The part does not exist
The bill of materials names a custom magnetic built for the demonstration board, with no public part number and no distributor. It was never meant to be bought.
Home / Services / Reference Designs
Reference Design Magnetics
A semiconductor reference design gets you most of the way, and then calls out a transformer with a part number you cannot buy, or one that was chosen for a demonstration rather than for your product. ConverNova designs and supplies that part. Based in Madrid, Spain, engineering answer usually within 24 hours, prototypes in 4 to 8 weeks, no minimum order quantity. Send the reference design number and the schematic page, and we take it from there.
The gap in the box
The silicon is documented. The magnetics rarely are.
Three versions of the same problem, and we see all three every month.
The part exists but is wrong for you
It was chosen for a 25 degree lab bench and a fixed load. Your product runs at 85 degrees, at partial load most of the day, in a housing with no airflow.
The design has moved on
You changed the switching frequency, the input range or the power level, and the magnetic from the reference no longer suits the converter you actually built.
What we need
Send the reference, not just the part number.
The reference design document tells us more about what the magnetic has to do than any specification sheet would.
Send us your spec- The reference
- The design number or the document. Most are public, and the schematic and waveforms in them are exactly what we work from
- What you changed
- Frequency, input range, power, load profile. If nothing changed, say so, because that is useful too
- Your environment
- Ambient temperature, cooling and the enclosure. This is usually where the reference part stops being suitable
- Your constraints
- Height, footprint, keepout, terminations and the standard you qualify against
- Your volume
- Whether this is a demonstrator, a first build or a production programme changes what we would design
Named designs
The reference designs engineers bring us.
If you are evaluating one of these, the semiconductor is fully documented and the magnetic component usually is not. These are the designs we are asked about most often, with the part each one actually needs wound. Send us the reference and the operating point you are moving to, and we aim to have the first proposal back inside 24 hours.
Texas Instruments
TIDA and TIDM designs, plus the PMP series
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| TIDA-01604 | 6.6 kW totem pole PFC for an HEV or EV onboard charger, 98.6 percent efficiency | PFC boost inductor, sized for ripple and for saturation at the line peak | Three interleaved phases only cancel ripple if the three inductors match each other. A catalogue part comes with a tolerance band, not as a matched set. |
| TIDM-02002 | 6.6 kW bidirectional CLLLC resonant dual active bridge for an onboard charger, 500 kHz | Main transformer with the leakage inductance as part of the resonant tank | The leakage is the resonant element, so it is specified to a window rather than minimised. Nobody sells a transformer to a leakage tolerance. |
| PMP21999 | 6.6 kW bidirectional CLLLC resonant converter using a PCB winding transformer, 500 kHz | Planar transformer, where the artwork fixes the leakage inductance | The winding is the board artwork, so the magnetic and the PCB are one part. There is nothing to order from a catalogue. |
| TIDA-010054 | 10 kW bidirectional dual active bridge for level 3 EV charging, 100 kHz planar magnetics | Planar transformer plus the DAB series inductance | The series inductance has to land on the value the control loop was tuned for, and the planar height is already fixed by the board stack. |
| TIDA-01606 | 11 kW bidirectional three phase three level T-type inverter and PFC | Boost and output filter inductors carrying heavy DC bias | Heavy DC bias takes the core towards saturation long before the thermal limit arrives, so the gap and the material decide the part, not the inductance. |
| TIDA-010210 | 11 kW bidirectional three phase ANPC on GaN, switching at 100 kHz | Filter inductors, small because of the frequency and hot because of it too | At that frequency the part is small enough that there is no surface left to lose the heat through. Thermal design comes before magnetic design. |
| TIDA-010938 | 10 kW GaN single phase string inverter with battery energy storage | Boost inductor and grid side filter inductors | Grid side filtering has to meet a conducted emissions limit, and that limit is met by the winding geometry as much as by the inductance value. |
