diy solar

diy solar

DIY Inverter

Good points. I think you are looking at paralleling H bridges with one mosfet per bridge. ZVS? Yes, you could squeak a little more efficiency out by paralleling many smaller bridges. How are more parts going to improve reliability tho? You must be considering redundancy. No? Would each bridge have it's own controller with a common current command or one controller with each bridge using the same gate signals? Your transformer would be plenty easy to wind.
 
I think you are looking at paralleling H bridges with one mosfet per bridge.

As said, I parallel circuits, each composed of a bridge and a transformer. That way you avoid paralleling MOSFETs and so you avoid the inconvenients of doing that.



ZVS isn't really related to what I was saying.


How are more parts going to improve reliability tho? You must be considering redundancy. No?

Not really, redundancy is a nice side effect but the main goal is to avoid unequal currents between multiple MOSFETs in // due to layout, inequal Vgsth, inequal switching timings, etc...

It also increase the surface area to power ratio so any heat generated is far easier to dissipate (lower temp = long lifetime/higher reliability).


Would each bridge have it's own controller with a common current command or one controller with each bridge using the same gate signals?

Ideally they would be staggered to limit current spikes, noise and ripple so one controller but multiple drivers. Else it would be only one controller with one driver for the simplicity (unless multiples drivers makes things simpler/cheaper of course).


Your transformer would be plenty easy to wind.

That's one of the advantages: smaller and easier to wind transformers instead of one big monster (also easier to manage mechanical problems like avoiding vibrations related problems for example).
 
As said, I parallel circuits, each composed of a bridge and a transformer. That way you avoid paralleling MOSFETs and so you avoid the inconvenients of doing that.




ZVS isn't really related to what I was saying.




Not really, redundancy is a nice side effect but the main goal is to avoid unequal currents between multiple MOSFETs in // due to layout, inequal Vgsth, inequal switching timings, etc...

It also increase the surface area to power ratio so any heat generated is far easier to dissipate (lower temp = long lifetime/higher reliability).




Ideally they would be staggered to limit current spikes, noise and ripple so one controller but multiple drivers. Else it would be only one controller with one driver for the simplicity (unless multiples drivers makes things simpler/cheaper of course).




That's one of the advantages: smaller and easier to wind transformers instead of one big monster (also easier to manage mechanical problems like avoiding vibrations related problems for example).
Thank you.
 
Classic toroidal core, not too high in frequency to limit switching losses (something around 100-200 kHz seems a good compromise), and Litz wire to limit copper losses ;)

Also, massively // topology with small transformers and only one MOSFET bridge per transformer (like having 20x 100 W inverters instead of 1 or 2 big one). That should allow for better efficiency, higher reliability, and easier design (current sharing becomes less critical).
I suggested parallel-series transformer confiquration bit earlier: primaries in parallel, secondaries in series. You get perfect current sharing between transformers and mosfets. And you can possibly use 1:1 transformers that simplifies the winding and sourcing parts.
 
As said, I parallel circuits, each composed of a bridge and a transformer. That way you avoid paralleling MOSFETs and so you avoid the inconvenients of doing that.




ZVS isn't really related to what I was saying.




Not really, redundancy is a nice side effect but the main goal is to avoid unequal currents between multiple MOSFETs in // due to layout, inequal Vgsth, inequal switching timings, etc...

It also increase the surface area to power ratio so any heat generated is far easier to dissipate (lower temp = long lifetime/higher reliability).




Ideally they would be staggered to limit current spikes, noise and ripple so one controller but multiple drivers. Else it would be only one controller with one driver for the simplicity (unless multiples drivers makes things simpler/cheaper of course).




That's one of the advantages: smaller and easier to wind transformers instead of one big monster (also easier to manage mechanical problems like avoiding vibrations related problems for example).
Perhaps you could select an off the shelf transformer and design the inverter circuitry around it's parameters. I would rather not be winding transformers by hand.
 
Perhaps you could select an off the shelf transformer and design the inverter circuitry around it's parameters. I would rather not be winding transformers by hand.
Off the self selection of suitable transformers is awfully thin. Aliexpress seem to have custom magnetics available nowadays but you can make it your lifelong task to try to beat the chinese transformer supplier to submission and supply what you asked. :LOL:
 
Perhaps you could select an off the shelf transformer and design the inverter circuitry around it's parameters. I would rather not be winding transformers by hand.

It would be a very low number of turns so not hard to do by hand ;)
 
I suggested parallel-series transformer confiquration bit earlier: primaries in parallel, secondaries in series. You get perfect current sharing between transformers and mosfets. And you casibly use 1:1 trancoax would sformers that simplifies the winding and sourcing parts.
I think this is a good approach too. I like coax. We did a bunch of coax transformer work at UW-Madison and they worked well. Litz wire would be a good choice too. So maybe 4 converters with 1:1 transformers. As BiduleOhm pointed out these would spread the heat around too. I was thinking open loop Push-Pull converters. I've many ZVS bridges and and looking for something new.
 
Yes. this is one of the construction methods. Sheets of copper are often used. They can be cut like paper dolls and folded back on each other to get more turns. Copper could also be used as bus bars plugged into the pc board. Instead of using 16oz copper to do a power board - use bus bars to get the current where you need it.
I just did a thumbnail calculation for a planar transformer. The skin depth of copper at 350kHz is about 110um. A 0.01" copper foil would be a good choice and can be cut with a pair of scissors. This may actually be an easy way to construct a planar transformer. Use kapton tape in between layers If a voltage doubler topology could be used then a 1:2 transformer would work great. Or the outputs of a pair of these (voltage doubler) converters using a 1:1 transformer could be put in series to get a good invertable bus. Your idea is gaining some traction on a small hilltop here in New Hampshire.
 
