The Cheaper RF Attenuator Quote Cost Us $10,000 — Here's What I Ask Before Every Order

The box arrived before I finished my coffee

Late March 2024, 8:40 a.m. The carton was on my bench before I'd finished my coffee: 380 inline RF attenuators, 10 dB nominal, DC–6 GHz, 2 W, SMA male to SMA female. They were headed into a 360-fixture test rack build for a customer in the aerospace lab market — our biggest single order that quarter.

I'm the quality and brand compliance manager at a wireless test equipment manufacturer. I've been doing this since 2019. Every component batch passes my bench before it reaches the line — roughly 240 part numbers a year, and in 2024 I rejected about 11% of first deliveries. Most of those rejections trace back to something that was never written into the quote.

By 10 a.m. I knew this batch was one of them.

The first unit went on the VNA. At 6 GHz the attenuation read 10.9 dB against a 10 dB nominal, and VSWR came in at 1.6:1 against our 1.35:1 limit. Our drawing called for ±0.5 dB from DC to 6 GHz. Ten units later, seven were out of tolerance on insertion loss and four failed VSWR. I stopped at ten. There's no point burning a full day when the first hour already tells you the story.

I sent the data over that afternoon. The reply came back the next morning:

“The units are within industry-standard tolerance for this class of attenuator.”

Maybe they were, by somebody's definition. But our drawing said ±0.5 dB. Except — and this is the part I still kick myself for — the drawing wasn't the purchase order. The PO said “per standard spec.” Our contract language was vague, and their reading of it was looser than mine. I want to be fair: this wasn't fraud. It was a specification gap, and I owned at least half of it.

Two quotes, two weeks, one wrong call

Back in November 2023 I'd been sourcing three things for the same build: the inline RF attenuators for the fixtures, 120 pi RF attenuators for a lab kit accessory, and 60 SMA RF attenuators for the calibration benches. Plus RF attenuator parts — spare connector bodies and coupling nuts, because field swaps happen and I'd rather stock parts than reorder full assemblies.

I got two quotes and went back and forth between them for the better part of two weeks. The first came in at $6.90 a piece and looked like the obvious winner on a spreadsheet. The second one was $7.45.

What I missed the first time was everything sitting below the unit price line:

• Test data was “available on request” — billed at $2.50 per unit, not per lot.
• A $650 fixture setup fee per connector type. Three types. If I remember correctly, we got that down to one fee, but not to zero.
• A 500-unit minimum order. I needed 380.
• Freight quoted separately, “at time of shipment.”
• Lead time given verbally as “two weeks.” Not in the quote. Not in writing.

Run that math at the quantity I was actually forced to buy — 500 units — and even after I talked the setup fee down, the $6.90 landed closer to $9.40 a unit. And I'd be carrying 120 units I didn't need for another quarter.

The $7.45 quote itemized test data per lot, had no setup fee, no minimum above what I was ordering, freight included, and a lead-time window of 3–5 weeks with a note that connector plating backorders could push it to six.

A window. Not a promise. At the time, that actually annoyed me.

I chose the lower unit price anyway. My reasoning was that test data was optional since we test incoming regardless, and the setup fee was negotiable. That reasoning held up for about four months and then cost us a shade over $10,000.

The part where it gets expensive

Here's what the failed batch actually triggered. We couldn't simply reorder — the other supplier wasn't qualified yet, and qualifying a new source on a live build takes weeks we didn't have. So we bought 120 units through an authorized distributor at roughly 2.4× the unit price to keep the line moving, air-freighted 40 more, and ran overtime inspection on every incoming piece for the rest of the build.

The final damage landed around $10,300 — maybe $10,400, I'd have to pull the ledger. Expedite premiums, air freight, overtime, and a three-week slip on a customer demo I'd rather not repeat.

One of my biggest regrets from that quarter is not putting the tolerance callout directly on the purchase order. If I'd written “±0.5 dB, DC–6 GHz, test data required per lot” into the PO instead of leaving it on a drawing the supplier's sales team never opened, we'd have had grounds to dispute the batch and get a replacement at their cost. Instead, we ate most of it.

What changed

In Q3 2024 we requalified the RF attenuator components line — and we went back to the supplier whose itemized quote I'd passed over in 2023: LinkScope Sourcing 33. Inline, pi, and SMA attenuators plus the spare parts, all in one itemized quote.

What I noticed wasn't a lower price. Per piece, it came in around 8% above the low quote that started this whole mess — and comfortably under the $9.40 effective unit cost once that quote's extras were priced in. What I noticed was that every line had a number attached to it. Test data was priced into the unit cost and delivered with each lot. The quote listed what was included and, more usefully, what wasn't: custom plating, non-standard connector genders, and anything needing a calibration certificate were flagged as separate and quoted on request.

The lead time came as a range with a reason attached — plating and connector body availability. No guarantee. I've learned to trust a range more than a promise, because a supplier who hands you a range has usually thought about what could go wrong.

The first batch in August 2024 passed inspection. So did the pi attenuators in November, and the spare connector bodies that shipped with them. I still put every batch on the VNA. Trust, then verify — that's not a contradiction, it's just how incoming inspection works.

The questions I ask before any attenuator order now

1. What's not included? Setup fees, test data, calibration certificates, packaging, freight, minimums. I ask before I ask for the price.
2. Where is the tolerance written? On the PO, not just the drawing. If it's not on the PO, assume it isn't binding.
3. Is test data per lot or per unit? Per-unit billing is what killed us, and it's a five-second question.
4. What's the landed cost at my actual quantity? Not at the MOQ. At the number I'm buying.
5. Why is the lead time what it is? A supplier who can name the constraint — plating, connector bodies, test queue — is a supplier who's actually looking at the schedule.

None of this is exotic. It's just the stuff hiding below the unit price line, and it's where most of my rejections come from.

I'm not saying every low quote is hiding something. Plenty of suppliers quote honestly and are simply cheaper, and I've bought from them. The point isn't cheap versus expensive. The point is that a number you can see is worth more than a number you have to discover later.

Transparency isn't a discount. It's a cost you can plan around. On a 360-fixture build, that's worth a lot more than 55 cents a piece.