EV Aluminum Electrolytic Cap Mounts: How They Work and Where They Fail
A DC-link capacitor the size of a soda can weighs 300-500 g and sits in a housing that vibrates at 10-2,000 Hz. EV aluminum electrolytic cap mounts are the brackets, clamps and end plates that keep that can in place. This page explains the mechanics, the tolerances that matter and the points where a well-drawn mount still fails in the field.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
What an EV capacitor mount actually has to do
A large-can aluminum electrolytic capacitor is a wound element inside a sealed can. The anode foil, separator and electrolyte are stacked under pressure, and the rubber bung at the bottom holds the whole stack. Nothing inside that can is rigid. Bolt the can to a chassis rail without a compliant interface and vibration passes straight into the winding, where the foil edges rub against each other.
So the mount is not a shelf. It is a spring element with a defined stiffness, a thermal path and a dielectric boundary. The clamp force has to survive thermal cycling from -40 °C to +125 °C without going slack, and it has to stay below the point where the can wall deforms or the bung extrudes.
For EV aluminum electrolytic cap mounts, three load cases drive the geometry: random vibration from the road and the motor, thermal expansion of the can relative to the housing, and assembly torque applied by the line operator. Designs that look fine in CAD often fail the third case, because a clamp rib that is 0.2 mm too tall transfers the torque into the can instead of into the bracket.
The mount also sets the creepage and clearance distance between the can body and any grounded metal. On an 800 V DC-link that distance is not a suggestion. It is fixed by the safety standard the pack is built to, and the mount is what holds it.
Alloy choice for EV aluminum electrolytic cap mounts
Most EV aluminum electrolytic cap mounts start as 6061-T6. It welds, anodizes cleanly, machines fast and holds ±0.005 mm on a well-fixtured 5-axis setup. Thermal conductivity is roughly 167 W/m·K, which is enough to pull heat out of the can base if the contact area is real and not just nominal.
7075-T6 enters when the bracket is also a structural node, for example a mount that ties the capacitor to the inverter housing and carries its own bending load. Yield strength is higher, but the alloy is less forgiving: it is harder to weld, it does not anodize to the same cosmetic standard, and stress corrosion cracking becomes a concern at sharp internal corners in a wet environment. Add radii.
Die-cast ADC12 is the production answer once the design is stable. It gets you near-net shape, integrated cooling fins and threaded bosses in one shot. The trade-off is porosity and dimensional scatter. A cast mount typically needs machined pads and bores for the critical interfaces, so the drawing has to separate as-cast tolerances from machined tolerances rather than calling one blanket value.
Stainless 304 or 316 is rare here because of weight and galvanic pairing with the aluminum can, but it shows up in small retention clips. If a stainless clip touches an aluminum mount, specify a finish on at least one side or expect galvanic corrosion at the contact line.
Flatness, hole position and the thermal runaway link
The can base is the main heat path out of a large electrolytic capacitor. If the mount pad is not flat, contact is reduced to a few high spots and the thermal resistance rises. The capacitor runs hotter, the electrolyte dries out faster, and ESR climbs. That is the slow version of the failure.
The fast version is a contact resistance hotspot on the electrical side. A mount that doubles as a busbar support can shift the terminal position by a few tenths of a millimeter. Terminal stress then concentrates at the weld between the tab and the foil stack. Under ripple current that joint heats, and a hot joint on a DC-link is exactly the condition that ends in thermal runaway.
This is why tier-1 drawings usually call out flatness on the mating face, position tolerance on the mounting holes, and a separate perpendicularity callout on the terminal axis. Blanket general tolerances hide the one dimension that matters.
Machining capability sets the floor here. Our 5-axis centers hold ±0.005 mm and Ra 0.8-1.6 μm on aluminum mating faces, with 100% inspection before shipment. The point is not the number on the certificate. It is that the flatness callout is measured on every part, not just the first article.
