Venatech Limited, an authorised Walsin Technology (Walsin) distributor in Hong Kong, ships to
North America from Hong Kong. There is no US entity. AI server power draws on a narrow band of
the passive catalogue: high-voltage MLCC for the PFC and DC link, high-capacitance MLCC at the
point of load, and 2 W chip resistors for sensing and bleed.
Does Venatech have a US warehouse?
No. The group’s operating entities are Venatech Limited in Hong Kong, 颐威纳（深圳）国际贸易有限公司
in Shenzhen, VENATECH PTE LTD in Singapore and the Jiangmen factory in Guangdong, with Chengdu
in preparation. Hong Kong is the authorised distribution entity and the shipping origin for a
North American ship-to.
Say it plainly, because the alternative is worse. A distributor that implies local stock it
does not hold creates a schedule risk the buyer discovers at the wrong moment. What Hong Kong
does provide is authorised-only sourcing direct from the manufacturer, original sealed
manufacturer packaging, full lot and date code traceability, and a Certificate of Conformity
on request.
Which Walsin MLCC covers the high-voltage rails in a server power supply?
The Middle to High Voltage series, 4,095 part numbers rated 200 Vdc to 4 kV. It is the second
largest series in Walsin’s 24,026-row MLCC extract after general purpose, and it is where the
snubbers, DC link bypass and PFC-stage parts live.
| Rated voltage | Parts in the 200V–4kV series | Example part | Where it lands in a server PSU | Match / mismatch |
|---|---|---|---|---|
| 200 Vdc | 654 | — | Secondary-side snubber | Match — widest low-HV coverage |
| 250 Vdc | 867 | — | Primary-side filtering | Match — largest single voltage band |
| 500 Vdc | 647 | — | Bulk bypass below the DC link | Match |
| 630 Vdc | 637 | — | 400 V DC bus bypass with margin | Match |
| 1,000 Vdc | 612 | 1206B472K102 |
Snubber across a switching node | Match — held in Hong Kong stock |
| 1,500 Vdc | 29 | — | 800 V bus with 2× derating | Mismatch — only 29 parts, thin selection |
| 2,000 Vdc | 356 | 1206B102K202 |
High-side snubber, isolation bypass | Match — but see the thickness note below |
| 2,500 Vdc | 28 | — | — | Mismatch — 28 parts across all case sizes |
| 3,000 Vdc | 212 | 1808B102K302 |
Isolation barrier bypass | Match — held in Hong Kong stock |
| 4,000 Vdc | 22 | — | — | Mismatch — 22 parts, datasheet check essential |
Two voltage bands are effectively empty. 1,500 Vdc has 29 parts and 2,500 Vdc has 28. A design
that lands on either should either move to 2,000 Vdc or plan around a very narrow selection.
The large-case series adds 501 more part numbers at 1825, 2220 and 2225, of which 83 are at
2,000 Vdc, 71 at 3,000 Vdc and 37 at 4,000 Vdc.
Why does the same Walsin part number appear with more than one thickness?
Because Walsin publishes more than one physical build against a single part number. This is the
most dangerous thing in the high-voltage data for a server power supply, where board-to-board
pitch and airflow clearance are fixed early and a 0.8 mm surprise becomes a 1.6 mm part.
Row in Walsin’s export for 1206B102K202 |
Length | Width | Thickness | Match / mismatch against a 1.0 mm height budget |
|---|---|---|---|---|
| 1 | 3.20 ±0.15 mm | 1.60 ±0.15 mm | 1.25 ±0.10 mm | Mismatch — exceeds the budget |
| 2 | 3.20 ±0.20 mm | 1.60 ±0.20 mm | 1.60 ±0.20 mm | Mismatch — worst case 1.80 mm |
| 3 | 3.20 ±0.15 mm | 1.60 ±0.15 mm | 0.95 ±0.10 mm | Match — but only just, at 1.05 mm worst case |
| 4 | 3.20 ±0.15 mm | 1.60 ±0.15 mm | 0.80 ±0.10 mm | Match — 0.90 mm worst case |
| 5 | none published | none published | none published | Mismatch — no dimensions given at all |
Unique to Venatech’s data. 399 Walsin MLCC part numbers appear more than once in the
24,026-row parametric export, and 381 of those 399 carry more than one distinct thickness
figure.1206B102K202is the worst case: five rows, four different thicknesses — 1.25 mm,
1.60 mm, 0.95 mm and 0.80 mm — plus one row with no dimensions at all. The part number alone
does not determine the part’s height. This count is published nowhere; it comes from counting
duplicates in Walsin’s own export.
The practical rule for an AI server build: specify the thickness on the BOM line, not just the
part number, and ask for it confirmed against the reel you will actually receive. A 1206
1206B222K102 has the same problem in miniature, appearing at 1.25 mm and 1.60 mm.
How much capacitance can a Walsin MLCC give at the point of load?
220 µF, in a 1210 or a 1206, at 4 or 6.3 Vdc. Walsin’s general purpose range tops out there.
