Cycle-proof cooling for 24/7 continuous operation

Yes. Water coolers for servo drives and inverters are a recurring case for us, often as a retrofit when an air-cooled drive in a sealed control cabinet reaches its limit. Enquiries come in regularly for drive series from well-known manufacturers, sometimes on the recommendation of the drive manufacturer itself. What we need is not the drive type alone but the heat load, the position and size of the cooling surface on the device and the hole pattern. The reason: the same series is built in different power classes, and the cold plate has to fit the specific device, not the name. Limit: we do not have a catalogue item called "cold plate for drive X". We manufacture it to fit, which drives the price for single pieces but covers every device variant.

Yes, cold plates in the 19-inch format are a recurring case in test bench construction, typically as a plate onto which test items or power assemblies are mounted. Real enquiries are around 430 × 800 mm with up to 2,500 W per plate at a 15 °C inlet temperature and a plate temperature that must not exceed 75 °C. That is a readily solvable task because the heat is distributed and the inlet temperature is low. What matters is the mounting side: if test items are changed frequently, the plate needs a hole or thread pattern rather than individual holes, and the surface has to stay flat. Limit: with frequent reconfiguration a plate with threaded inserts lasts longer than threads cut directly into aluminium, because threads in soft material wear with repeated assembly.

A cold plate removes heat to wherever the coolant carries it, so the limit is set not by the plate but by the circuit: volume flow, inlet temperature and the capacity of the heat rejection unit. For orientation from real projects: designs from a few hundred watts to over 10 kW per plate are everyday work, in one case with two IGBT modules on a shared plate. The decisive factor is the heat flux density at the source, not the total power: 2 kW over 300 cm² is uncritical, 2 kW over 30 cm² calls for an internally structured plate with a channel directly beneath the source. Limit and a frequent misunderstanding: a cold plate does not cool below the temperature of the coolant. If you need component temperatures below room temperature, you need a chiller, not a cold plate.

We face-machine the mounting face and, on request, measure and record the flatness achieved. We regularly supply flatness measurement reports and records of hole spacing to series customers. Why that is more than a formality: on a power module, flatness directly determines the contact resistance, and no thermal paste compensates for a tenth of a millimetre of waviness. State in the drawing over which area the flatness applies: across the whole plate or only across the module footprint. On large plates that is a considerable difference in effort and price. Limit: very tight flatness over long lengths requires machining after joining and in some cases stress-relief annealing; that extends throughput time and should be planned in early.

Yes. Integration into existing parts is one of our most frequent applications: we bring cooling channels directly into a choke box, a production tool or a magnet system instead of attaching a separate cooler. The advantage is the heat path as much as the installation space: every separation between part and cooler costs temperature, and an integrated solution does not have that separation. The prerequisite is enough wall thickness for the channel routing and a part that can still be machined after joining. What we need for it: the existing part as a STEP model and a statement of which faces are to be cooled. Limit: we cannot press or weld into parts made of castings or of unknown material without material testing. A bolted-on plate is then the safe route.

In test and measurement, reproducibility is often more important than the absolute value: all plates have to behave the same so that measurement results stay comparable. For series we therefore fix the process parameters and inspect the critical features (flatness, channel geometry, leak tightness) to a defined inspection plan, and we supply the measurement reports with them. What creates scatter is rarely the machining but the raw material: different batches and especially different material forms such as cast and rolled material behave differently during joining. For reproducibility-critical series we therefore fix the material form and document the batch. Limit: if you need a scatter figure for thermal resistance, it has to be determined by measurement. Series scatter cannot be derived from a simulation.

The pressure drop is not determined by the plate alone but mostly by bends, transitions and fittings: in one customer project the pressure drop fell significantly simply by leaving out the rotary joints at the connection. That is the most important practical insight: if you want to reduce pressure drop, start at the connection routing, not at the channel structure. For designs we state calculated curves; measurement on the real set-up is more accurate because your adapters and hoses contribute. Many customers measure pressure drop against volume flow themselves and characterise their adapters in the process, which is a sensible approach. Limit: pressure drop and thermal resistance work against each other; a plate with a very low pressure drop cools less well at the same geometry.

Yes, several consumers on one circuit are common. The only question is whether they sit in series or in parallel. In series, each consumer heats the medium for the next, so the last in the chain has the worst conditions; in parallel, every consumer gets cold medium, but the split of volume flow has to be controlled via the channel cross-sections, otherwise the flow takes the path of least resistance. With set-ups combining consumers of very different temperature (drive controllers and braking resistors, for example) we deliberately fix the order: the most temperature-sensitive component comes first. Limit: with widely differing heat loads on one circuit, a hydraulic design of the overall system is necessary; optimising the individual plate alone is then misleading.

We start from your load case: heat load, distribution across the area, permissible component temperature, inlet temperature, volume flow and installation space. This yields the Rth budget. We test it in thermal simulation against variants of the channel routing before design work begins.

We calculate with the load profile across the shift, not with the data sheet maximum, and plan reserve for contamination and ageing of the circuit. A cooler that fits exactly when new is too small after two years.

Copper, copper-nickel, stainless steel and aluminium. The choice follows three criteria: thermal conductivity, media compatibility and mechanical requirement. We often combine materials: conductive material at the heat source, resistant material at the medium.

Pressure testing of every part is our standard; for demanding applications we additionally test with a helium leak test and document the leak rate per part. We set the target leak rate before design.

From a 30 mm microchannel cooler to a 3 m (30,000 mm) cold plate, and up to a fully cooled large precision part. What is possible is not decided by size alone but by material, design and tolerance: give us your dimensions and we will tell you in which design we can produce them.

Yes: stators, spindles, housings and structural parts. In machine tools, thermal drift of the structure is often a bigger lever than cooling the electronics.

Through geometry: no dead volumes, adequate cross-sections, flushable channel routing. We also match material and medium, because deposits usually arise from the material pairing.

Samples in around 6-8 weeks as standard, 2-4 weeks by express. For series production we manufacture in a highly automated plant in Germany; cost-sensitive standard parts come from our joint ventures in Asia.

Yes, that is the norm for us. We sign your non-disclosure agreement or provide ours. We never pass on customer projects or application details.