Matched Thin Film Resistor Pairs for a 4–20 mA Input in Belgium: 428 of 2,016 Cells Hold Two

A matched thin film resistor pair for a 4-20 mA input needs two part numbers sharing one decade, one TCR grade, one tolerance and one footprint. We built that grid from Walsin’s export: 2,016 cells, 452 hold a part, 428 hold two or more, 24 hold exactly one. There is no 250 Ω part.

---

Your 4–20 mA input stage specifies a 250 Ω burden resistor. We searched all 93,914 thin film rows and all 223,376 discrete thick film rows for 250 Ω and found zero. The E96 ladder goes 243, 249, 255. That is the sort of thing you want to know before the schematic is frozen, not after.

Venatech Limited, an authorised Walsin distributor in Hong Kong, holds all 20 thin film parametric export files from Walsin Technology (Walsin). This post is measured from the eleven non-empty ones — Thin_Film_Precision_Chip_Resistor_WF_Series_0603_160.xlsx and its seven siblings, Tantalum_nitride_Thin_Film_Chip_Resistor_MF_Series_1.xlsx and two Anti_Sulfuration_Thin_Film_Chip_Resistor_SF_Series_*.xlsx sheets. We grouped every row into a (resistance decade, TCR grade, tolerance, case size) cell and counted the part numbers in each. Every part number below is real.

How many catalogue cells can supply a matched thin film pair at all?

428 of 2,016. We built the full grid — 7 resistance decades from 1 Ω to 10 MΩ, 6 TCR grades, 6 tolerances, 8 case sizes — and counted part numbers per cell. 1,564 cells are empty. 24 hold exactly one part number, which cannot form a ratio pair. 428 hold two or more, which is the minimum a ratio-matched divider needs.

Cell population	Cells	Share of 2,016	What it means for a pair
Empty	1,564	77.6%	Nothing to buy at that combination
Exactly 1 part number	24	1.2%	One leg only — no ratio pair possible in-cell
2 or more part numbers	428	21.2%	A ratio pair is buyable inside one cell
Non-empty total	452	22.4%	—

We have not seen this counted anywhere. The useful number is not 452 but 428, because a divider is two parts and the second one has to come from the same cell if the grade, tolerance and footprint are to be identical. The distinction costs 24 cells, and we show below that 22 of those 24 sit in one decade.

Does the decade a 4–20 mA sense resistor sits in hold pairs?

Yes, and generously. A 4–20 mA loop converted to 1–5 V needs a burden resistor near 250 Ω, which puts it in the 100 Ω to 1 kΩ decade. We counted that decade on its own: 75 of its 288 cells hold a part number, and every one of those 75 holds at least 205 part numbers. Not one single-part cell exists in that decade.

Case size	Non-empty cells in 100 Ω – 1 kΩ	Smallest cell	Tightest TCR at this size	Tolerances offered
0201 (0603)	6	205	≦±25 ppm/°C	±0.1%, ±0.5%, ±1%
0402 (1005)	6	205	≦±3 ppm/°C	±0.01% only
0603 (1608)	6	205	≦±3 ppm/°C	±0.01% only
0805 (2012)	6	205	≦±3 ppm/°C	±0.01% only
1206 (3216)	26	205	≦±3 ppm/°C	all six
1210 (3225)	5	205	≦±5 ppm/°C	±0.01% only
2010 (5025)	15	205	≦±5 ppm/°C	all six
2512 (6432)	5	205	≦±5 ppm/°C	±0.01% only

So for the decade a process-control input actually lives in, pair availability is not the constraint. 1206 is the broadest footprint by a wide margin with 26 live cells, and 2010 second with 15, because those are the two sizes the tantalum nitride MF and anti-sulfuration SF series occupy. The four smallest sizes each offer six cells and a single tolerance.

Can I buy a 250 Ω sense resistor from this catalogue?

