High voltage MLCC thickness is not a smooth function of rated voltage. We measured the thickness field on all 4,094 real rows of Walsin’s Middle and High Voltage series: 1,733 clear a 1.20 mm worst-case height. Above 2,000 Vdc, none do. At 400 V and 450 V, none do either.
Venatech Limited, an authorised Walsin distributor in Hong Kong, holds the 24,026-row combined export 00_ALL_MLCC_parts_combined.xlsx from Walsin Technology (Walsin) and the 13-page approval sheet 04_Middle_and_High_Voltage.pdf. Every thickness below was read from the export and then checked against the external dimensions table on page 3 of that sheet, because a height taken from an export alone is a figure we will not stand behind. Every part number is real.
I have a 1.2 mm height budget and a 1,000 V bus — does a Walsin part exist?
Yes. 123 rows, 118 distinct part numbers. We filtered all 612 rows at 1,000 Vdc in 00_ALL_MLCC_parts_combined.xlsx on worst-case height — nominal thickness plus the positive tolerance, which is the only number a stack-up can use. 123 of them land at or below 1.20 mm, on two published thicknesses: 0.80 ± 0.10 mm, a 0.90 mm worst case, and 0.95 ± 0.10 mm, a 1.05 mm worst case. They sit in three case sizes: 1206 on 94 rows, 0805 on 19 and 1210 on 10.
| Walsin part number | Size | Dielectric | Capacitance | Published thickness | Worst case | Match / mismatch vs a 1.20 mm budget |
|---|---|---|---|---|---|---|
1206B152J102 |
1206 | X7R | 1500 pF | 0.80 ± 0.10 mm | 0.90 mm | Match — 0.30 mm of margin, highest capacitance that clears |
1206B102J102 |
1206 | X7R | 1000 pF | 0.80 ± 0.10 mm | 0.90 mm | Match |
0805B122K102 |
0805 | X7R | 1200 pF | 0.80 ± 0.10 mm | 0.90 mm | Match — smallest footprint that clears at 1 kV |
1206N470J102 |
1206 | NP0 | 47 pF | 0.95 ± 0.10 mm | 1.05 mm | Match — 0.15 mm of margin |
1210N100J102 |
1210 | NP0 | 10 pF | 0.95 ± 0.10 mm | 1.05 mm | Match |
1210N682J102 |
1210 | NP0 | 6800 pF | not published | unknown | Mismatch — no thickness in the export; symbol K decodes to 2.00 ± 0.20 mm on the sheet |
1206N182K631 |
1206 | NP0 | 1800 pF | 1.60 ± 0.20 mm | 1.80 mm | Mismatch — 0.60 mm over budget |
1812B103K352 |
1812 | X7R | 10 nF | 2.50 ± 0.30 mm | 2.80 mm | Mismatch — the only 3,500 Vdc row in the catalogue |
Note the sixth row. 1210N682J102 publishes no thickness at all, and the height has to be recovered from the Symbol letter against the dimension table in 04_Middle_and_High_Voltage.pdf — which is the finding further down.
How many thicknesses does Walsin publish across the 4,095 high-voltage rows?
Nine nominal values, on 4,094 real rows. The series holds 4,095 rows in 00_ALL_MLCC_parts_combined.xlsx; one jams the series name and a tab character into its Size field, so we exclude it and count 4,094. We counted all 4,094. The T column carries 16 distinct strings resolving to nine nominal heights, and every one of the nine appears in the external dimensions table on page 3 of Walsin’s approval sheet with a symbol letter beside it.
| Nominal thickness | Symbol in the datasheet | Rows | Worst case | Voltage range in the export |
|---|---|---|---|---|
| 0.50 mm | N |
38 | 0.55 mm | 200 V to 250 V only |
| 0.60 mm | A |
370 | 0.70 mm | 200 V to 630 V |
| 0.80 mm | S, X, B |
1,055 | 0.87 to 0.95 mm | 200 V to 2,000 V |
| 0.95 mm | C |
270 | 1.05 mm | 200 V to 2,000 V |
| 1.25 mm | D, I |
1,434 | 1.35 to 1.45 mm | 200 V to 4,000 V |
| 1.60 mm | G, P |
371 | 1.80 to 1.90 mm | 200 V to 4,000 V |
| 2.00 mm | K |
292 | 2.20 mm | 200 V to 4,000 V |
| 2.50 mm | M |
197 | 2.80 to 3.10 mm | 200 V to 4,000 V |
| 2.80 mm | U |
2 | 3.10 mm | 450 V only |
Nine heights for a range spanning 200 V to 4 kV, and the thickest of the nine exists at exactly one voltage step, which is not the highest one.
Which heights stop existing above a given voltage?