Above 6.3 Vdc the ceiling drops to 100 µF, which covers a 1.8 V or 1.2 V core rail with
margin but not much more.
| Part | Case size | Dielectric | Capacitance | Rated voltage | Thickness | Match / mismatch for a 1.2 V core rail |
|---|---|---|---|---|---|---|
1210X227M6R3 |
1210 | X5R | 220 µF | 6.3 Vdc | 2.50 ±0.30 mm | Match — highest capacitance available, tall |
1206X227M6R3 |
1206 | X5R | 220 µF | 6.3 Vdc | 1.60 +0.30/−0.10 mm | Match — same value, 0.9 mm shorter |
1206X107M6R3 |
1206 | X5R | 100 µF | 6.3 Vdc | 1.60 +0.30/−0.10 mm | Match — lower value, same envelope |
0805X107M6R3 |
0805 | X5R | 100 µF | 6.3 Vdc | 1.25 ±0.20 mm | Match — 100 µF in an 0805 footprint |
1210X107M160 |
1210 | X5R | 100 µF | 16 Vdc | 2.50 ±0.30 mm | Match — for a 12 V intermediate bus |
1210S107M6R3 |
1210 | X6S | 100 µF | 6.3 Vdc | 2.50 ±0.30 mm | Match — X6S, wider temperature range |
Every one of these is ±20% or ±10% and all are Class II dielectrics, so DC bias derating
dominates the real delivered capacitance. A 220 µF X5R at 6.3 Vdc operated at 5 V will not
deliver 220 µF. Size the bank on the derated figure from the datasheet curve, not the printed
value.
Which chip resistors handle server power sensing and bleed?
The Thick Film High Power series, 40,255 part numbers, and the Wide Terminal Power series,
4,294. Between them they cover 1/8 W to 3 W and the 2512, 2010, 1225, 1020 and 0612 case sizes
that a bleed resistor or a rail-discharge network actually uses.
| Part | Series | Case size | Resistance | Rated power | TCR | Match / mismatch for a 2 W bleed path |
|---|---|---|---|---|---|---|
WF25P1R00FTL |
High Power | 2512 (6432) | 1.00 Ω ±1% | 2 W | ≦±100 ppm | Match — standard 2 W footprint |
WK25L1R00FTL |
Wide Terminal Power | 1225 (3264) | 1.00 Ω ±1% | 2 W | ≦±200 ppm | Mismatch — same power, 2× looser TCR |
The wide terminal part is the interesting one for thermal control. It rates the same 2 W in a
shorter, wider body, which puts the terminations closer to the copper and moves heat out
faster. The cost is TCR: ≦±200 ppm/°C against ≦±100 ppm/°C on the 2512 high power part. For a
bleed path that does not care about drift, take the thermal win. For a sense element, do not.
The High Power series power breakdown is 10,268 parts at 1/2 W, 7,933 at 1/8 W, 5,147 at 1/4 W,
4,154 at 2 W, 3,324 at 1 W, 3,322 at 3/4 W, 1,950 at 1.5 W, 1,662 at 3 W, 1,660 at 1/3 W and
835 at 1/5 W. Note that the 1.5 W and 1/5 W parts — 3,615 between them once the wider catalogue
is counted — are missing from Walsin’s own public filter dropdown and are invisible to anyone
selecting through the website.
What precision is available for a feedback divider?
±0.01% across every case size from 0402 to 2512, and nothing looser at any of them. Walsin’s thin
film WF series is 58,314 part numbers, and the tolerance grid is not what a designer expects: the
looser grades exist only at 0201, so tolerance and case size are coupled in a way no datasheet
summary makes obvious.
| Case size | ±0.01% | ±0.1% | ±0.5% | ±1% | Match / mismatch |
|---|---|---|---|---|---|
| 0201 (0603) | 0 | 1,674 | 1,916 | 1,916 | Mismatch — no ±0.01% part at 0201 |
| 0402 (1005) | 6,359 | 0 | 0 | 0 | Mismatch — ±0.01% only, nothing looser |
| 0603 (1608) | 7,562 | 0 | 0 | 0 | Mismatch — ±0.01% only |
| 0805 (2012) | 7,985 | 0 | 0 | 0 | Mismatch — ±0.01% only |
| 1206 (3216) | 8,427 | 0 | 0 | 0 | Mismatch — ±0.01% only |
| 1210 (3225) | 7,216 | 0 | 0 | 0 | Mismatch — ±0.01% only |
| 2010 (5025) | 7,436 | 0 | 0 | 0 | Mismatch — ±0.01% only |
| 2512 (6432) | 7,823 | 0 | 0 | 0 | Mismatch — ±0.01% only |
The grid is effectively two separate catalogues. At 0201 the choice is ±0.1%, ±0.5% or ±1% and
there is no ±0.01% part at all. From 0402 upward the only published tolerance is ±0.01%. A
divider specified at ±0.1% in an 0402 footprint has no direct thin film answer in this data; it
has to move to 0201, move to thick film high precision, or accept ±0.01%.
