here, I ended up using a Allegro UGN-3503 linear Hall sensor. That sensor is about 4 times less sensitive than the A3515 unit. The reason....the greater horizontal mass drift even here would have exceeded the limits of the A3515 very quickly. This was due in large part to the confining area available with that particular instrument.
If you have a similar circuit build and have experimented
with a four pole neodymium magnet around the sensor you use, you will quickly
see that the distance between them has a great bearing on the sensitivity
of the whole unit. The closer the two, the higher the sensitivity,
and the further apart, the lower it becomes. In a sense, this distance
could be
considered a circuit "gain" adjustment. Probably
for this reason, it would be a good idea to limit the distance to no more
than 1/4", between the Hall and the magnet used, and then experiment with
shorter ranges, and actual circuit gains, to find the right distance for
your instrument. In my S-G design, the distance is about 1/32th inch
separation (with the A3515), while the horizontal distance was in the neighborhood
of about 3/16th inch (with the UGN3503).
I would really strongly suggest the use of surplus 4 pole neodymium magnets. The primary reason is that they exhibit a very sharp magnetic "junction" between the "two" magnets they inhabit in one piece of material. Other magnets of various materials, just don't have this key element of a sharp "junction" or "boundary", and are not worthwhile to use. Of course most magnets are 2 pole; i.e., they only have a north and south pole (No sharp change) like the 4 pole. Most of the surplus variety neodymium magnets have a characteristic "butterfly" or "wing shape". Earlier models but with 2 pole configuration often were painted white, and are not what we want here. If you have access to a possible 4 pole neodymium magnet in a surplus store, and run, say, a paperclip (magnet preferred) over the flat long surface, you will see that in the middle, it reverses polarity sharply. The paperclip will tend to jump from one side to the other in a short space of movement aroundthe middle of the length. Unfortunately, in many surplus web sites, they don't specify whether they are 2 pole or 4 pole magnets, and you may have to buy a variety to find one. One tip is that alot of the newer 4 pole neodymium magnets are covered with what looks like a chrome coating all over.