Four of the nine, and the cut-offs are sharp. 0.50 mm stops at 250 V: all 38 rows at that thickness sit at 200 or 250 Vdc, every one of them 0402. 0.60 mm stops at 630 V — 370 rows, none above it. 0.80 mm and 0.95 mm both stop at 2,000 V, between them 1,325 rows. Above 2,000 Vdc the thinnest thing Walsin’s published catalogue offers here is 1.25 mm nominal, 1.35 mm worst case.
| Rated voltage | Rows | Thinnest worst-case height | Rows clearing 1.20 mm |
|---|---|---|---|
| 200 Vdc | 654 | 0.55 mm | 429 |
| 250 Vdc | 867 | 0.55 mm | 524 |
| 400 Vdc | 10 | 1.80 mm | 0 |
| 450 Vdc | 19 | 1.80 mm | 0 |
| 500 Vdc | 647 | 0.70 mm | 293 |
| 630 Vdc | 637 | 0.70 mm | 283 |
| 1,000 Vdc | 612 | 0.90 mm | 123 |
| 1,500 Vdc | 29 | 0.90 mm | 2 |
| 2,000 Vdc | 356 | 0.90 mm | 79 |
| 2,500 Vdc | 27 | 1.35 mm | 0 |
| 3,000 Vdc | 212 | 1.35 mm | 0 |
| 3,500 Vdc | 1 | 2.80 mm | 0 |
| 4,000 Vdc | 22 | 1.35 mm | 0 |
1,733 of 4,094 rows clear a 1.20 mm worst-case budget. 1,500 Vdc is the thin spot in the middle of the range: 29 rows exist and exactly two of them clear 1.20 mm. The single 3,500 Vdc row in Walsin’s published catalogue, 1812B103K352, publishes 2.50 ± 0.30 mm — a figure we established and published previously — so a step up from 3,000 V to 3,500 V for margin doubles the component height, while a step to 4,000 V does not.
Why are 400 V and 450 V thicker than 2,000 V?
Because they are not the same kind of part. All 29 rows at 400 and 450 Vdc are X7R, and all but five are 68 nF or larger, up to 1 µF. The thinnest worst-case height at either step is 1.80 mm, double the 0.90 mm available at 2,000 Vdc, and zero of the 29 clear a 1.20 mm budget. The two thickest rows in the entire 4,094-row series sit here: 1812B105K451 and 1812B105J451, both 1812 X7R 1 µF at 450 Vdc, both 2.80 ± 0.30 mm.
We have not seen this counted anywhere. The height floor in Walsin’s published high-voltage range is not monotonic in rated voltage, and the discontinuity is not at the top — it is at 400 and 450 V, where the parts stop being small-value NP0 chips and become bulk X7R. A German design that derates a 1,000 V bus part down to a 450 V part to buy margin elsewhere can double its height doing it. 1206B104K451, a 1206 X7R 100 nF at 450 Vdc, is the thinnest row at that step and it is still 1.60 ± 0.20 mm.
What does a 1.20 mm budget cost you in capacitance at 1,000 Vdc?
A ceiling of 1,500 pF. We sorted all 123 qualifying rows in our extract by capacitance. The highest value that clears is 1,500 pF, held by 1206B152J102, a 1206 X7R part at 0.80 ± 0.10 mm. Of the 123 rows, 101 are NP0 and 22 are X7R, and the NP0 rows top out at 1,000 pF.
That is the real cost of the height constraint. Walsin’s approval sheet puts the series capacitance ceiling at 1 µF. Hold the height to 1.20 mm worst case on a 1,000 V rail and the catalogue available to you ends at 1.5 nF — roughly three orders of magnitude below what the series can do if height is free. On a DC-link snubber or a high-voltage divider that is a topology decision, not a part decision.
Which high-voltage rows publish no thickness at all, and can the height be recovered?
65 rows publish no thickness, and the height is recoverable on 64 of them. The T, L and W cells are empty on 65 of the 4,094 rows. But 64 of those 65 still carry a letter in the Symbol column, and that letter is the thickness symbol printed in Walsin’s own dimension table — N is 0.50 ± 0.05 mm, B is 0.80 ± 0.10 mm, D is 1.25 ± 0.10 mm, K is 2.00 ± 0.20 mm. Decode the symbol and the height comes back.
Exactly one row of the 4,094 carries neither a thickness, nor a symbol, nor a capacitance: 0603N102J251, a 0603 NP0 part at 250 Vdc with three empty decision fields.
The symbol is not a free pass either. We checked the Symbol letter against the T string on all 4,029 rows that publish both, and 19 disagree against the dimension table in 04_Middle_and_High_Voltage.pdf. Eight are 1812 rows whose T reads 2.50 ± 0.30 mm while the symbol is D, G or K — letters the datasheet assigns to 1.25, 1.60 and 2.00 mm. Four are 1210 NP0 3300 pF rows at 1,000 Vdc, 1210N332F102 and its three tolerance siblings, with T at 2.00 ± 0.20 mm and symbol M. Four are 1812 NP0 560 pF rows at 3,000 Vdc, including 1812N561F302, same contradiction. One, 1812B222M302, carries a lower-case m the datasheet does not define. One more, 1210B224K401, publishes 2.50 ± 0.60 mm — a tolerance that appears nowhere in the table.