Note also that nine of Walsin’s twenty thin film exports return zero rows, including the whole
AEC-Q200 WF-Q professional series and the TTL high power low-TCR current sensor series. If a
server design needs an AEC-Q200 thin film part, the parametric data does not exist and the
selection has to come from the datasheet.
What is held in Hong Kong for a high-voltage BOM line?
In the 23 September 2026 stock snapshot, 183 distinct MLCC order codes sit at 200 Vdc or above
and 68 at 1,000 Vdc or above. The held high-voltage lines are concentrated at 1206, 1808 and
1812 in X7R and NP0 — exactly the case sizes a snubber and isolation bypass network uses.
Worked examples from that snapshot: 1206B102K202CT at 1206 X7R 1000 pF 2,000 Vdc with a
standard pack quantity of 3,000; 1808B102K302CT at 1808 X7R 1000 pF 3,000 Vdc with a pack
quantity of 1,000; and 2220B104K102CT at 2220 X7R 100 nF 1,000 Vdc. Stock is a dated snapshot.
Treat the shape as stable and the quantities as not.
What delivery terms work for a North American ship-to?
DDP, DAP and EXW, arranged to suit the customer’s terms and destination. The Hong Kong entity
handles offshore trade and foreign currency settlement, so the invoicing currency is a separate
question from the delivery term. Ask for all three quoted rather than assuming one, because which
is cheapest depends on your own forwarder and import handling, not on the components.
FAQ
Does Venatech hold stock in the United States?
No. The group’s operating entities are Hong Kong, Shenzhen, Singapore and the Jiangmen factory in Guangdong, with Chengdu in preparation. Hong Kong is the authorised Walsin distribution entity and the shipping origin for a North American ship-to, with DDP, DAP and EXW delivery terms available.
What is the highest voltage Walsin MLCC available?
4,000 Vdc, but the selection is thin: 22 part numbers in the 4,095-part Middle to High Voltage series and 37 in the 501-part large-case series. The densest high-voltage bands are 250 Vdc with 867 parts, 500 Vdc with 647, 630 Vdc with 637 and 1,000 Vdc with 612.
Why do two Walsin datasheet rows show different thicknesses for the same part number?
Because Walsin publishes more than one physical build under one part number. 399 MLCC part numbers appear more than once in the 24,026-row export and 381 of those carry more than one distinct thickness. 1206B102K202 has five rows and four thicknesses. Specify thickness on the BOM line.
What is the highest capacitance Walsin MLCC?
220 µF, in X5R at 4 Vdc and 6.3 Vdc, available in both 1206 and 1210. Above 6.3 Vdc the ceiling is 100 µF. All are Class II dielectrics at ±10% or ±20%, so DC bias derating governs the delivered capacitance and the bank must be sized on the datasheet curve.
Which Walsin chip resistor gives 2 W in a 2512?
The Thick Film High Power series, which has 4,154 part numbers at 2 W. WF25P1R00FTL is a 1.00 Ω ±1% 2 W part at 2512 with TCR within ±100 ppm/°C. The Wide Terminal Power series offers the same 2 W at 1225 with looser TCR of ±200 ppm/°C.
Does Walsin publish an AEC-Q200 thin film resistor?
Walsin lists a WF-Q professional AEC-Q200 thin film series, but that export returns zero rows, as do eight of Walsin’s other twenty thin film exports including the TTL high power low-TCR current sensor series. The parts may exist; the parametric data does not, so selection has to come from the datasheet.
What tolerance can I get in an 0402 thin film resistor?
±0.01% only. Walsin’s thin film WF grid publishes 6,359 parts at 0402 and every one is ±0.01%. The looser grades — ±0.1%, ±0.5% and ±1% — exist only at 0201. A divider specified at ±0.1% in an 0402 footprint has no direct thin film answer in this data.
Can I get high-voltage MLCC from Hong Kong stock?
In the 23 September 2026 snapshot, 183 distinct MLCC order codes sat at 200 Vdc or above and 68 at 1,000 Vdc or above, concentrated at 1206, 1808 and 1812 in X7R and NP0. Stock is a dated snapshot that moves, so confirm the specific line rather than relying on the count.
Next step
If an AI server or power conversion board has a high-voltage or high-capacitance line that is
hard to place, the thickness is usually the blocker rather than the electrical spec. Send the
BOM lines and Venatech Limited, an authorised Walsin distributor in Hong Kong, will return the
rated voltage, every published thickness for that part number, the pack quantity and whether the
line sits in Hong Kong stock.
Send your BOM to Venatech · Browse the
Walsin chip resistor range · Read
MLCC DC bias and voltage derating and
Wide terminal power resistors: 1225, 1020 and 0612.
Cross it yourself. The cross-reference tool takes a single part number or a whole BOM column and returns the Walsin candidates with every field that differs stated plainly — qualification stays with you. Ready stock shows what is held in Hong Kong today, with an RFQ against any line.