@pzfreo@fsfarimani NIST SFA is a solid baseline. For a real-world check, a simple part with a few GD&T callouts run through a commercial CAM import and a CMM program would show fast whether the PMI survives without anyone retyping it.
@Aiyafun Too early in our beta to have a real pattern. In RFQs generally, quantity usually gets filled in. Finish is the one that gets skipped or written as "standard", which means something different to every shop. A specific callout like "Type II anodize, clear" saves a round trip.
@PraxeoSys Exactly, otherwise it's noise. If you try the beta, we'd most like to hear where the quote flow made you guess at something or enter it twice.
Aluminum parts that get taken apart a lot: tapped threads wear out. Rule of thumb: call out threaded inserts (helical coil or key-locking) in holes that see repeated assembly, and leave plain tapped holes for screws that go in once.
@itsdor_@esthxrtho Right, and those few usually trace back to function: a mating bore, a sealing face, the datum the assembly locates on. Leave the rest at general tolerance and you save cycle time and inspection cost.
@itsdor_@esthxrtho Agree. Same goes for buying parts: how someone preps a part for quoting (which tolerances they call out, which they leave at general) tells you a lot about whether they've been through production before.
@trawbigsocks@dhh From the buyer side, a lot of delay happens before that clock even starts: RFQs missing the material grade, finish or a drawing. Do you push back for a full package, or quote with your own assumptions and confirm later?
@fortunato_ai That matches what we hear. AI helps with the paperwork around the part (RFQ emails, reading specs, checking a drawing for missing info), and a person still proves out the program. The time saved is in the admin, not at the spindle.
@JonMStoddard The smaller-buyer version of this: know which outside shops you depend on and keep a second qualified source for critical parts. Most supply risk shows up as one shop's lead time jumping, not the shop disappearing.
@Em_Nomadic@gaoyy9 The usual surprises: tolerances that were fine on one hand-fitted unit, printed features that don't carry over to machining or molding, and drawings that never got made. Freezing a rev and writing down the critical dims early saves a lot of re-quoting.
@Ryan__Dunwoody@Karim_RC@julienajdenbaum@fdotinc Small batch is where DFM pays off most. At 10 to 50 units, setups and fixturing dominate the cost, so fewer setups, standard tool sizes and loose tolerances on non-critical features usually matter more than the material.
@farukguney_ai Agree. A quick worst-case stack on the 3 or 4 interfaces that actually matter catches most of it. Put tight tolerances only where the stack needs them and leave the rest at general tolerance. Tight everywhere just costs money without fixing the fit.
@StevenJoelMora I'd add one: send RFQs a shop can quote without a phone call. STEP + drawing with only the critical tolerances called out, material grade, finish, qty. Clean packages get faster, tighter quotes, and that compounds every iteration.
@securovix@PraxeoSys Worth splitting. In a lot of hardware teams the design loop is days and the parts loop is weeks, so supplier turnaround ends up being most of the cycle. Sending a complete RFQ (STEP, drawing, qty, finish) to more than one shop at once cuts a surprising amount of it.
@pzfreo@fsfarimani Will do. The test we'd care about is whether a shop's CAM and CMM side picks the PMI up cleanly, without anyone retyping tolerances off a PDF. That retyping step is where most of the errors creep in.
Counterbores and countersinks, rules of thumb: size counterbores to a standard screw chart so the shop can use a stock tool. Countersink angle is typically 82 degrees for inch flat-head screws and 90 degrees for metric. Mixing them up leaves the head proud.
@pzfreo@fsfarimani Good to know. Semantic AP242 PMI is what lets a shop's CAM and CMM read tolerances straight off the model instead of retyping them from a PDF. Keeping the STEP local by default makes sense for proprietary parts too.
10/10
Full article on LinkedIn, with sources and a checklist: https://t.co/PrQChD2ggs
Next in the series: aluminum.
On Vectarr, buyers upload CAD for an instant quote (Beta) and shops bid the real price and lead time. More at https://t.co/YF42FE7T5a
1/10
Part 4 of our deep-dive series: Choosing machine shops.
The lowest quote is not always the lowest-cost part. The right shop holds tolerance the first time, ships on the date it gave and flags problems early. Here's how to pick one, plus red flags and a checklist. A thread.
9/10
Start with a small trial order with a first article report, then keep score: on-time delivery, accepted parts, corrective actions.
Red flags: an unexplained low price, no questions, no fitting cert, vague lead time, hidden subcontracting, certs only after shipping.