No. Zero of 93,914 thin film rows and zero of 223,376 discrete thick film rows carry 250 Ω. We searched both exports for the exact value. The nearest parts are 249 Ω, which the thin film export holds on 98 part numbers across all eight case sizes, and 255 Ω. A 249 Ω burden resistor turns 20 mA into 4.98 V rather than 5.00 V, a 0.4% span error that your reference and gain setting have to absorb.

That is a design decision, not a procurement one, and it is better taken deliberately. Of the 98 part numbers at 249 Ω, 28 are at 1206, 20 at 2010, 12 at 0201, 8 each at 0402, 0603 and 0805, and 7 each at 1210 and 2512. The tightest combination available at that value is ±0.01% with ≦±3 ppm/°C, at four footprints.

What does one real 250 Ω requirement return?

Five real candidates and no exact value. We set the requirement at 250 Ω, ±0.1%, ≦±25 ppm/°C, 0805 — a plausible 4–20 mA burden specification — and read every row that came close.

| Part number | Series | Size | Resistance | Tolerance | TCR | Rated power | Match / mismatch vs 250 Ω, ±0.1%, ≦±25, 0805 | |—|—|—|—|—|—|—| | — no row — | any | any | 250 Ω | — | — | — | Mismatch — 250 Ω appears in no thin film or discrete thick film row at all | | WF08R2490TTL | WF precision | 0805 | 249 Ω | ±0.01% | ≦±25 | 1/8 W | Match on size and TCR; mismatch — value 0.4% low, tolerance 10× tighter than asked | | WF08B2490TTL | WF precision | 0805 | 249 Ω | ±0.01% | ≦±3 | 1/8 W | Match on size; mismatch — value 0.4% low, tolerance 10× tighter, TCR 8.3× tighter | | SF20U2490BTL | SF anti-sulfuration | 2010 | 249 Ω | ±0.1% | ≦±25 | 1/2 W | Match on tolerance and TCR; mismatch — value 0.4% low, 2010 footprint, adds sulfur resistance | | MF12W2490BTL | MF tantalum nitride | 1206 | 249 Ω | ±0.1% | ≦±10 | 2/5 W | Match on tolerance; mismatch — value 0.4% low, 1206 footprint, TCR 2.5× tighter, different film material | | WF12B2490TTL | WF precision | 1206 | 249 Ω | ±0.01% | ≦±3 | 1/4 W | Mismatch on three fields; 2× the power rating of the 0805 part |

Nothing here is a drop-in and we will not pretend otherwise. We will not tell you a part is equivalent. We will show you where it differs. The honest route is to accept 249 Ω and absorb 0.4% in the gain stage, or to move the burden value and the reference together. Confirm against the datasheet you request from us and qualify in your own circuit.

Which footprints can supply a ±3 ppm/°C pair in the same decade?

Four, and only at ±0.01% tolerance. We counted every cell at the ≦±3 ppm/°C grade: 19 of a possible 336 are non-empty, all 19 at ±0.01%, and all 19 hold two or more part numbers. They sit at 0402, 0603, 0805 and 1206 and nowhere else. 0201, 1210, 2010 and 2512 hold no ≦±3 cell at any decade.

Decade	0402	0603	0805	1206
1 Ω – 10 Ω	—	32	32	32
10 Ω – 100 Ω	205	205	205	205
100 Ω – 1 kΩ	205	205	205	205
1 kΩ – 10 kΩ	185	205	205	205
10 kΩ – 100 kΩ	—	122	185	205
100 kΩ – 1 MΩ	—	—	—	16
1 MΩ – 10 MΩ	—	—	—	—

WF06B2490TTL at 249 Ω and WF06B4990TTL at 499 Ω are a real pair from one cell: same 0603 footprint, same ±0.01%, same ≦±3 ppm/°C, same 1/10 W, a 2:1 ratio. WF08B2490TTL and WF08B4990TTL are the same pair one footprint up, at 0805 and 1/8 W. That is what a matched pair looks like when it is actually available.

Where does pair availability actually break?

