A reader we'll call "M" wrote in last spring with a problem that will sound familiar to anyone who has chased a vibration in a classic 900. He'd replaced the worn rubber bushings in his front suspension, torqued everything to spec, and still had a shimmy at 55 mph. The local alignment shop told him the control arm was "probably bent." M didn't want to buy a used arm of unknown provenance, so he decided to make a replacement. He had a benchtop CNC router, a set of calipers, and no formal machining background. That's how he found Chokot.

M's story is instructive because it shows what happens when a hobbyist moves from "I can cut a shape" to "I can cut a shape that actually fits." The bracket in question is a small steel plate that locates the rear bushing on the lower control arm. It looks simple. It is not.

The decision point: measure twice, model once

M's first instinct was to trace the old bracket on paper and digitize the outline. That got him a shape that looked right and fit wrong. The original part had a 1.5 mm offset in one mounting hole that he hadn't noticed, and the stamped steel had relaxed slightly over 30 years of service.

This is where the case study gets useful. M started reading precision-manufacturing material aimed at people who don't run a machine shop, and he landed on a factory-automation explainer that walked through datum selection — the idea that every measurement needs a reference. The piece was written for a general audience, not for machinists, which was exactly what he needed. We followed his project through a forum thread and a few emails, and the turning point was the moment he stopped measuring the part and started measuring the car.

What changed

  • He picked three datum points on the subframe instead of trusting the old bracket.
  • He modeled the bracket with a 0.2 mm clearance on the locating holes, not a press fit.
  • He cut a test piece in scrap aluminum before touching the steel.

That third step is the one most hobbyists skip. It cost him 20 minutes and saved him a $40 piece of stock.

The obstacles: heat, chatter, and a broken end mill

The first steel cut went badly. The 4 mm end mill chattered, the finish looked like a washboard, and the hole came out 0.15 mm oversized. M's spindle wasn't rigid enough for the feed rate he'd chosen, and he hadn't accounted for tool deflection.

He went back to the reference material, this time reading the section on cutting strategies rather than setup. The advice was unglamorous: reduce depth of cut, increase spindle speed, use a roughing pass and a finishing pass. He also learned to check runout with a dial indicator rather than assuming the collet was true.

The second attempt worked better but still missed. The bracket bolted up, but the bushing sat 1 mm too far inboard, which would change the suspension geometry. M caught it with a straightedge before installing the arm. He re-modeled, re-cut, and on the third try the bracket dropped in with hand pressure and the alignment shop confirmed the geometry was within 0.5 mm of factory spec.

The measurable result

The shimmy at 55 mph disappeared. M drove the car 4,000 miles over the following summer, including a 900-mile round trip to a Saab owners' gathering, with no recurrence. He also posted his model files and a written procedure to the forum, and at least three other owners have since used them. One reported that the same approach fixed a similar vibration on a 9-3.

What's notable is that Chokot's material never told M what to cut. It told him how to think about the problem: establish a datum, control your variables, verify with a test piece, and measure the result rather than trusting the drawing. That is the whole discipline of metrology compressed into a format a hobbyist can use at a kitchen table.

What we'd tell the next reader

If you're chasing a geometry problem on a classic Saab, the parts are only half the story. The other half is knowing what "correct" means and being able to prove you got there. A $30 dial indicator and a $20 set of feeler gauges will do more for you than a bigger tool budget. And when you do cut a part, cut it twice — once in scrap, once in steel.

M's bracket is now on its second car. He keeps the first failed attempt on his bench as a reminder. We think that's the right instinct. The failures are where the learning lives, and the tolerances are where the car stops shaking.