Transformer with high isolation in DC-DC Converter

[PartialDischarge]

Need guidance with a transformer for a low voltage dc-dc converter, the objective is to obtain around 5-10W at the secondary but with a lot of 50Hz isolation between pri-sec, I'm talking about 10kV or more.

I'm not a magnetics/transformer expert but I'm committed to learn about this, from my understanding could a flyback transformer be designed to provide higher isolation compared to a transformer given the same wire? My reasoning comes from the fact that since a flyback transformer is really two coupled inductors, the primary and secondary can be winded far apart from each other and not overlapping like in a transformer??? I'm I right or not?

If not what would be the best approach?

(Wouldn't want to discuss here about the need to use special insulated wires, epoxies or oil, just the type of transformer)

[jbb]

I think a standard flyback could be bad news here. In order to get the isolation you need, the primary and secondary will be quite far apart, which increases lekagae inductance. Flyback supplies don’t like leakage inductance.

Not sure what the ideal topology would be - probably something half bridge or full bridge.

There’s also the question of output voltage control. How much tolerance do you have? Could be best to run the isolated DC / DC open loop ( or controlled) and the fit a buck regulator on he secondary side.

[PartialDischarge]

There’s also the question of output voltage control. How much tolerance do you have? Could be best to run the isolated DC / DC open loop ( or controlled) and the fit a buck regulator on he secondary side.

Thats an option although I can manage a specialty opto isolator that matches the isolation of the transformer

[amspire]

I'm not a magnetics/transformer expert but I'm committed to learn about this, from my understanding could a flyback transformer be designed to provide higher isolation compared to a transformer given the same wire? My reasoning comes from the fact that since a flyback transformer is really two coupled inductors, the primary and secondary can be winded far apart from each other and not overlapping like in a transformer??? I'm I right or not?

The further apart the two windings are, the more the leakage inductance of each winding. For a flyback circuit, this usually means you have to absorb much more energy in snubbers on the primary and the efficiency goes down.

But you are right, it still works.

You could use either a bobbin with two sections or a single section bobbin with sufficient layers of mylar transformer tape to safely isolate two windings from each other and the magnetic core (which is usually earthed). Getting it right at 10kV is not necessarily that easy. You also have to consider breakdown between wire layers in each single winding. For HV flyback transformers, you often have to put a layer of mylar tape between each winding layer.

[PartialDischarge]

If you decide to go the hard way, don't use flyback. Consider push pull, half bridge or LLC.

Arent those toopolgies overkill for 10W output?, although I guess is that or a very low efficiency flyback due to leakage inductance as others have posted...

[Rog520]

The flyback is much more cost-effective than bridge or push-pull, etc, and driving it is also simpler matter. 10kV isolation between windings could probably be achieved without causing excessive leakage inductance. Although windings can theoretically be physically isolated quite far from each other on a common core (separate legs, for example), you will end up with large leakage inductance and a miserably inefficient transformer...so plan on layering them together. Use some triple insulated polyimide for starters, and then probably several layers of mylar tape between windings. It's probably not necessary to insulate between each layer within a winding as there will be very little potential difference between layers (as you mentioned that the overall design was low voltage). Allow adequate margin at the edge of all windings using margin tape. Potting the finished transformer in a suitable compound would also help if you can set up the equipment to do it properly (i.e. vacuum chamber). There are other various winding techniques (such as a split primary) which can reduce leakage inductance significantly. Always a good idea to figure a snubber into your design; there are techniques which are non-dissipative if you find that a standard RCD snubber is biting into your efficiency too much.

[jbb]

There are plenty of LLC drivers out there now. Just add 2 MOSFETs and some gate resistors. It takes a bit more design effort than a flyback but I really suggest you give it a go.

A properly designed LLC will actually use the leakage inductance in operation to make the converter soft switch. A flyback supply will instead need snubbers to burn off the energy storage in the leakage inductance.

