Ultra-wide input DCDC

[Hawaka]

Hello,

I’m looking for the following DCDC converter:

- Vin: ~15Vdc - 410Vdc
- Vout: 12Vdc (5V would also be ok)
- Pout: 3Wmax

The unit shall withstand to be plugged to 410Vdc and inrush current shall be limited. I do not care for efficiency, it just need to work, be cheap (<30USD) and available (50pcs). The output voltage does not need to be perfect, it will mostly be used to power a fan.

I have been looking at a lot of DCDC converter and regulator (like LTC3895, EC7AW18-72S12, LTC7801) and I though I could somehow reduce the voltage to power one of these IC but still havn’t found a viable solution that doesn’t require a lot of engineering. MP9488 would have been perfect, but is nowhere near available…

Does anyone has something on-hand or know an easy way to do that?

Thanks in advance.

[SiliconWizard]

With such a high input voltage, your best bet is a flyback converter IMO.

[uer166]

Do you have an aux rail you can use for the converter power? Or is the converter supposed to start up and operate with that high range as well.. Does it need isolation? Technically you can use any buck controller with sufficiently low minimum duty cycle rating, assuming non-isolated and you taking care of the aux supply for the controller itself somehow.

[SiliconWizard]

Thinking about it, such a wide Vin will be tough to handle with a classic flyback design. It is pretty uncommon.

Can we dare ask why such a wide Vin range? Where does that possibly come from?

As uer166 evoked, you could just use any buck converter with an external MOSFET with Vds > 450V DC, but the devil is in the details. First you'd need to figure out how to supply the converter itself. (Thus the aux rail question uer166 asked.) It's a bit of a catch-22 situation.

One possibilty would be to use a high-voltage, low power linear regulator just to power the converter itself. Something like this: https://www.onsemi.com/products/power-management/linear-regulators-ldo/ncp785a
Assuming your converter itself wouldn't draw more than about 10 mA. You'd need to add a serious tank capacitor in order to provide the current spikes during the switching of the MOSFET. Now this linear regulator will require serious heatsinking too. About 440 V x 10 mA => 4.4 W of dissipation. If a low Iq is not a requirement and availability difficukt, you could of course implement this linear regulation with a high Vds MOSFET, Zener diode and a few components. Wouldn't be adequate for providing the fan supply itself (3W of output from this would mean a LOT of wasted heat), but for powering a buck converter, that should be OK.

[uer166]

MP9488 would have been perfect, but is nowhere near available…

UCC28910 is a direct competitor that does something very similar, it's more available but not sure if the voltage goes low enough.

See https://www.ti.com/lit/an/snva750/snva750.pdf?ts=1668199411660

[Marco]

I think the UCC28910 circuit should start up at 15V, but would get a bit screwy with very low loads afterwards.

[oz2cpu]

the design challenge is great..
I design switchmodes for a living, and i think it is possible, i would go for a two stage solution
with auto change, but that is maybe only me :-)

but i dont understand the user senario ?
you say it is supposed to drive a FAN
the fan is 12V and 3Watt..
your input is from 15 to 410V DC !!
that is where most people think ?? WHY ?? what is the point ??
how is this "problem" even possible ?? as a use case ??

please do explain ?
you dont know if you got 15V or 410V just like that ?

[NiHaoMike]

your input is from 15 to 410V DC !!
that is where most people think ?? WHY ?? what is the point ??
how is this "problem" even possible ?? as a use case ??

Sounds like a solar input, open circuit voltage easily extends into the hundreds of volts and the lower bound is "as low as practical".

[uer166]

Sounds like a solar input, open circuit voltage easily extends into the hundreds of volts and the lower bound is "as low as practical".

That does not sound like it; even at the absolute lowest insolation in the evening, the max power point is not at 15V for a 450V Voc panel, but maybe in the order of >50% of Voc_max.

[Hawaka]

Do you have an aux rail you can use for the converter power? Or is the converter supposed to start up and operate with that high range as well.. Does it need isolation? Technically you can use any buck controller with sufficiently low minimum duty cycle rating, assuming non-isolated and you taking care of the aux supply for the controller itself somehow.

No aux rail available, only the 15-410Vdc input is available. Isolation is not needed.

Thinking about it, such a wide Vin will be tough to handle with a classic flyback design. It is pretty uncommon.

I did think of a flyback but that's a considerable work to design, choosing the transformer and make it work for such a wide input.

Can we dare ask why such a wide Vin range? Where does that possibly come from?

Can't say too much, but the 15 to 410V comes from a capacitor bank and the circuit is a safety element that needs to work standalone, from the bank empty (or almost empty) to full charge. The fan is a part of it but by far the biggest power draw.

One possibilty would be to use a high-voltage, low power linear regulator just to power the converter itself. Something like this: https://www.onsemi.com/products/power-management/linear-regulators-ldo/ncp785a

I did find the ncp785a during my research, but couldn't find a way to make use of it. That's a good idea to use it to power a converter.

If a low Iq is not a requirement and availability difficukt, you could of course implement this linear regulation with a high Vds MOSFET, Zener diode and a few components.

