Hi,
We got a requst to design a DC Power Supply which will be sourced from solar panel array connected in series. At peak the DC volt is rated to 1200VDC, typically 800VDC.

Honestly this is a bit above my capabilities, nevertheless I took the challenge as part of learning curve. I am studying about solar array output, and it does not look like a dc power source.

I'm also need to address the method of reducing the 1200VDC to 5V, definitely need to use switching regulator. Component selection for high voltage will be a issue too.

Guys, I am willing yo burn candle and time to experiments to develop reliable solution.

Really appreciate if you guys can point me in the right direction or give me some advice or reference design.

Looking for in hearing from you.

Let me understand , you need 5V@1A from 1200VDC SOLAR PANEL, what is the main load for the solar panel. what is the system power range. Do you need isolation from solar panel return.. Can you be bit more specific..

If it were me, right off the top of my head I would try a 2-switch flyback. Use triple insulated wires on the xfmr. That is, assuming your load is small. Kadir mentioned 1A. Seems like a tiny load to me...anyways, give us some more details if you can.

Check out STMicroelectronics 1500V MOSFET family (ie part # STW9N150). These look pretty sweet. You might also try looking around for IGBTs at this voltage level.

Also, you can always step down to something higher than 5V, and then place a linear regulator or another buck stage on the secondary side to help out. Again, this would depend on the type of power level you are working with. Are we talk ing about 5W, or 5kW?

Found a 1300V/33A very fast IGBT from STMicroelectronics (STGW38IH130D) for about half the unit price for the MOSFET listed in my above post. You will find at these voltages, IGBTs are much more cost effective than MOSFETS (just be sure to get ones that are optimized for fast switching). If using the 1300V device, maybe place some polypropylene decoupling caps across the switch leg to deal with any unclamped stray inductance from the DC bus.

Also foubnd a whole bunch of 1700V IGBTs.

What is your load and what is your switching frequency?

I am not sure whether u ll find similar kind of reference design in the internet or not. But I am sure this forum will help you to get it done or at least in a comfortable zone. Pls reply to the question asked by the forum member. In addition to that, we generally find challenges in device selection, transformer insulation, clamping ckt etc for this kind of SMPS. Is this something a 5W bias power supply?

first of all thanks guys for reading and considering solution for my design problem, please bare few more hours, I'm traveling and really hard to type long text on S3, once reach home (eta 3 hours), I reply with requested inputs

@ KADIR,

This solar panels are part of a 10MW solar farm, panels are connected in series to become arrays and arrays are parallel connected to Solar Inverter. Which convert the DC to AC, typically 415V, 3-phase or 240V, 1-phase.

My objective is to tap the arrays point, where I have +ve and -ve line (from series connected panels). This point specified by panel manufacture provide 1200VDC, which is called "Maximum power point tracking (MPPT)"

More on this at http://en.wikipedia.org/wiki/Maximum_power_point_tracking

Typically solar array will varies the output voltage with current and also amount of sun light, as such the DC input to my DC-DC will varies between 500VDC - 1200VDC.

The output from this should be 5V @ 1A (or 5W max).

No need to isolated.

@ Jay,
I am okay with using two stage DC-DC or using drop the voltage to 48V then using linear to get 5V. I will check out the IGBTs your proposed. Any idea on the control side for the switching?

I am facing headache on the xfmr as the current vendor need details of the xfmr design, number of turn and wire size. It's quite hard for me, I used to having software doing this for me, like Power Integration, PI Expert Suit.

My load is microprocessor and some passive component. As for switching frequency, its not decided yet, as I indicated all this fairly new and sorting out one matter a time. I think the frequency will be high in the range of 100KHz + / -, but at 1200V, I am concern of the EMI.

Thanks for the "polypropylene decoupling caps" hint, noted it.

@Amita,

Yes, I am sure member in the forum will be a great guide for me, as for the 5W Power Supply, its another Yes, only concern is the high DC input.

Is it too difficult to get connection from, lets say at the middle of the panel or even lower point since power is very little and used to power micro only.. Then use fix ratio self driven isolated half bridge controlled IRxxxx and put series pas regulator on secondary side.. Done No complication ... Noting.

At this voltage level, probably the most rugged transformer design would be to use triple insulated wire. I think Virginia Insulated Products makes wire with 4 layers...(not 100% sure but check them out. I like their products). Also take a look at Rubadue and TEX-E. But as Kadir mentioned, is there any way to connect to a lower potential?

Do you really need to connect your 5W microcontroller to the 1200V solar panels?

If you can power it from the grid (415Vac) you will probably find a standard power supply and it is probably more reliable, as the panels will not generate power at night or when it is too cloudy.

If you can’t connect a power supply to the grid and the microcontroller is needed only when the sun is shining, isn’t it possible to install a very small solar panel (say, 15W) to power only the microcontroller? In this case a simple regulator will do the job.

Maybe these suggestions have already been considered, but it is worth exploring some alternatives before going to the trouble of connecting a small microcontroller to the HV installation.