What must a German engineer confirm before a height-constrained design freeze?
Three figures, and none of them is the capacitance. Confirm the worst-case height, not the nominal — the positive tolerance here runs from 0.05 mm to 0.60 mm, so the nominal understates the stack-up by up to 24%. Confirm it against the datasheet and not the export, because 65 rows publish no thickness and 19 contradict their own symbol. Then confirm the dielectric strength requirement for your rail, which the export has no column for at all.
- Name the part number and ask for the sheet. We hold 28 Walsin datasheets, 451 pages, and send them on request rather than describing them. The public URL for every MLCC series is in
INDEX_MLCC_series.csv. - Cross the whole list in one pass. Paste up to 500 part numbers or drop an Excel or CSV into the free cross reference tool, use the low-profile special-requirement filter, and read every stated difference rather than a claimed match.
- Check what was physically in Hong Kong on ready stock — a dated snapshot of part numbers as at 23 September 2026, not a live feed.
For the capacitance side of the same range see the high-voltage MLCC capacitance limit, for the series overview middle and high voltage MLCC, 200 V to 4 kV, and for an 800 V traction example EV traction inverter passives in Germany. The full range sits on the Walsin MLCC page.
The short version
High voltage MLCC thickness buys you less than the voltage ladder suggests. Across all 4,094 real rows in Walsin’s published Middle and High Voltage range there are nine nominal heights, and 1,733 rows clear a 1.20 mm worst-case budget. A 1,000 V bus with a 1.2 mm budget has 118 distinct part numbers available, capped at 1,500 pF. Above 2,000 Vdc nothing clears 1.20 mm. At 400 and 450 Vdc nothing clears it either, because those 29 rows are bulk X7R and the two thickest parts in the series — both 1 µF 1812 — sit at 450 V, not at 4 kV. 65 rows publish no thickness, 64 of them recoverable from the datasheet’s symbol letter, and 19 rows carry a symbol that contradicts their own thickness. Qualification stays with you.
Send us the high-voltage lines off your German bill of materials with the height budget written next to them, and we will come back with the worst-case figure from the approval sheet rather than the nominal from the export, and name every row where the two sources disagree. And a question worth asking your own stack-up: when you derated to buy voltage margin, did anyone re-check the height?
Frequently asked questions
Does a Walsin high-voltage MLCC exist for a 1,000 V bus under a 1.2 mm height budget?
Yes. We filtered all 612 rows at 1,000 volts direct current on worst-case height: 123 rows and 118 distinct part numbers land at or below 1.20 millimetres, at 0.80 plus or minus 0.10 and 0.95 plus or minus 0.10 millimetres. They sit in 1206, 0805 and 1210 case sizes.
What is the thinnest Walsin high-voltage MLCC at each rated voltage?
The worst-case floor is 0.55 millimetres at 200 and 250 volts, 1.80 millimetres at 400 and 450 volts, 0.70 millimetres at 500 and 630 volts, 0.90 millimetres from 1,000 through 2,000 volts, and 1.35 millimetres at 2,500, 3,000 and 4,000 volts. The single 3,500 volt row is 2.80 millimetres worst case.
Which MLCC heights disappear above a given voltage?
Four of nine. The 0.50 millimetre height exists only at 200 and 250 volts, across 38 rows. The 0.60 millimetre height stops at 630 volts, across 370 rows. Both 0.80 and 0.95 millimetres stop at 2,000 volts. Above 2,000 volts the thinnest published nominal is 1.25 millimetres.
Why are Walsin’s 400 V and 450 V high-voltage MLCC thicker than its 2,000 V parts?
Because those 29 rows are bulk X7R rather than small-value NP0 chips, and all but five are 68 nanofarad or larger, up to 1 microfarad. The thinnest worst-case height at either step is 1.80 millimetres, double the 0.90 millimetres available at 2,000 volts. Zero of the 29 clear a 1.20 millimetre budget.
How much capacitance can I get at 1,000 Vdc under a 1.2 mm height budget?
1,500 picofarad. We sorted all 123 qualifying rows by capacitance and the highest value that clears is a 1206 X7R part at 0.80 plus or minus 0.10 millimetres. Of the 123 rows, 101 are NP0 and those top out at 1,000 picofarad. The series itself reaches 1 microfarad if height is free.
What do I do when a Walsin high-voltage row publishes no thickness?
Decode the Symbol column against the external dimensions table on page 3 of the Middle and High Voltage approval sheet. 65 of the 4,094 rows publish no thickness, length or width, and 64 of those still carry the sheet’s thickness symbol letter, so the height is recoverable. One row carries no symbol either.
Is the thickness in the export reliable enough to freeze a design on?
Not on its own. We checked the Symbol letter against the thickness string on all 4,029 rows that publish both: 19 disagree against the sheet’s own dimension table, one of them using a lower-case symbol the sheet never defines. Ask for the approval sheet for your exact part number and work from the worst case.