| TIDA-010933 | 1.6 kW four input bidirectional GaN microinverter with storage | Isolation transformer and input inductors, at very tight volume | When volume is the binding constraint, the only way through is a bespoke window and a termination designed around whatever space is left. |
| TIDA-010966 | Bidirectional isolated dual bridge series resonant DC-DC for residential storage, 40 to 60 V | Low voltage high current transformer plus the resonant inductor | Low voltage and high current means few turns in wide conductors. That is foil, and nobody stocks foil wound to your window. |
| TIDA-011012 | 50 kW SiC based modular solid state transformer | Medium frequency cell transformer, designed for partial discharge as well as hipot | Designed to a partial discharge level and not only to a hipot. That is an insulation system decision, and it is taken at the winding. |
| TIDM-02021 | 11 kW three phase bidirectional onboard charger with a 2.5 kW low voltage DC-DC | Single stage series resonant transformer facing the mains | One stage from AC to DC puts all of the isolation into a single transformer that sees the grid. Creepage and clearance set the build before the losses do. |
| TIDM-02013 | 7.4 kW EV and HEV bidirectional onboard charger | Bidirectional transformer working over two different voltage ranges | The same winding runs in both directions over different ranges. Sizing it for the charging direction alone is the classic mistake. |
| TIDA-010081 | 1 kW AC-DC with a CCM PFC front end and an isolated LLC stage | CCM PFC inductor and the LLC transformer, in series | Two magnetics sharing one thermal budget in one enclosure. Split it badly and the supply runs hot without either part being wrong on its own. |
| TIDM-1001 | 500 W two phase interleaved LLC resonant converter | Two resonant tanks that have to match each other | Interleaving only works if the two tanks are the same. Leakage tolerance stops being a line in a datasheet and becomes the specification. |
| TIDM-RESLLC-DCDC | 300 W digitally controlled LLC half bridge with synchronous rectification | LLC transformer with a low resistance secondary | With the diode drop gone, winding and interconnection resistance are what is left to argue about, and both are geometry decisions. |
| TIDA-00512 | LLC resonant half bridge DC-DC with digital control, 350 to 400 V in and 12 V out, 340 W | High ratio transformer with a high current secondary | Few secondary turns carrying a lot of current is the point where foil starts beating wire, and foil is wound to your window or not at all. |
STMicroelectronics
STEVAL and STDES boards
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| STDES-7KWOBC | 7 kW onboard charger: interleaved totem pole PFC and a dual full bridge LLC stage | Two 510 uH PFC inductors and the LLC transformer | Two inductors at a declared value means two parts that have to agree with each other, not two parts that each meet a tolerance on their own. |
| STEVAL-DPSTPFC1 | 3.6 kW bridgeless totem pole PFC with digital control and inrush limiting | Boost inductor, and the current ripple decides its core | The ripple, not the average current, is what the core sees. A part chosen from an inductance table has been chosen on the wrong number. |
| STDES-2KW5CH48V | 2.5 kW 48 V industrial battery charger: boost PFC on the L4984D into a full bridge LLC on the L6599A | PFC inductor and the LLC resonant transformer | A 48 V output at this power means a secondary that is mostly copper cross section. The winding sets the size of the part, not the core. |
| STEVAL-ISA172V2 | 2 kW fully digital AC-DC: interleaved PFC into a phase shifted full bridge with ZVS | Interleaved PFC inductors, PSFB transformer and the output choke | Three magnetics in one box, each one heating the other two. That thermal interaction is exactly what a catalogue part cannot know about. |
| STDES-DABBIDIR | 25 kW bidirectional dual active bridge for EV charging and storage, 800 V in and 400 V out at 100 kHz on ACEPACK 2 SiC modules, 98.4 percent peak | Dual active bridge transformer and the series inductance that carries the power transfer | In a dual active bridge the series inductance is not a parasitic, it is the element that sets how much power moves at a given phase shift. You do not pick that off a shelf and hope. |
