100228-2028 EST
Dustin:
Your question lacks a lot of necessary detail. Thus, I will make some assumptions.
Assumptions:
At the main panel with no load you read 240 V. Loaded with 100 A steady state, max for your shop, (a continuous 100 A on a 100 A would be too high), this drops to 218 V. If the source and load were resistive, then the source impedance as seen at the main panel from and including the transformer and wire is about 22/100 = about 0.22 ohms. Power company is probably using small wires over a long distance from their transformer to your meter, and if it is a 25 KVA transformer then it has substantial internal impedance relative to your load. What is the distance from the transformer to your main panel?
Do you know what is the transformer size? Sometimes it is marked on the outside. What is the distance from the transformer to your main panel?
The additional drop from the main to the shop, a 100 ft distance, is 6 V at an assumed 100 A produces a line impedance of 6/100 = 0.06 ohms. That might be #8 copper.
At the main and the subpanel you need to define the no load voltages, and the voltage changes for a known load current. I suggest that to run the tests everything is turned off at the main except the breaker to the subpanel.
Would a boost transformer be useful, probably. But you really need to define your load currents, wire sizes, and material.
What are your large motors, size? Is there a lot of inertia load? How long to get up to full speed?
Is inrush your major problem?
Your present result may be what you should expect.
.