With a bit of luck, you might find that you can get a stable-enough output voltage using input voltage feedforward rather than optocoupler feedback. Having fewer parts across the isolation gap might be worth extra effort.

[BrianHG]

Approved 10kv isolated 6 watt DC-DC converters already exist:

https://www.digikey.com/products/en/power-supplies-board-mount/dc-dc-converters/922?k=dc-dc+converter&k=&pkeyword=dc-dc+converter&pv2187=u6W&FV=2dc19ee%2C2dc1bff%2C2dc1c58%2C2dc2032%2C2dc236c%2Cmu10kV%7C2226%2Cffe0039a&quantity=0&ColumnSort=1000011&page=1&stock=1&nstock=1&pageSize=100

The question is, do you go out and make a circuit and custom transformer and hermetically seal/pot it and hope for 10kv isolation with trouble free home made driver operation at different power levels and temperatures, or, do you just spend the 42$ USD and say the hell with it, I know I have full medical grade 10kv isolation guaranteed, with all the driving electronics hermetically sealed, with over current and thermal protection, completely specked & tested good and approved.

One minor issue, this is 10kv dc isolation, 5kv ac.

LOL, I wonder if placing 2 in series will give you 20kv dc isolation.

[PartialDischarge]

Approved 10kv isolated 6 watt DC-DC converters already exist:

As you mentioned it is only 5kV for the 1min AC test, which is what counts. However the overall idea is more complex and we may want to design our own solution. This people have high isolation power supplies:

http://www.siebel-elektronik.de/en/products/power-supplies/

[PartialDischarge]

A properly designed LLC will actually use the leakage inductance in operation to make the converter soft switch. A flyback supply will instead need snubbers to burn off the energy storage in the leakage inductance.

Thats a good point on the LCC, however I can also come up with a clever snubber in the flyback configuration that will turn back most of the energy into the primary rail again, and I'm not referring to typical lossless snubbers.

[BrianHG]

I guess a simple extra thick split-bobin switching transformer wouldn't be good enough in your application.

[Nauris]

Need guidance with a transformer for a low voltage dc-dc converter, the objective is to obtain around 5-10W at the secondary but with a lot of 50Hz isolation between pri-sec, I'm talking about 10kV or more.

I think you need the rotating type of transformer for example two brushless motors connected with insulating shaft. It can provide very high isolation easily.
Motors are cheap these days and if it has quality bearings runs for more than ten years continuous at modest speed no problem.

If you are going to oil fill it you can use simple plain (journal) bearings for practically infinite life.

[Kleinstein]

There is not big problem getting the required insulation if there is enough spaces size. So it gets difficult if you need it small - but no problem if it is allowed to weight a few pounds.  I have used an E-beam evaporator that used just grid connected 60 Hz iron core transformers with 20 kV (or similar) isolation rating. AFAIR they were only about 100 VA, but something like 20 pounds (10 kg) and big.  Like a really large toroidal core with both windings encapsulated in thick plastics, well separated and likely  >10 mm or insulation between the windings and core. The high current (e.g. 20 mm²) and high insulation cables were kind of odd - normal cables with extra tubing around.

Especially if starting from a low voltage, a push pull or similar driver is not at all complicated. Regulation could be just on the secondary side. For poor coupling Royer type converters work reasonably well.

At lower power there are acoustic transformers, that also allow for very good insulation, as the power transmitting part is something like a rod of glass.

[PartialDischarge]

Never heard of rotating or acoustic transformers, TI has an interesting app note about the latter:

http://www.ti.com/lit/an/snoa819/snoa819.pdf

[nctnico]

Need guidance with a transformer for a low voltage dc-dc converter, the objective is to obtain around 5-10W at the secondary but with a lot of 50Hz isolation between pri-sec, I'm talking about 10kV or more.

A wireless energy transfer setup is a good solution. A project I worked on required somewhere around 35kV between the PSU and electronics box (used in a high voltage testing lab). I designed a wireless energy transfer system for that. Getting a few cm of distance between the transmitter and receiver isn't a problem.