| STDES-6KWHVDCDC | 6 kW full bridge LLC for EV charging infrastructure, 700 to 800 V in and 200 to 1000 V out, 98 percent peak, with the secondary interchangeable between centre tapped and full wave | LLC transformer in two variants, single or dual, plus the gate drive transformer | ST ships one PCB that accepts two different magnetic solutions, because the output range you want decides which. The magnetic is not a component of the variant, it is the variant. |
Infineon
EVAL and REF boards
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| REF-DAB11KIZSICSYS | 11 kW bidirectional CLLC resonant DC-DC on CoolSiC, for onboard and offboard charging | Resonant transformer, with the leakage inductance treated as a circuit element | In a DAB the leakage carries the power. It is designed to a value, and a transformer bought on turns ratio alone will not transfer the power you expect. |
| REF-11KW-PFC-SIC-QD | 11 kW and 7.3 kW multilevel ANPC bidirectional PFC, single and three phase | Boost inductors, with the multilevel waveform changing the ripple spectrum | A multilevel waveform moves the loss into harmonics no datasheet curve covered. The core material has to be chosen against the real spectrum. |
| EVAL-3K3W-TP-PFC-SIC | 3.3 kW totem pole PFC on CoolSiC | Boost inductor, the part that usually sets the height of the board | When the inductor sets the height, the only route to a thinner product is a different winding geometry. That is a redesign, not a reselection. |
| EVAL-5K5W-3PH-LLC-SIC2 | 5.5 kW three phase interleaved LLC, 380 to 420 V in and 50 V out on 650 V CoolSiC G2, close to 99 percent and 170 W per cubic inch | Three LLC transformers with their resonant inductances, matched to each other | Three interleaved phases only share the current if the three magnetics are the same. The tolerance between units stops being a line on a datasheet and becomes the specification. |
| EVAL-3KW-2LLC-C7-20 | 3 kW dual phase LLC for the isolated high voltage DC-DC stage of a telecom or industrial supply, 600 V CoolMOS C7 with digital control | Two LLC transformers and their resonant inductances | Two phases instead of three does not make the matching easier, it makes any mismatch show up as a current imbalance you can measure on a scope. |
| REF-8KW-HFHD-PSU | 8 kW high frequency, high power density supply for AI data centre racks | Isolation transformer designed to the rack slot it has to fit | Power density is a volume target before it is an electrical one. A magnetic designed to a box is a different exercise from one chosen from a catalogue of boxes. |
onsemi
SEC evaluation boards
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| SEC-6D6KW-OBC-TTP-GEVB | 6.6 kW onboard charger power stage, interleaved totem pole PFC with 650 V MOSFETs and SiC, 97 percent typical | Interleaved totem pole PFC inductors | Interleaving at this power only pays if the phases share current. Matched inductors are a manufacturing decision, not a purchasing one. |
| SEC-3PH-11-OBC-EVB | Three phase 11 kW onboard charger platform, PFC and LLC with AEC qualified 1200 V 80 mohm SiC, above 98 percent PFC and 97 percent LLC | Three phase PFC inductors and the LLC transformer | Automotive qualification reaches the magnetic too. The insulation system and the potting have to survive the same thermal cycling the semiconductors are rated for. |
| SEC-25KW-SIC-PIM-GEVK | 25 kW DC fast charger on a SiC power integrated module, three phase PFC into a dual active bridge, bidirectional for 400 V and 800 V batteries, above 96 percent | Three phase PFC inductors and the dual active bridge transformer | One converter that has to serve both a 400 V and an 800 V battery is two turns ratios and two isolation problems asking to be solved by one part. |
Wolfspeed
CRD reference designs
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| CRD-22DD12N | 22 kW bidirectional DC-DC converter with 1200 V SiC, 98.5 percent peak in both directions, 8 kW per litre | CLLC transformer and resonant inductance at high power density | Eight kilowatts per litre leaves the magnetic no spare volume for a thermal margin. Every millimetre of window is already accounted for. |
| CRD-22DD12N-U2 | 22 kW bidirectional CLLC using top side cooled U2 package devices | CLLC transformer inside a top cooled mechanical stack | Top side cooling changes where the heat leaves the board, so a transformer that worked in the bottom cooled version is now in the wrong place thermally. |
| CRD-22DD12N-J2 | 22 kW bidirectional CLLC built on an insulated metal substrate board | CLLC transformer on an IMS board | An IMS board pulls heat sideways instead of upwards, so the winding termination becomes a thermal path. That is a construction decision. |
