It was a Tuesday in July 2023 when the RFS 8110 showed up. The box was smaller than I expected: plain cardboard, no marketing stickers, no big logo. Inside was a silver housing with two N connectors and a packing slip that said "RFS 8110, filter, 50 ohm, N/N." The warehouse guy set it on the counter and said, "Looks generic." I agreed. That should have been my first clue.
I've been handling RF equipment orders for about six years. In that time I've personally made—and documented—13 significant mistakes, totaling roughly $46,000 in wasted budget. The RFS 8110 order cost me about $1,270 on paper and closer to $2,800 once I counted labor and the rental. Now I maintain our team's RF ordering checklist, mostly so nobody has to repeat it.
If you've ever searched for "rfs equipment" or "rfs filter" and ended up staring at a model number you don't recognize, this story is for you.
What Is an RFS Filter?
The short definition: RFS stands for Radio Frequency Systems, a manufacturer of RF transmission components. An RFS filter is a frequency-selective device used in transmit and receive paths to keep wanted frequencies in and unwanted frequencies out. RFS equipment includes antennas, Cellflex cables, leaky feeders, combiners, filters, and other site infrastructure. A big part of their U.S. manufacturing and support runs through De Soto, KS.
In practical terms, a filter like the 8110 is a tuned cavity. The center frequency, bandwidth, insertion loss, return loss, rejection, impedance, and power rating all matter. The 8110 family is not one generic part; each variant is tuned to a specific band. Per RFS product documentation, a filter order is incomplete if any of those parameters are missing.
Per RFS product documentation: ordering an RF filter requires center frequency, bandwidth, insertion loss, return loss, rejection, and power handling. If you're only specifying a model number and connectors, you haven't specified the filter. RFS datasheet / application note, accessed January 2025
"What is a filter?" is one of those questions that sounds easy. A filter is just a gate. It lets your signal through and stops everything else. The hard part is that "everything else" is defined by the band plan, not by the filter's general shape. I learned that with my first 8110 order.
How I Ordered the Wrong RFS 8110
We were refurbishing a site for a public-safety client in central Kansas. The client needed a transmit filter for a UHF repeater. I was busy, and I took a shortcut: I searched the RFS catalog, found "8110," saw N to N connectors, and assumed it was the right filter. Actually, I didn't even look up the dash suffix. I just added "RFS filter 8110" to the purchase requisition under notes.
What I should have been looking at was the band plan. If I remember correctly, the repeater used a receive frequency around 454 MHz and a transmit frequency around 459 MHz. The 8110 variant I chose was specified for a lower segment—maybe 420–450 MHz—though I might be misremembering the exact split. The point is, the passband didn't cover our transmit offset. The connectors were fine. The impedance was fine. The tuned cavity was wrong.
From the outside, RFS equipment looks like a generic aluminum box. The reality is that the internal dimensions, the tuning slugs, and the test sheet are all specific to a frequency band. Our vendor's invoice even listed the model number with the band suffix. I saw "N" for connector and stopped reading.
What Happened Next
The filter arrived in five days from De Soto, KS. It came with a printed test sheet showing the return loss across the test band. In my opinion, that test sheet is where you can catch most problems. I didn't read it closely. I saw the response curve and thought, "That's a filter curve." It was, but it was a filter curve for the wrong band.
We still installed it. I told myself, "It's a cavity—we can tune it." That was a lie. The 8110 cavity can be tuned, but only within a narrow range. The "just tune it" thinking comes from an era when filters were wider and spectrum was less crowded. Today, with narrowband fleets and tight site permits, you don't get to slide a filter 15 MHz from where it was built without killing the performance.
Dana, our field tech, put the filter inline and ran a spectrum sweep. The insertion loss on the transmit side was about 1.5 dB higher than the datasheet promised, and the rejection skirt wasn't clean. She looked at the screen and said, "This isn't going to work." She was right. When she swept the test unit side by side with the spec sheet's curve, I finally understood why insertion loss and rejection are shown on the same chart: they trade against each other. We pulled the filter, put the old one back in, and I called our RFS rep.
Dodged a bullet in a small way: the client wasn't scheduled to go live for another ten days. The bad news was that the correct 8110 variant wasn't in our distributor's local stock. It would have to ship from De Soto, KS—again—and the replacement lead time was three weeks. We couldn't wait three weeks. So I paid for an emergency rental filter from a nearby supplier and had the correct RFS unit shipped on air freight. (Should mention: the rental was technically for a different site, but the frequencies happened to match. That's a level of luck I won't count on again.)
The Hard Numbers
Let's put the waste in plain terms. The original RFS 8110 order cost $1,240 plus $28 for expedited ground shipping. I remember the number because I had to explain it in a project review. Actually, the total was $1,268. The rental filter cost $400 for two weeks. The extra site visit cost about $600 in labor and truck time. Then there was the expedited reorder of the correct filter: $850 including freight. That put us at roughly $2,700—and that doesn't include the hour I spent on hold with the supplier or Dana's dignity after I blamed the filter first.
I'm not sharing this because the RFS equipment was bad. It wasn't. The mistake was mine: I treated a precision part number like a generic SKU. "RFS 8110" is not "a filter." It's a family. Within that family, the specific variant matters. The same mistake would have happened with any good manufacturer.
The Checklist That Finally Worked
After the project closed, I built a pre-order checklist for any RF filter or passive component. It's not complicated. It's the thing I skipped:
- Center frequency / band split: Does this variant cover our transmit, receive, or both? Write the exact licensed frequencies.
- Passband / insertion loss: At the band edges, what's the expected loss? Compare with the datasheet.
- Rejection requirement: Which out-of-band signals are we trying to suppress? Check the datasheet's rejection curves, not just the headline number.
- Power and impedance: 50 ohm? Average and peak power? If the product datasheet says "non-standard," why?
- Connectors and mechanical layout: N, DIN, or 7-16? Input and output orientation on the actual mounting bracket.
In the 12 months since I made that checklist, it has caught 47 potential errors. "Potential" because some are found by our purchasing team before I see them, but the list is the reason. It's not a sexy habit. It's a $2,700 one.
I recommend this checklist for anyone buying RFS equipment—or any manufacturer's equipment, honestly. But I'll add the honest limitation: this checklist is for straightforward transmit/receive filtering. If you're dealing with a multi-band combiner, a tower-mounted amplifier, or a custom notch situation, get an engineer who does this daily. The 8110 worked for us once we matched the variant. It's not the right answer for every site.
Wrapping Up: What Is a Filter?
So, what is a filter? After that July, I'd describe it as a gatekeeper with a specific accent. It only likes the frequencies it was tuned for. Everything else gets reflected or absorbed. If you give it the wrong passband, it doesn't matter how pretty the metal housing is, how fast it shipped, or how much the connector torque was checked.
RFS equipment, the 8110, De Soto, KS, and "what is a" are all connected in one practical answer: a filter is a precise, band-specific RF component, and the model number alone doesn't tell you enough.
Check the band. Check the test sheet. Check the variant. That's the entire lesson—and it cost me roughly $2,700 to learn it.
