Voltage regulation?

I am getting varied voltage readings in my 2000 XJ. One minute it is fine right around 13.9 volts then one minute it spikes all the way over to 19+ volts and stays there for a few minutes. When it is not spiking it will drop as low as 10 or so volts then rise to around 15 volts. Any ideas as to what would cause this problem? Is the voltage regulation internal to the alternator or is it controlled by the ECM?

It's in the ECM. I had a 92 doing the same thing, changed the computer & the problem was solved. JIM.

What is the best way to test for this? I need to be sure before I puchase a used or new ECM. I found my Haynes and it is saying the voltage regulation is internal to the alternator.....? I did replace my alt back in July with an Autozone special.

Hence, the reason most people consider the Haynes to be crap. Autozone specials don't have a great reputation either, but I'm not sure how you can blame this on a poor alternator. Is the connector on the back of the alternator making good contact and not falling apart?

True for 1990 and before - they used Delco alternators (typical for AMC - Delco or Motorola) and the Delco SI and CS series have internal regulators.

Chrysler uses Nippondenso alternators, and they use external regulators. 1991-up has the regulator circuit in the PCM.

However, it's entirely possible you've got a ground floating. Can you reproduce the problem while sitting still? If you can, do this:

Shut down.
Take out your booster cables, and connect the engine block (as close to the alternator as possible) directly to the battery - post. Watch for "mechanical interference" (getting the booster cable caught in moving parts is a Very Bad Thing (tm))
Restart, and try to reproduce the problem again. If you cannot, it was probably a ground issue - check, clean and/or service as necessary. If you can, move the engine end of the cable to a clean spot on the chassis and repeat - if that solves the issue, then check your chassis grounds and clean/service as necessary.

This costs a lot less than buying parts, and problems like this are often fixed incidental to replacing the part anyhow. Therefore, you don't know that you actually solved the wrong problem! This would probably take you about a half-hour, and this is the sort of test that costs nothing but your time one after-noon. If enhancing the ground doesn't fix it, then you know - and can replace parts with confidence that you actually do have to spend the money on them.