| CRD-30DD12N-K | 30 kW DC-DC converter for off board EV fast charging, above 98 percent peak | High power isolation transformer and the resonant inductance | At 30 kW the transformer is the largest single part in the converter. Nothing about it is a selection: it is designed to the cabinet it has to fit. |
| CRD-020DD17P-J | 20 W auxiliary power flyback converter on a 1700 V SiC device | Flyback transformer with reinforced isolation from the high voltage DC link | Twenty watts, and still nothing about it is standard: the isolation it has to hold off a 1700 V rail decides the bobbin, the margin tape and the winding order. |
| CRD-02AD065N | 2.2 kW bridgeless totem pole PFC | PFC boost inductor | A totem pole holds the inductor in continuous conduction across the whole line cycle, so the core loss follows the line rather than sitting at one operating point. |
| CRD-03600AD065E-L | 3.6 kW bridgeless totem pole PFC with synchronous SiC MOSFETs | PFC boost inductor at the higher current | Same converter, more current. The winding loss grows faster than the power does, which is where the strand choice stops being optional. |
| CRD-06600DD065N | 6.6 kW high frequency LLC | LLC transformer and the resonant inductor | High frequency is the stated point of the design. Raise it and the split between core loss and winding loss moves, so the litz construction has to move with it. |
| CRD-06600FF065N-K | 6.6 kW bidirectional totem pole PFC and CLLC in one high power density design | PFC inductor and the CLLC transformer with resonant elements on both sides | Bidirectional means each winding is the primary at some point in the day. The symmetry that buys you is also a constraint on how the turns can be arranged. |
| CRD-22AD12N | 22 kW bidirectional active front end | Three phase line inductors | Three inductors that have to match each other, and that see what the grid does rather than what a bench supply does. |
| CRD60DD12N-GMB | 60 kW dual active bridge isolated DC-DC on Gen 4 baseplate-less SiC modules, 99.2 percent peak and above 98.8 percent anywhere over 20 kW | Isolation transformer whose leakage inductance is the series inductance, with no separate inductor at all | This is the clearest case on the page. Wolfspeed deleted the series inductor by designing its value into the transformer leakage. That number comes from the winding geometry and from nowhere else, so the part cannot exist before the design does. |
| CRD-60DD12N-K | 60 kW three phase interleaved LLC converter | Three LLC transformers and their resonant inductors | Interleaving at 60 kW means a mismatch between the three does not average out, it concentrates in whichever phase happens to be lowest impedance. |
| CRD-60DD12N | 60 kW interleaved boost converter | Interleaved boost inductors | Interleaving only cancels the input ripple if the inductors are the same inductor. The cancellation is a specification on the set, not on each one. |
Power Integrations
RDK reference design kits and DER design examples
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| RDK-1054Q | 120 W supply with a planar transformer replacing the 12 V battery in an 800 V BEV, InnoSwitch3-AQ with a 1700 V SiC switch | Planar transformer rated for an 800 V bus | An 800 V bus in a vehicle sets the isolation before anything else is decided, and a planar stack has to deliver it in a height nobody will give you more of. |
| RDK-952Q | 86 W supply for an 800 V bus using InnoSwitch3-AQ with a 1700 V SiC switch | Flyback transformer with reinforced isolation to the 800 V rail | Reinforced isolation at 800 V doubles the creepage. The winding arrangement that met it at 400 V does not survive the move. |
| RDK-840Q | 30 W emergency or backup supply inside a traction inverter, InnoSwitch3-AQ | Flyback transformer qualified for the inverter environment | It lives inside the inverter, so it sees that enclosure temperature and that vibration. The electrical design is the easy half. |
| RDK-848 | 42 W dual port isolated flyback with 30 W USB PD 3.0 and a 12 W USB-A port, InnoSwitch3-Pro PowiGaN | Multi output flyback transformer | Two outputs that have to cross regulate means the coupling between secondaries is a specification, and coupling is decided by where the turns sit. |
| RDK-747 | 65 W charger using InnoSwitch3-EP with a 750 V PowiGaN switch | Compact flyback transformer at high frequency | At charger volumes the transformer is a cost line, not just an electrical one, and that is decided by turns count and by how fast it can be wound. |