I agree Fellipe. Seems a little overkill...5W from a 1200V DC bus. A small dedicated solar panel sounds like a good idea.

@KADIR
Its not possible to get other termination, the arrays are layout in such configuration, the proposed device will be installed in the "Combiner Box", where multiple arrays will come, each will have fuse and then connected in parallel to feed the DC/AC inverter

@Jay
Hi, thanks for the lead, I will check them out. As for lower source of DC, its not possible.

@Fellipe
As I mention above, the device will be mounted within the "Combiner Box", where its still DC, the 415V will be very far away, currently the application we pull 48VDC from the inverter room and this is networked to all "Combiner Box", we are trying to eliminate this.

As for no generation or at night, its not a issue, the device will not work and we are okay.

We did analysis the possibility putting operate panel for the device only, but in practical application this does not meet markets needs.

@Jay
My wish/hope is the same, but the challenge in to get powered from the existing source.

I have seen devices which directly connect to 800V system, mine is another 400VDC up. Agree its over do, but that is the challenge I need to over come.

As of now, I feel this can be done, I need to give good though on xmfr insulation, core design, switch speed, high voltage isolation and also generated noise.

I am compiling all information and will start exploding some chip in few weeks time... ;-)

As you said, "Which convert the DC to AC, typically 415V, 3-phase or 240V, 1-phase",
so you don't need to design DC-DC converter, you just buy a AC-DC adapter, connect to 240V, 1-phase source, that's it.

Hi,
Thanks for your input, a typical 10MW solar farm will have around 50-60 combiner box. currently its done as you propose. since combiner box has DC, I was presented with this challenge.

Objective or rather my design challenge is to *create* a DC-DC converter which will give a 5V output.

What about this:

Use a small battery to provide the initial power supply to your controller. Then use an auxiliary winding off your transformer to power the IC and charge the battery once the converter comes to life. The function of the battery would be to simply provide an initial source of power to your IC (kind of like the drop down resistor in an offline supply).

You could probably use any controller you wanted, since it would be completely isolated from the DC bus. Use gate drive transformers to power your switches to ensure total isolation.

interesting approach. I will explore, off hand sounds good. We still need sense device to detect if solar is generating enough voltage for us to start our switching, this could be resistor divider and opto-coupler.

This is certainly the design from Hell. 1200 V leaves black spots on your fingers (done that, been there). So isolation is mandatory. Triple insolated wire is the easiest method to wind the transformer. Always keep in mind the creepage and clearance requirements, otherwise it is a custom bobbin.
The topology can only be a transformer-isolated buck converter even switched several times a second. A ripple regulator might be the best (a voltage hysteretic controller). the uA78S40 or the MC34066 would be a good reference.
Driving a floating gate is always a problem with power loss. If cost is of little issue, please check out SiLabs which has a capacitive isolated gate driver, which I had defined in 2002 (5000VAC isolation). It requires a MOSFET (IGBT) local voltage to drive the device. This can be done through stacked resistors or stacked depletion-mode Mosfets (constant current diodes) (refer to ONSEMI) (600V each with Mohm balancing resistors) which I defined in 1997. The constant current diode provides constant current for any voltage above 30 V.
PFM would be the best control method. This minimizes the 1200V switching loss and maintaines the output voltage. Can be constant ON-time switching with output ripple regulation.
As I said, the application from Hell.

Hi, Marty,
Thanks for your details hints... really shows you know a lot. Yup agreed thid challenge is really from Hell, the more I time I spend the more complicated it looks....

In two week time I should end my research and start prototyping....

This kind of smps is really a special type design. We must enter inside the layout to properly make the PCB layout. We need to be done a lot with defining creepage and clearnace among HV line (+ve and -ve), and other digital level signals. Otherwise it will fire if you ignore HV and LV lines. And Sir Marty said, there minimum 5KV isolation required between primary and secondary and here isolated SMPS is must requirement. when I started a similar development I was also like you. Rather than development I was seeking for getting confidence over high voltage testing. Finally with the help of all great Engineers done a lot with HV testing.....Its really a fun and after doing it you will feel very confident. Take it as a challenge and make it happen. You will enjoy the development, though it may take some time to development.....!!!

Maybe I got walked on. I know ya'll  can design what I need. The concept alone should earn a patent. My idea is very marketable. I'll spell it out.

 "Wireless" [ha ha] camera. Camera sends pics to base via antenna. Cool! Not wireless; 120 vac must be supplied to camera convertor to supply 5vdc @ 1 amp to run camera. This is not convenient to camera placement. I don't need to string wires through the trees ; so my camera can send me pics by antenna. Are you catching on now?

 Components required:   Rechargeable battery that can support 5vdc @ 1amp for 14 or more hours. { For running the camera at night.}
      
     A solar panel that can fully charge the battery for the next night; and also fully power the camera during the day.

  A control card that knows when to store and send. It will maintain the battery and  keep the camera online 24/7. No wires from another power source. Camera will be totally wireless and very mobile. Can be moved from tree to tree. 

 Anyone getting this concept?