Surface finishing: where thermal and dielectric needs meet
Anodizing is the default finish for aluminum mount hardware, but the type matters. Clear anodize is a thin dielectric layer and gives modest corrosion protection. Hardcoat anodize is thicker and much more wear resistant, which is what you want on a clamp face that slides during assembly. Conductive anodize exists precisely because the other two insulate, and sometimes you need the mount to be a ground path.
If the mount carries current or provides a ground bond, an insulating anodize layer is a defect, not a feature. Mask the contact areas or specify conductive anodize and check the resistance after coating. A 20 μm hardcoat layer on a grounding pad will pass visual inspection and fail a bond test.
Bead blasting before coating controls gloss and removes machining marks, but it also changes the surface roughness that a thermal interface relies on. For mounts that use a gap pad or thermal grease, keep the mating face as-machined at Ra 1.6-3.2 μm and mask it from blasting.
Laser marking is the practical way to keep traceability on the part. Minimum character height is 1.5 mm on our equipment, which is large enough to survive anodize and still be readable in a service bay.
Where EV aluminum electrolytic cap mounts fail in service
The most common field failure is not a cracked bracket. It is a loose one. The clamp force drops because the elastomer insert took a compression set, or because the bolt lost preload through differential thermal expansion. The capacitor starts to move by a fraction of a millimeter, and that is enough to fret the can surface.
The second is creep of the aluminum itself. A clamp arm loaded near yield at 125 °C will slowly deform. Design the arm so that the operating stress stays well below yield at the top temperature, and add a rib instead of thickening the whole section. Thick sections hold heat and distort during machining.
The third is a machining-driven failure: a hole pattern that is fine on the nominal part but drifts across the lot. If hole position varies by 0.15 mm and the mating bracket is also at its limit, assembly forces the can into a bind. The operator torques the bolt harder to close the gap, and the can wall deforms.
The fourth is corrosion at the interface between a stainless fastener and an aluminum mount in a humid underhood environment. It shows up late, after the warranty period in some cases, and it loosens the joint from the inside.
What to check before outsourcing the mount
Ask how many setups the shop plans. A mount with a machined bore, a flat pad and a terminal clearance slot needs at least three faces. If the answer is three separate 3-axis setups, budget for alignment error and plan on a looser bore-to-pad relationship. One 5-axis setup is not a luxury here, it is the tolerance strategy.
Ask what happens to the mating pad during finishing. A shop that anodizes the whole part without masking will hand you an insulating layer exactly where you need electrical or thermal contact. That is a process question, not a drawing question, and it should be answered before the first part is cut.
Ask for the inspection report, not just the certificate. On a safety-relevant mount, the flatness and position numbers matter more than the ISO logo on the letterhead. Our quality system covers ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022, and we inspect 100% of parts before shipment.
Finally, ask about the prototype-to-production path. Machined 6061 for validation and die-cast ADC12 for volume is a normal sequence. If the shop can only do one of the two, you will rebuild the fixture and the quality baseline at the worst possible time.
How we machine a capacitor mount from drawing to shipped part
This is the sequence for a typical 6061-T6 mount with a machined bore, a flat mating pad and threaded inserts.
- 1Review the drawing and run a DFM passWe check flatness callouts, hole position tolerances, wall thickness at the clamp ribs and whether the terminal axis has its own perpendicularity callout. Quotation and free DFM analysis come back within 12 hours.
- 2Pick the alloy and stock form6061-T6 plate for machined prototypes. If the drawing specifies a cast body, we flag the machined interfaces that must be toleranced separately from as-cast surfaces.
- 3Set up 5-axis workholdingClamping ribs, undercuts and the mounting boss usually sit on three or more faces. One 5-axis setup cuts them together and removes the alignment error that stacked fixtures introduce.
- 4Cut the critical faces and boreMating pad flatness and the capacitor bore are machined in the same setup at ±0.005 mm. Roughing leaves 0.3-0.5 mm for finishing to control heat distortion in thin ribs.
- 5Deburr and prepare for finishHand deburr at the terminal clearance and the rib roots. Mask the mating pad and any ground contact before anodize, bead blast or laser marking.
- 6Inspect and shipRaw material check, in-process monitoring, final inspection on 100% of parts. Reports on request. Parts ship in 3-5 days.