| RDK-1066 | 10.9 W dual output isolated flyback adapter, InnoSwitch3-EP with a 1250 V PowiGaN switch | Small dual output flyback transformer | Eleven watts and a 1250 V switch means the insulation system, not the power, sets the size of the part. |
| RDK-641 | 40 W variable output supply using InnoSwitch3-Pro | Flyback transformer across a wide output range | A variable output means the transformer never sits at one operating point. It has to be right across the range, not optimal at one place in it. |
| RDK-752 | 15 W wide range input dual output converter, InnoSwitch3-EP with a 725 V silicon switch | Dual output flyback transformer for a wide input range | A wide input range moves the flux swing with the line. The core is chosen for the worst corner, which is rarely the nominal one. |
| RDK-801 | 100 W flyback three way dimmable LED ballast using LYTSwitch-6 and HiperPFS-4 | PFC inductor and the flyback transformer | Dimming takes the converter down to a fraction of full load, where a transformer sized only for full power starts behaving differently. |
| DER-715 | 24 W multi output flyback for monitors using InnoMux2-BL, one CV output and four independently regulated CC outputs | Five output flyback transformer | Five windings regulated to three percent without a downstream converter puts the whole burden on the winding geometry. That is as custom as magnetics gets. |
Navitas
GaNFast evaluation boards
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| NVE107C | 650 V bidirectional GaN evaluation board | Bidirectional inductor working in both directions | Bidirectional operation means the part is never at a single duty point. The ripple and the bias change sides, and so does where the loss lands. |
| Single Stage BDS | Single stage bidirectional switching platform built on GaNFast power ICs | Single stage magnetic carrying the whole conversion | One stage means one magnetic doing what two normally share. There is nowhere to hide a compromise. |
Microchip
dsPIC33 based demonstration platforms
| Reference | What it is | The magnetic it needs | Why it is not a catalogue part |
|---|---|---|---|
| 3.8 kW and 7.6 kW dsPIC33C totem pole | Totem pole PFC stage of an onboard charger, dsPIC33C with SiC | PFC boost inductor at two power levels off one topology | The same topology at two powers does not scale by scaling the inductor. Core loss and copper loss move in opposite directions with size. |
| 11kW PFC Demonstration Application | Three phase PFC stage at 11 kW on dsPIC33 | Boost inductors, three of them, matched to each other | Three phase PFC puts the matching requirement on three parts at once. A tolerance that is fine individually is not fine as a set. |
| dsPIC33C 4 kW DC-DC demonstration | Isolated DC-DC stage on dsPIC33C with SiC | Isolation transformer and the output inductor | The isolation requirement comes from the end application, not from the demo. That is the part of the design that always has to be rebuilt. |
| Two/Three-Wheeler EV Charger with CAN Control | 1.5 kW onboard charger for light EVs, with CAN control | PFC inductor and the isolation transformer, at a price sensitive volume | At this volume the magnetic is a cost decision as much as an electrical one, which is exactly where a stock part stops being the cheap option. |
ConverNova is an independent magnetics manufacturer and is not affiliated with, endorsed by or sponsored by any of the companies above. Every designator is the trademark of its owner and is used here only to identify the design it names. Each one links to the vendor's own page so you can check the specification at source.
Straight answers
Reference designs, the usual questions.
Can you build exactly what the reference specifies?
Usually, and sometimes we will suggest you do not. A reference magnetic is
designed to make the semiconductor look good on a bench. If your operating point is
different, copying it faithfully reproduces a compromise that was never yours.
Do you work with SiC and GaN reference designs?
Yes, up to 2 MHz switching. Those are the designs where the reference magnetic
is most often the weak point, because the winding loss at those frequencies is where the
efficiency actually goes.
How fast can you turn one round?
We aim for an engineering answer inside 24 hours, and 4 to 8 weeks from an agreed design when
the raw materials are in stock. If your evaluation window is tighter, a
fast sample may be the better route.
What is the minimum order?
One, which is usually exactly how many an evaluation needs.
Start here
Send the reference design and what you changed.
There is a file field on the specification form for the schematic and the waveforms.