1kW 200/400V dc/dc converter design tip

[Benta]

Which inductor have you chosen? This is a key question.

[Amper]

Even though his is a different kind of inverter and the power is multiple times larger i like to see some mildly similar things for comparison:

[coppercone2]

hmm, I am just curious. Would tantalum caps work, or maybe wet tantalums, to get size down while increasing reliability close to film caps? (if they are properly over rated). ?

like hermatic ones.

[David Hess]

hmm, I am just curious. Would tantalum caps work, or maybe wet tantalums, to get size down while increasing reliability close to film caps? (if they are properly over rated). ?

like hermatic ones.

Tantalum parts could be used but would be awfully expensive and they are only available up to about 50 to 100 volts for hermetic packages.  You might find them in a military, aerospace, or extreme industrial application.  Polymer electrolytics would be better for lower ESR and higher ripple current rating but they are only available in even lower voltages.  That leaves aluminum electrolytics, film, and ceramic.

[jbb]

Well, a SiC MOSFET + Schottky diode is almost certainly the right tool for this job. A Gallium Nitride (GaN) device might be a bit better. PCB layout will be absolutely critical to get the best performance out of your devices.

You could reduce the diode conduction losses by using a second MOSFET as a synchronous rectifier.

You mention using a ‘transistor’ instead. MOSFETs, IGBTs and BJTs are all types of transistor.

In terms of changing the switching frequency, you’ll have to look at the knock-on effects on inductor current ripple and therefore turn on and turn off currents. It would also change the point at which you go to discontinuous conduction mode.  If you want to go full nerd on it (and why not ) with an optimiser I suggest you design for peak efficiency at 50% load but still check temperatures are OK at 100% load.

For improved efficiency at light load you could consider using a burst mode.

[Amper]

You mention using a ‘transistor’ instead. MOSFETs, IGBTs and BJTs are all types of transistor.

In case you meant me, i said bipolar transistor and thats referring to ordinary old school bjt, not fet, no igbt, not even darlington.

The point i wanted to make is that for this task you really dont need insane frequencies, power densities  and ultra modern parts at all. They used low tech even for the time and the  iron core inductor was only a 50mm toroid including winding.

[webgiorgio]

I haven't found an inductor, I have to look in the bin of old large salvaged inductors.
Online I keep getting common mode Toroidal chokes (with two windings in opposite direction) like this one.
https://www.elfa.se/en/toroidal-choke-mh-10-wuerth-elektronik-744834101/p/11064067
https://uk.rs-online.com/web/p/leaded-inductors/8711385/
I guess that they are suited to be used with near zero flux, so I can't simply connect the two windings in series.

I hanven't look up a transistor, which was suggested as more efficient than a mosfet. Can it be?

When I say "transistor" I mean BJT.

@Amper Very nice video! The small one gives an idea on how my converter will look like.

For improved efficiency at light load you could consider using a burst mode.

What is a simple way to do this?
A second 555 timer with duty cycle dependent on converter current, to enable/disable the PWM of the first 100 kHz 555? (too rough?)
Is a IC available to do this?

[richard.cs]

Another topology to consider; a half bridge driving a 1:1 isolation transformer, the output from the transformer is full wave rectified and placed in series with the input supply.
Key features:
Converter only handles half the output power so 500 W
Square wave output so minimal output capacitance is needed
Two switches and two capacitors on the input
Four diodes on the output, you could reduce this to two by going centre tapped but would probably push the copper losses up

Thoughts?

[fourtytwo42]

I think there is a safety aspect here that does not seem to have been considered in this thread. The GTI you name IS NOT ISOLATED meaning the PV side is at GRID POTENTIAL moreover the GTI does not have arc sensing meaning there can be a fire hazard on the PV side. Both these issues make it unsuitable for DIY electronics additions to the PV side.

As an alternative had you thought of extending the existing string, are the new panels on the same roof face, if so you can add them as long as N*VOC <=700V.

Another alternative would be to use a separate GTI for the new panels as GTI's are quite happy to operate in tandem (on the GRID side) or replace your existing GTI with something more suited to your combination of panels.

[coppercone2]

Are there any available tantalum caps on the market that would make that gigantic lump of foil capacitors go away?

Why are hermatic ones only built to 100v?.

[Benta]

@webgiorgio:
The coils you are linking to are common-mode noise suppression inductors and will in no way store the energy you are looking for. The current rating is for the 50 Hz going through.
The kind of coil you are looking for will be around the size of two cigarette packs. Also core and copper loss will be significant and way higher than your estimations.
That's why I suggested full bridge or forward.

[webgiorgio]

Another topology to consider; a half bridge driving a 1:1 isolation transformer, the output from the transformer is full wave rectified and placed in series with the input supply.

Neat!
If it is a single secondary with full bridge, the current can be is positive and negative in the winding.
If it is center taped (is actually twice longer), each half winding will conduct for half period.
So, I think the losses in the copper are the same, so, better with 2 diodes. Same amount of copper in both cases, if it is center taped the section is half and the length twice.
This would also reduce the losses, a it is dealing with half the power. In this case I need to make it Vo=Vin.


The GTi (grid Tied inverter) complies with the grid requirements, without isolation. It has a isolation fault monitoring, that would disconnect the AC in case of leak current to ground.
I can't add inverters neither change number and orientation of panels, ans I need to comply with technical regulations.
I am allowed only to add "optimizers" (dc-dc) on the DC side.

[Benta]

Neat!
If it is a single secondary with full bridge, the current can be is positive and negative in the winding.
If it is center taped (is actually twice longer), each half winding will conduct for half period.
So, I think the losses in the copper are the same, so, better with 2 diodes. Same amount of copper in both cases, if it is center taped the section is half and the length twice.
This would also reduce the losses, a it is dealing with half the power. In this case I need to make it Vo=Vin.

No. The center tapped configuration has significantly higher copper losses, as only half of the secondary is used at a time and it only has half the winding space = double resistance.
At these voltages I see no reason to avoid a bridge rectifier, this is only relevant in low voltage applications where diode drop is an issue.

[David Hess]

Are there any available tantalum caps on the market that would make that gigantic lump of foil capacitors go away?

I do not think so but there might be such a thing for aerospace and military applications.  They also make really big ceramic capacitors for aerospace and military applications but they are similarly expensive.

Why are hermatic ones only built to 100v?.

The solid hermetic parts are available up to like 75 volts and the wet tantalums are available up to like 125 volts.  Considering their cost, they make even expensive film capacitors the cheap option and the way to go.