Classic toroidal core, not too high in frequency to limit switching losses (something around 100-200 kHz seems a good compromise), and Litz wire to limit copper losses ;)

Also, massively // topology with small transformers and only one MOSFET bridge per transformer (like having 20x 100 W inverters instead of 1 or 2 big one). That should allow for better efficiency, higher reliability, and easier design (current sharing becomes less critical).
Still looking at the toroid and the planar. Your Toroid looks to be the easiest way to go. I'll call Rubadue Sue and see if she can hook me up with some Litz. Thanks.
 
Sorry, didn't mean to brush you off. Yes, it would be a fairly easy task to cobble one together. At lower power levels the choices to get from 24v to an invertable voltage level for 120vac are plenty. A simple forward converter would be easy. The inverter topology stays the same regardless of power levels. 4 transistors wired in a H configuration is the normal. I prefer the high frequency design, but at 100w-200w the low frequency design could be the way to go. A 60Hz transformer to get to 120v are all over the place. Buck-boost transformers would play well and you could get 120-0-120 very easy with one. https://www.ebay.com/itm/Jefferson-...595160?hash=item3ff40f7c18:g:1xIAAOSwZ3Ff7Ke4
 
Sorry, didn't mean to brush you off. Yes, it would be a fairly easy task to cobble one together. At lower power levels the choices to get from 24v to an invertable voltage level for 120vac are plenty. A simple forward converter would be easy. The inverter topology stays the same regardless of power levels. 4 transistors wired in a H configuration is the normal. I prefer the high frequency design, but at 100w-200w the low frequency design could be the way to go. A 60Hz transformer to get to 120v are all over the place. Buck-boost transformers would play well and you could get 120-0-120 very easy with one. https://www.ebay.com/itm/Jefferson-...595160?hash=item3ff40f7c18:g:1xIAAOSwZ3Ff7Ke4
Thank you. I just wanted something small and simple to cary on my mobility scooter to power up a laptop or something simple.
Thanks for your reply. :)
 
Open Source Inverter Specification



  • Inverter is 48vdc+- to 120vac. 5kW with 100% overload.
  • Overload capability will be a function similar to a motor thermal overload. Tjmax <150c in overload. PCB tracks <105c
  • A high frequency design will be the goal. This would require a DC-DC converter(s) front end in addition to the basic inverter. Initial design would tend towards a basic low frequency inverter that can drive a 50Hz-60Hz transformer
  • Low frequency design is flexible voltage designed to drive many LF transformer ratios. 230v is doable.
  • High frequency design would most
  • Thermal simulations performed with a crossflow blower seem the likely way to go. A pair of extrusions could be interleaved with converter on one side and inverter on opposite side. One blower would be required to cool both converters.
  • Switching frequency at or above 10kHz with possible reduction in carrier frequency in overload condition
  • Output Inverter transistors would be TO-247 style. Silicon Mosfets look to be the best choice with 2 on each corner.
  • Input DC-DC converter most likely resonant switched open loop Push-Pull design. Two parallel input series output converters with 1:1 transformer may prove best design. Voltage doubler design would provide 1:4 voltage boost.
  • DC-DC converters may use toroidal and/or planar transformer sets.
  • Topology would be plug-in control boards with 4layer 6oz copper power boards
  • Units will be parallel capable with 240v or 120-0-120 volt split phase operation.
  • Output frequency selectable 50Hz/60Hz
  • On board EMI filters at both input and output.
  • Digital display with operation parameters and fault annunciation.
  • Auto-restart capable in event of fault.
  • Integrated fuse TBD
  • Cost target is always zero. We see how close we can get.
  • Size target is 250mm x 400mm x 150mm for high frequency design
  • No built in chargers of any kind. MPPT chargers are a dime a dozen and a PFC design would complicate and slow effort.
  • PCB designs in Altium. Other packages as well.
  • Thermal modeling is Sauna
  • Spice modeling is LT Spice.
  • Tenor autem omnino documentum Inverter will be provided.
 
I suggested parallel-series transformer confiquration bit earlier: primaries in parallel, secondaries in series. You get perfect current sharing between transformers and mosfets. And you can possibly use 1:1 transformers that simplifies the winding and sourcing parts.
I think you are onto something. I am looking at using two push-pull converters with secondaries in series. If i use a voltage-doubler output on each converter I am at 1:1 ratio. I absolutely have to keep leakage low as I intend to go for small snubbers and I believe planar may be better choice there. Using 10mil copper puts me in the skin effect sweet spot and can cut it with a pair of scissors. interwinding capacitance will be high, but I am willing to live with that (I think) The toroid was suggested with Litz. I would need a gapped toroid to account for any flux imbalances tho still on the table. Looking at some SiC switches now. I expect to have a preliminary push-pull design in LT Spice this weekend. I'll throw it up on my webpage.
 
Hi there. Is there a reason why ther seem to be no 3.5V inverters? Other than lack of demand. I'd really like to make one but the kits only seem to go down to 12V. I have basic electronics knowledge but nothing like circuit design.
 
Yep, there's is: for example a 3.5 kW inverter powered by only 3.5 V would need 1 kA which means very very large wires, fuses, connections, etc... to the point it's impractical (or at least very expensive).

So unless you only need a few hundreds W then it's not a good idea.
 
Yep, there's is: for example a 3.5 kW inverter powered by only 3.5 V would need 1 kA which means very very large wires, fuses, connections, etc... to the point it's impractical (or at least very expensive).

So unless you only need a few hundreds W then it's not a good idea.
yeah what about for 120A?
The PCB etc on a 1500W 12V inverter would be rated for that current.
 
Back
Top