Above 1 MΩ, and almost entirely in one decade. Of the 24 cells that hold exactly one part number, 22 are in the 1 MΩ to 10 MΩ decade, and every one of those 22 holds the same single value: 1.00 MΩ. The other two are WF04Z1002TTL at 10 kΩ, 0402, ≦±5 ppm/°C, and WF08Z1003TTL at 100 kΩ, 0805, ≦±5 ppm/°C.

The 22 break down cleanly. Twenty are tantalum nitride MF parts at 1206 — MF12W1004BTL, MF12F1004BTL, MF12R1004BTL and MF12Q1004BTL are four of them, one per TCR grade at ±0.1% — and the MF series holds exactly one resistance value above 1 MΩ. The remaining two are WF20W1004TTL and WF20F1004TTL at 2010 and ±0.01%. So if a high-impedance leg above 1 MΩ has to be ratio-matched inside one grade and footprint, this catalogue cannot do it at ≦±10 or ≦±15 ppm/°C. We have not seen that counted anywhere either.

What does the catalogue not tell a Belgian process engineer?

Everything about the matching itself. We read all nine column headers in every one of the 20 thin film files. There is no ratio-tolerance column, no TCR-tracking column, no matched-pair designator, and no operating-temperature, lifecycle or qualification column anywhere in the thin film export. Two parts from the same cell share a grade; they are not specified as a tracked pair.

That matters for a 4–20 mA input because the error that reaches your ADC is ratio error, not absolute error. Two parts each at ±0.01% can sit at opposite ends of their windows, and two parts each at ≦±3 ppm/°C can drift in opposite directions, which is up to 6 ppm/°C of ratio drift across the pair. Our earlier post on precision divider ratio accuracy works that arithmetic through.

What should a Belgian plant or OEM ask for?

Ask for the cell, not the part. Belgian chemical, pharmaceutical and process plants run thousands of 4–20 mA loops, and an input card redesign is the kind of change that drags an installed base behind it. The question that gets a useful answer is “which cell holds both my legs”, not “what is the cross for this part number”.

1. State the decade, the grade, the tolerance and the footprint as one requirement. That is the cell. If it holds two or more part numbers you have a pair; if it holds one you have a problem, and 24 cells do.
2. Accept the E96 value or move the reference. 250 Ω does not exist in either export. 249 Ω does, on 98 part numbers.
3. Keep high-impedance legs below 1 MΩ if they have to be matched. Above 1 MΩ, 22 of 46 live cells hold a single value.
4. Cross the whole BOM in one pass. Paste up to 500 part numbers or drop your Excel or CSV into the free cross-reference tool. It decodes 18 manufacturers’ part numbers, is matched against Walsin’s published catalogue across 800,000+ crossable parts, and states every difference from your original rather than hiding it.
5. Check what is physically in Hong Kong on ready stock. That is a dated snapshot as at 23 September 2026, not a live feed, and thin film has zero representation in it — ask us directly rather than assuming.

For the material behind the MF parts in the single-part cells, see tantalum nitride thin film resistors. The full range is on the Walsin thin film range page.

The short version

A matched thin film resistor pair for a 4-20 mA input is a cell question, not a part question. The grid of 7 decades, 6 TCR grades, 6 tolerances and 8 case sizes holds 2,016 cells: 1,564 are empty, 24 hold exactly one part number and 428 hold two or more. In the 100 Ω to 1 kΩ decade a burden resistor sits in, 75 of 288 cells are live and every one holds at least 205 part numbers, so pairs are abundant there. ≦±3 ppm/°C pairs exist in 19 cells, all at ±0.01%, at 0402, 0603, 0805 and 1206 only. 22 of the 24 single-part cells are above 1 MΩ and every one of them holds the same 1.00 MΩ value. And 250 Ω does not exist: zero rows in 93,914 thin film and zero in 223,376 discrete thick film.

Send us the decade, the grade, the tolerance and the footprint together and we will tell you whether that cell holds two part numbers, name them, and state every field where each differs from what you asked for. Qualification stays with you, and we will not imply otherwise.

So: when your input card specification says 250 Ω, which of the two does it actually mean — 249 Ω, or a reference you have not re-scaled yet?

---

Frequently asked questions

---