[PartialDischarge]

Need guidance with a transformer for a low voltage dc-dc converter, the objective is to obtain around 5-10W at the secondary but with a lot of 50Hz isolation between pri-sec, I'm talking about 10kV or more.

A wireless energy transfer setup is a good solution. A project I worked on required somewhere around 35kV between the PSU and electronics box (used in a high voltage testing lab). I designed a wireless energy transfer system for that. Getting a few cm of distance between the transmitter and receiver isn't a problem.

What frequency did u use? Could you achieve at least 2 or 3W ??

[nctnico]

Need guidance with a transformer for a low voltage dc-dc converter, the objective is to obtain around 5-10W at the secondary but with a lot of 50Hz isolation between pri-sec, I'm talking about 10kV or more.

A wireless energy transfer setup is a good solution. A project I worked on required somewhere around 35kV between the PSU and electronics box (used in a high voltage testing lab). I designed a wireless energy transfer system for that. Getting a few cm of distance between the transmitter and receiver isn't a problem.

What frequency did u use? Could you achieve at least 2 or 3W ??

I used around 190kHz and could reach a maximum power around 30W to 40W at 35mm distance (IIRC) and around 70% efficiency but the power limit was not in the off-the-shelve coils but in the transmitter circuit I designed.

[David Hess]

Isolation just depends on the transformer and primary side regulation to control the output voltage can be done with a flyback or inverter design.  Offhand I do not know of any stock 10+ kilovolt isolation transformers so I would plan on winding one.  I have seen some which used a toroid or even an elongated toroid with a winding at either end.  Each winding was well insulated from the core with tape or a bobbin and I would definitely vacuum pot the whole thing at least conformally.  Leakage inductance is higher with widely separated windings but that just means that the design needs to take that into account.

A flyback transformer needs to have a high saturation current necessitating an air gap which would normally be distributed but an inverter transformer is more like a pulse transformer with low saturation current.

An inverter design would be easier than a flyback design.  If primary side feedback is used then the voltage drop of the rectifiers on the secondary is compensated for using diodes in the feedback circuit on the primary but I doubt you need that kind of accuracy unless you want to do away with a secondary side voltage regulator.

How were you planning on testing them for sustained isolation voltage?  I guess an old CRT oscilloscope anode supply could be used for testing; many of them are in the 12 to 18 kilovolt range.

[PartialDischarge]

Isolation just depends on the transformer and primary side regulation to control the output voltage can be done with a flyback or inverter design.  Offhand I do not know of any stock 10+ kilovolt isolation transformers so I would plan on winding one.  I have seen some which used a toroid or even an elongated toroid with a winding at either end.  Each winding was well insulated from the core with tape or a bobbin and I would definitely vacuum pot the whole thing at least conformally.  Leakage inductance is higher with widely separated windings but that just means that the design needs to take that into account.

Exactly my thoughts, since the primary rail will be around 24V the switching mosfet can easlily be a 800V o 1000V part (for other stuff I use 2.5kV switched mosfets...). I'll probably design an active snubber that can turn most of the energy in the spikes back to 24V again.

A flyback transformer needs to have a high saturation current necessitating an air gap which would normally be distributed but an inverter transformer is more like a pulse transformer with low saturation current.

I was thinking flyback with powder core, the materials sintered in these "may" help with the isolation too

How were you planning on testing them for sustained isolation voltage?  I guess an old CRT oscilloscope anode supply could be used for testing; many of them are in the 12 to 18 kilovolt range.

Power frequency isolation tests must be carried at 50Hz, and although this is a dc-dc it is part of a bigger system that will be connected to a MV power frequency line.
At home right now I can generate up to 160kV DC floating, but in AC I'm limited to around 18kVrms using a casted 70lbs phase-phase isolated transformer.  Looking forward to buying a 100kV phase-neutral one.

