Hi. Does anyone know of a manufacturer making small, low cost off the shelf, inductors designed for withstanding say 400V for 0.5µs repetitively for many years? I am concerned about turn-to-turn breakdown in standard inductors. The ones I have un-wound so far have had the start and finish of the winding overlapping. Manufacturers don't specify the working voltage.

I have seen this too with start and stop overlapping. I hasn't stopped people from using them, but I have always flagged it as a failure risk. There is no reason they can't wind them differently, but they are not usually aware of the need.

Any of the magnetics vendors out there, perhaps you can respond to this need.

Also, normal low-cost surface-mount inductors rely on the ferrite core as an insulator, but I know from bitter experience that ferrite can break-down spectacularly. I have no way of predicting at what voltage ferrite will break-down. Another one for any magnetics vendors out there...

You can add PCB spark gaps to divert a voltage surge. I did a project where they needed high impedance but also needed to meet surge, we used gas discharge tubes to protect the inductor and a MOV on the other side. The inductor will fail during a surge event if not properly designed, magnet wire insulation is only rated for what the manufacturer lists depending on single, double, heavy type of coating.

Hi Stephen,
The design guidelines used by a magnetics vendor I have worked with for years recommends no more than 120V between turns when using heavy build magnet wire.
Layer insulation must be specified by the designer to prevent high voltage turns on adjacent layers from being in contact as well as sleeving on lead breakouts if needed. If using a toroid, progressive winding may be used to reduce layer to layer voltage and capacitance which has the added benefit of raising the self-resonant frequency of the inductor.

What kind of voltage waveform is applied? Is it a resonant half-sinusoid, or a square-wave like from a HV buck converter? Inter-turn or inter-layer voltage gradient guidelines should be a function of dv/dt: a steep dv/dt may cause repetitive partial-discharge or corona, which can degrade insulation and result in delayed failure. One way to improve the voltage gradient capability is to vacuum encapsulate the windings as is common for high-density, high voltage power supplies, but this is not an inexpensive process.

It must be custom made. There is nothing unusual in this application but I am not sure if somebody actually manufactures such kind of inductors. Nelson can definitely help but it is a custom component anyway. I also know several manufactures which are capable of doing it but it will be custom made product.

The worst case in this regard occurs when the start of a winding on a flanged bobbin is routed radially outward for termination. Some bobbons do not have deep slots to accommodate the winding start termination, and we then have a situation where the termination for the start winding on the inner flange surface of the bobbin is at right angles to the edges of the winding. Nothing can be more dangerous, as the force loading from resulting point contact is very high. During temperature cycling, sooner or later the varnish coating will wear through and the winding will short circuit. True, cambric or another insulating material can be used for the radially outward section of the start winding, but it will age or cold flow over time, particularly at high temperatures. I've been successful in drilling a small hole in the bobbin wall, just above the winding surface for entrance of the winding start end. Bobbins like those made for some of the Mag-Inc E-cores are particularly notorious regarding absence of deep winding exit slots.

Winding failure due to overlapping start and finish / phase crossovers is a very common failure on motors even without PWM drive which adds more issues. I agree with the consesus above that this will probably require a custom part. I looked at doing a low power and cost PSU for 5V off rectified mains formed from a buck at a very low duty cycle using a power integrations part (Linkswitch) quite some time ago. In the end I used a little axial rod choke.

Hi. Some great comments. I had convinced myself that I would have to design a custom inductor until Mr Arulaane suggested the Würth parts. I will try the Würth parts first but design a custom part otherwise maybe on a single layer solenoid for simplicity.
The fact that Würth have created a special choke confirms to me that standard chokes are not up to the job.

Maybe the excellent people in this group could suggest an accelerated ageing test? I was thinking a 4x frequency series resonant circuit to get say 2x voltage across the choke. Is there a more scientific way? I guess my test won't detect the cold-flow and wire chaffing issue suggested by Mr Kernei.

Stephen I am not sure of a standards dictated test for this but I would add elevated humidity to this to gain confidence unless of course you are confident of your humidly levels.

There is magnet wire and then there is magnet wire. The type we use in my Mil/Hi-Rel designs uses a (usually heavy) coating called either: a) PTX (polythermalize), b) PTZ (same), or c) GP-200. This is quite different from the usual solderable magnet wires, but the amount of voltage stress turn-to-turn is much higher than the cheap stuff. Turn-to-turn of 1 KV is no problem (that's 2 layers of the coating, one on each of the two touching wires). However, we usually don't push it anywhere near that hard. I completely agree that the magnetics shop needs to have a good spec from the customer so they can factor-in those sorts of issues during the design. And it's important to make sure you use a good quality magnetics shop, whether for standard or for custom magnetics parts. Call them up and talk to the applications engineer or better yet, the actual designer if they let you. If they don't let you, it may be time to consider using another magnetics shop that will address your concerns and let you talk to the engineer that will be or has design(ed) the part you need.

The conclusion appears to be to design the part myself in collaboration with the magnetics vendors and test to destruction.
Many thanks for all your comments.