Encapsulation resin accounts for approximately 4–8% of the material costs of a medium-frequency transformer (MFT); when expressed as a percentage of the total SST assembly cost, this amounts to 3–5%. The following section breaks down this cost structure step by step.
First: the most expensive component in the entire system is not SiC.
SST is divided into three stages: rectification, isolation, and inversion. The component that incurs the majority of the cost is the core component of the intermediate stage: the medium-frequency transformer. According to publicly disclosed information from manufacturers, the MFT accounts for 60%–70% of the total equipment cost and is also the most technically challenging component. What makes it so difficult? For the same power rating, the MFT must be packed into a volume less than one-tenth that of a conventional transformer. There is little dispute regarding this stage. The real ambiguity lies in the next stage: who else takes their share of the 60%–70% accounted for by the MFT?
Second: magnetic cores and copper wire account for 60% of the total.
The raw-material composition of high-frequency isolation transformers is actually quite straightforward: by project, amorphous/nanocrystalline cores account for 33%–35% of MFT costs; windings (Litz wire) account for 30%–35%; insulation materials, potting materials, and structural hardware account for approximately 30%–37%; and the combined cost of the core and windings exceeds 60%. This explains why whenever copper prices fluctuate, the entire industry follows suit: copper is used in the windings, which constitute about 30% of the total; when copper prices rise by 30%, the transformer’s BOM is driven up by roughly 10 percentage points. The remaining 30% contains insulation paper, the core frame, clamps, the housing, and the potting compound.
Third: How many points are there in the adhesive?
At this stage, publicly available data are no longer presented in separate sections. No manufacturer would list “the percentage of potting compound” in their promotional materials; however, this can be inferred. For dry-type transformers, the following can be used as a reference: epoxy resin (cast material) accounts for 8%–12% of the total equipment cost, which also includes a portion of the insulation system. In contrast, the potting layer of an MFT is much thinner and uses far less material than that of a power-frequency dry-type transformer; it falls into the category of “low volume, high unit price” materials. If we break it down as 10%–15% for structural components, 5%–8% for insulation materials, and 4%–8% for potting compound, the remaining 30% roughly corresponds to this breakdown. Thus, for an MFT: internal potting compound accounts for 4%–8%; for SST: 60%–70% × 4%–8% = 2.4%–5.6%—that is, 3–5 percentage points. This figure is a calculation rather than a specific quote from any particular manufacturer. Nevertheless, it aligns with the practical experience gained from conversations with MFT manufacturers in the industry: potting compound is not a primary cost item and is often categorized under “auxiliary materials” in quotes.
Why is 3% worth introducing?
Because the cost proportion and the failure rate follow two entirely different curves, the adhesive accounts for only 3–5 percentage points of the BOM. However, after an SST undergoes several hundred hot and cold cycles, the first point where problems typically emerge is not the magnetic core or the windings, but cracking of the casting layer and interfacial delamination. I have seen far too many cases where, in an effort to save a few hundred yuan on adhesive
by switching to a cheaper material that costs only two percentage points less, the casting layer begins to crack after only 300 cycles of hot and cold cycling, forcing the entire batch of prototypes to be reworked; the cost of one rework is enough to purchase adhesive for several years. This is the dilemma of this stage: it is so inexpensive that it is hardly worth selecting carefully; yet it is so critical that choosing the wrong material means starting over from scratch.
Who is funding this: four floors—don’t just focus on the first floor.
Regarding material supply, I categorize it into four types based on the mode of cooperation, which is more useful than simply comparing parameters by brand. The first tier: International specialized systems. Huntsman is the most prominent player in this category; Huntsman Araldite CW5879/HW5879 is a Class H epoxy vacuum casting system specifically developed for “SST high-frequency windings with SiC integration,” with a nominal coverage of 20–150 kHz. Furthermore, Huntsman has defined its own application boundaries: for conventional scenarios where the hotspot temperature is ≤120 °C, it offers a cost advantage over high-Tg systems. Henkel’s STYCAST represents another line; it is more expensive but offers comprehensive certification, making it widely used in export and automotive-grade projects. The second tier: Domestic comprehensive manufacturers, represented by the JOME brand, which offers both epoxy and silicone lines. JOME’s JOME6307 is a general-purpose epoxy benchmarked against STYCAST, designed for medium-to-high voltage windings; JOME6318LB is a high thermal conductivity, low CTE version, which explicitly states on the datasheet that it requires heating and curing above 80 °C followed by full post-curing. While these process prerequisites are typically listed at the very bottom of the datasheet, they are the most critical factors. JOME produces both its epoxy (GREPOXY™) and silicone (GRESIL®) lines, providing long-term support for motors, batteries, and power electronics. The key difference from the first two tiers lies in the mode of cooperation: formulations are more frequently co-developed and tuned based on specific operating conditions, rather than simply applying a ready-made grade. The fourth tier: Regional supporting manufacturers. In the Yangtze River Delta and Pearl River Delta regions, there are several specialized plants that focus exclusively on power insulation potting; they offer rapid prototyping and can provide process optimization services on-site. Some already possess dedicated SST formulations, making them ideal for rapid iteration during the prototype stage.
The first point of contact for the adhesive is not, in fact, the complete machine manufacturer.
Many people immediately turn to Delta, Sungrow Power Supply, and China XD Electric. The direction is correct, but the sequence is wrong. The resin for SST complete machine manufacturers is basically supplied by medium-frequency transformer suppliers. The ones truly engaged in resin selection, resin testing, and cold/heat cycling are these companies: Xinte Electric (301120), Wuxi Xien Electric, Jingquanhua (002885), Igor (002922), Kelike, Maoshuo Power Supply, Changzhou Yinhe Century Microelectronics, and Feilo Audio Special Electromagnetic Division. Among them, Xinte Electric deserves special mention: it specializes in high-frequency transformers for SST and medium-voltage high-frequency isolation transformers for data.
centers. In collaboration with Vertiv (NVIDIA’s official power partner), it co-developed a 2MW rack-mounted SST, which progressed from prototype verification to small-batch production by June 2026, and Vertiv placed its first small-batch prototype orders. What does reaching “small-batch” production signify? The formulation is essentially finalized. Another recent update: Tianzheng Electric’s first SST entered mass production on October 16. The moment when complete machine manufacturers begin production coincides with the period when transformer suppliers expand capacity and relock their supply chains.
Why are there so few grades available for this type of rubber?
There are many epoxy manufacturers worldwide, but only a handful can provide a grade specifically designed for SST high-frequency windings. The reason is simple: a single epoxy resin must undergo thermal cycling, partial discharge testing, and long-term aging—all of which can take years. While a complete equipment manufacturer may release several product variants per year, the specialized grades available from material suppliers remain limited to just a few. Therefore, when selecting a material, the most important question is not “How good is its thermal conductivity?” but rather “For which operating conditions was this grade originally designed?” If epoxy from a power-frequency transformer is used in a 100 kHz intermediate-frequency transformer, no apparent problem may be visible in the parameter sheet; however, once the material has undergone several hundred thermal cycles, the issue becomes apparent.
Finally, let’s take a final look at the overall picture.
Based on 3–5 points per complete unit, the material cost for a 2.5 MW SST typically ranges from several thousand yuan to just over 10,000 yuan, depending specifically on the encapsulation volume and unit price. For an item costing only a few thousand yuan, it determines whether a piece of equipment costing several hundred thousand yuan can operate for ten years. Regardless of how this cost is calculated, it should never be treated as a “supporting material” on the procurement schedule.