Alloy and process comparison for capacitor mounts
Ranges reflect typical automotive EV practice, not a guarantee for every program.
| Option | Best for | Watch out for | Typical finish |
|---|---|---|---|
| 6061-T6 machined | Prototypes and low-volume mounts | Lower yield than 7075 under bending | Clear or hardcoat anodize |
| 7075-T6 machined | Structural mounts with bending load | Stress corrosion at sharp corners | Hardcoat anodize |
| ADC12 die cast | Runs above a few thousand parts | Porosity at thin ribs and bosses | Bead blast plus machined pads |
| 6063 extrusion | Long straight rails and spacers | Loose tolerance on the cut ends | As-machined or anodized |
| 5052 sheet | Thin clamp straps and covers | Springback after forming | Powder coat or anodize |
| 304 stainless clip | Small retention features | Galvanic pair with aluminum | Passivation |
Symptom, cause and corrective action
| Symptom | Likely cause | What to do |
|---|---|---|
| Capacitor moves in the clamp | Elastomer compression set or lost bolt preload | Switch to a higher-temperature elastomer, add a belleville washer |
| Can surface fretting marks | Micro-motion below the visible threshold | Increase clamp stiffness, add a compliant liner |
| Rising ESR over service life | Thermal path degraded by poor pad flatness | Tighten flatness callout, mask pad from coating |
| Hot terminal after ripple test | Terminal position shift from loose hole tolerance | Add position tolerance on the mounting holes |
| Crack at rib root | Stress concentration plus 7075 corrosion | Add fillet radius, switch finish, review stress |
| Rust staining near fasteners | Galvanic pair between steel and aluminum | Specify finish on one side, use coated fasteners |
When to machine and when to cast
If you are still tuning clamp force or the terminal geometry, machine 6061-T6 and keep the design free to change. Once the drawing is frozen and volume passes a few thousand parts, move to die-cast ADC12 with machined interfaces and accept the porosity risk in exchange for near-net shape and integrated fins.
Questions engineers ask about capacitor mounts
How flat does the capacitor mating pad need to be?
It depends on whether the pad is a thermal interface or just a locating surface. For a thermal pad, flatness in the 0.05-0.10 mm range over the contact area is a common starting point, and the surface should stay as-machined so a gap pad can conform.
If the pad only locates the can, a looser callout works. The mistake is applying one blanket tolerance to both and losing the thermal path in the process.
Can the mount be anodized if it also carries a ground bond?
Not with standard clear or hardcoat anodize. Both are dielectric. Mask the grounding contact area, or specify conductive anodize and verify resistance after coating.
A 20 μm hardcoat layer will pass a visual check and fail a bond test, so the inspection step has to be defined on the drawing.
What is the risk of using 3-axis machining instead of 5-axis?
Each extra setup adds a datum shift. If the bore and the mating pad are cut in different setups, the relationship between them carries the sum of both fixture errors.
For a mount where terminal alignment matters, that stack-up can push the terminal out of position and concentrate stress at the tab weld.
Does the clamp need a compliant element?
Usually yes. Aluminum against aluminum with a bolted joint loses preload as the joint cycles thermally. A thin elastomer or a spring washer keeps the force within a usable band across -40 °C to +125 °C.
Pick an elastomer rated above the top operating temperature, or it will take a compression set and the clamp will go slack.
How do you keep the prototype and the production part equivalent?
Define which surfaces are machined on both versions and hold those to the same tolerance. The as-cast surfaces can be looser, but the bore, the pad and the hole pattern should not change between prototype and production.
Keeping both under one quality system avoids rebuilding the inspection baseline when the process changes.
Send the drawing and get a DFM review
Upload the mount drawing and we will return a quotation with free DFM analysis within 12 hours. We machine from one prototype to 10,000+ part runs, with no minimum order quantity and 100% inspection before shipment.
12-hour quote±0.005 mm toleranceIATF 16949:2016NDA on request
More from the shop floor
We publish setup notes, tooling trials and inspection data from the factory floor.