[Kleinstein]

The usual power core does not provide much insulation, but the low µ could be a problem with the rather large diameter core.

To get a good insulation it takes some separation and thus a larger than normal core, with sufficient space for the windings and extra insulation.

Especially with only 24 V input voltage there is no problem using a bridge or push pull driver, despite of the low power. As feedback is not that easy due to the high insulation voltage, one could consider using a forward converter and separate regulation of the secondary side. Of cause it depends on the regulation requirements.

[PartialDischarge]

Especially with only 24 V input voltage there is no problem using a bridge or push pull driver, despite of the low power.

How does a bridge or p-p configuration help in increased isolation vs a flyback?? Because the problems associated by using a flyback are secondary to me (emi, secondary regulation, increased leakage indctance...) and I'll probably can manage them

[T3sl4co1l]

I was thinking flyback with powder core, the materials sintered in these "may" help with the isolation too

FYI, powder is conductive, if not at low voltages then at modest voltages (breakdown).  Ferrite too.  If coated, the breakdown is only a few thousand, typically tested at 2500VAC or the like.

If you don't need agency approval, common mode chokes can hi-pot 10kV or more, and are okay as forward mode transformers.  Regulation is poor (as is LLC, until you add a postreg or feedback).

Tim

[nctnico]

I was thinking flyback with powder core, the materials sintered in these "may" help with the isolation too

FYI, powder is conductive, if not at low voltages then at modest voltages (breakdown).  Ferrite too.  If coated, the breakdown is only a few thousand, typically tested at 2500VAC or the like.

Precisely. Which is why wireless is such a good and low cost (using off-the-shelve components) alternative. You can put a thick piece of plastic between the two parts you want to isolate.

[PartialDischarge]

Precisely. Which is why wireless is such a good and low cost (using off-the-shelve components) alternative. You can put a thick piece of plastic between the two parts you want to isolate.

That's a neat solution, I still have to gather more technical info on topologies, characteristics and implementation

[David Hess]

To get a good insulation it takes some separation and thus a larger than normal core, with sufficient space for the windings and extra insulation.

Somewhere I have an entire power line frequency high isolation power supply module where the transformer has its core and windings suspended with like an inch air gap between them.  The specifications even include the isolation capacitance in picofarads between the primary and secondary.

Power frequency isolation tests must be carried at 50Hz, and although this is a dc-dc it is part of a bigger system that will be connected to a MV power frequency line.
At home right now I can generate up to 160kV DC floating, but in AC I'm limited to around 18kVrms using a casted 70lbs phase-phase isolated transformer.  Looking forward to buying a 100kV phase-neutral one.

It sounds a little like something I almost worked one once which involved attaching a load cell to a high tension power line.  My idea was to use optical power transfer through a fiber optic line with a beam splitters so that an optical modulator at the load cell end could send the data back.

[PartialDischarge]

It sounds a little like something I almost worked one once which involved attaching a load cell to a high tension power line.  My idea was to use optical power transfer through a fiber optic line with a beam splitters so that an optical modulator at the load cell end could send the data back.

The optimal solution to that would be to have a small ac/dc non-isolated converter from MV to 5V for example, that powers the cell amplifier, ADC, and a low power radio link transmitter (LoRa maybe).

That’s what I have in prototype right now, a non-isolated converter from MV (12kv.. 20kV) to 5-12V that can put 300mA to 1A (still in testing). Of course it has to withstand impulse lightings as if it were an inductive transformer and other fun stuff in the standards: for phase to neutral MV transformers for example IEC 60044 states that a voltage increase of 90% must be tested during 8h (in case of a ground fault in isolated neutral systems). IEEE C57.13 however states otherwise. Anyway I still have to do a lot of testing in this thing.

Now I’m studiying the possibility of making the same but isolated for low power appplications.

[nctnico]

What kind of bitrate do you need? One interesting aspect of wireless energy transfer is that it is very similar to RFID and hence it is not difficult to implement communication.