Efficiency is not a requirement at all and heatsinking neither. I might look more into that idea actually, sounds cheap and easily doable.

I think the UCC28910 circuit should start up at 15V, but would get a bit screwy with very low loads afterwards.

The 15V requirement is not very strict, it could start-up a bit later (let's say up to 20-25V max). The load is constant, as the fan will always be on.

[mk_]

I think the UCC28910 circuit should start up at 15V, but would get a bit screwy with very low loads afterwards.

The 15V requirement is not very strict, it could start-up a bit later (let's say up to 20-25V max). The load is constant, as the fan will always be on.

A simple Buck from (Vin_active: 18V - 80V) down to 15V and a standard Flyback from (Vin_active: 75V - 410V) down to 15V. The Buck will be seperated with something similar like a loaddump-Protection (some zener, resistors, bjts and a PFet will do it, if powerlosses are ok a NFet is ok too).

Both supplys maybe seperated by diodes...

[NiHaoMike]

That does not sound like it; even at the absolute lowest insolation in the evening, the max power point is not at 15V for a 450V Voc panel, but maybe in the order of >50% of Voc_max.

For an aux power supply, it can be of advantage for it to power up as early as possible and allow monitoring logic to boot up. As it would only need to extract a few watts from a several kW array, the voltage that would let it happen can be very low compared to the full power voltage.

[ali_asadzadeh]

Read this app note, it's almost what you want

[Marco]

The 15V requirement is not very strict, it could start-up a bit later (let's say up to 20-25V max). The load is constant, as the fan will always be on.

AFAICS after startup the UCC28910 supply capacitor is recharged from the output voltage during the free wheeling period, with a 3 Watt load that should almost certainly happen often enough to supply it.

[Terry Bites]

webench.ti.com/power-designer

[langwadt]

https://eu.mouser.com/datasheet/2/609/8315fa-2955684.pdf  ?

https://eu.mouser.com/ProductDetail/Analog-Devices/LT8315IFEPBF?qs=u4fy%2FsgLU9M2Ku8ATK2ZiA%3D%3D

[Hawaka]

https://eu.mouser.com/datasheet/2/609/8315fa-2955684.pdf  ?

https://eu.mouser.com/ProductDetail/Analog-Devices/LT8315IFEPBF?qs=u4fy%2FsgLU9M2Ku8ATK2ZiA%3D%3D

Damn that IC looks exactly like what I need and is cheap and available. I will surely look more into it, but many thanks already!

I have spent quite some times on the Analog website and their parametrics search, but that IC never pop-up 

[Marco]

The problem with flyback is when Vi/Vo gets close to the winding ratio ... you can see it in the datasheet as current for 50V doesn't go beyond 200 mA.

[Hawaka]

The problem with flyback is when Vi/Vo gets close to the winding ratio ... you can see it in the datasheet as current for 50V doesn't go beyond 200 mA.

Yeah, and I also find that the transformer is always tricky to do or hard to find something off the shelves. Especially for very small number.

But the LT8315 has a non-isolated typical schematic in the datasheet which looks very interesting, and I think it's the first time I see a flyback controller that can be used in a non-flyback way.

[Marco]

That's actually the same topology from the UCC28910 link.

[langwadt]

The problem with flyback is when Vi/Vo gets close to the winding ratio ... you can see it in the datasheet as current for 50V doesn't go beyond 200 mA.

Yeah, and I also find that the transformer is always tricky to do or hard to find something off the shelves. Especially for very small number.

But the LT8315 has a non-isolated typical schematic in the datasheet which looks very interesting, and I think it's the first time I see a flyback controller that can be used in a non-flyback way.

yo could start here: https://www.digikey.com/en/products/detail/analog-devices-inc/DC2478A/10239926

[Smokey]

The problem with flyback is when Vi/Vo gets close to the winding ratio ... you can see it in the datasheet as current for 50V doesn't go beyond 200 mA.

Yeah, and I also find that the transformer is always tricky to do or hard to find something off the shelves. Especially for very small number.

But the LT8315 has a non-isolated typical schematic in the datasheet which looks very interesting, and I think it's the first time I see a flyback controller that can be used in a non-flyback way.

yo could start here: https://www.digikey.com/en/products/detail/analog-devices-inc/DC2478A/10239926

That's a killer power supply.  Low component count too.  I'm sure I'll end up using that at some point.  Thanks.

[Hawaka]

For any body interested, I did end-up building the solution with the LT8315. It does work really good and with very minimal temperature raise.

The only problem I encountered is when hot-plugging at 400V, the LT8315 would always explode. As I have a very low input capacitors (to avoid inrush current and arcing), the dV/dt is extremely high. I believe that the internal power FET is in a undefined state without power and would end-up in short-circuit for a bit too long.

I had put that project on the side but it would be nice to finish it so that I can post a working solution.

[Jeroen3]

Yes, inrush is a problem. Many of the example circuits of regulators like this include a fusible resistor.

I've made one with a MP173 and a discrete linear regulator in parallel, because getting those primary side regulators working in the low range is problematic because their internal power supply doesn't work yet.

[ali_asadzadeh]

You can share the working design here.