[SOLVED] Drop 450VDC from solar panels to 375VDC or less ?

[hughmanoid]

I have a 7 KW grid tied solar array that is useless when the power grid goes down.

I would like to use some off the shelf VICOR DC-DC converters that are rated to put out 400 watts when fed 200 to 400 VDC.
They put out 24 VDC and can be trimmed down to 13.6 (car battery voltage). or I can use a Newmar voltage reducer (20 to 50 vdc in , 13.8 out @ 35amps max.
From there I can use an inverter or two to get 120VAC to get emergency power in daylight.
I would hope to pull an amp or 2 from the 15 amps max the panels put out and get a kilowatt or so.
Maybe enough to run a refrigerator or charge an electric wheelchair and recharge a laptop etc...
Under load, the panels drop to maybe 385 volts but depends on season, clouds, temperature,,,.

Rewiring the panels from 4 strings of 10 (450 VCD) to 5 strings of 8 (360 VDC) is the logical solution but would probably cost a lot. Changing to a battery backup solar system would cost much more.
Another inverter, batteries and having an electrician rewire the electrical service to code ? Ouch.
--------------------------------------------------------------------------------------------
So I can burn off power a few ways but was hoping to find a SMPS that drives an external FET.
---------------------------------------------------------------------------------------------
BAD Idea 1: use incandescent light bulbs to drop 100 volts.
Oops: inrush current and low initial resistance of bulb overvoltages converters. No load on converters means less current in light bulbs and too much voltage.

Idea 2: Use three 24 volt 50 watt zeners in series to run at 1/2 power (1 amp per series) and parallel more series strings to get more current. Aim a fan at the the big heat sink and done.
Oops: when the solar voltage goes down, I am still burning a few hundred watts under load.

Idea 3: A series pass emitter or source follower (IGBT or NFET) with 375 volts worth of zeners at the gate. A crowbar to blow the input fuse if the FET(s) fail (shorted drain to source).
The source follower should drop the least voltage / power when the input voltage is below the 375 volt zener stack.
Oops: If the solar array only drops to 425VDC and I pull 2 amps, 100 watts through one or more 600 volt high current FETS may be hard to manage.

Idea 4: A buck circuit to chop the DC (discontinuous mode ?) when over 375 volts and then stay full on below 375 all the way down to 200 and eventually 0. Still will use a crowbar.
Oops: Inductors and capacitors big and bulky

I have tons of power resistors, zeners, and a few ways to get 170, 450, or 1100 VDC to test with (at very low current). I don't intend to hook up to the solar until I get something that works.

Any suggestions ?

 

[BradtheRad]

I have experience with PV panels, deep cycle batteries, and inverters.

It will be difficult to get solar panels to power an inverter. Panel output V varies with load.

Batteries are needed for both storage and voltage stabilizing.

My inverter would not operate unless the input voltage was inside a certain range (namely typical battery volt level).

My unloaded panel put out 23 VDC. I must conclude that my inverter would only operate off a panel if I were to put a dummy load on the panel so as to bring voltage down to mimic a battery. Then I would attach the intended load.

If the load were to change (example, refrigerator motor cut off), then panel output would rise or fall accordingly. The inverter would stop operating (probably).

Does your system have batteries?

 

[betwixt]

I have a similar problem but haven't had time to fully evaluate the options. I wouldn't recommend dropping the high voltage though, all that does is waste the excess as heat, no matter what components you use.

My best first guess at the solution is a distributed (one per panel or pair of panels) change-over switch, possibly using MOSFETs so that the whole array can be configured in different ways. for example, all panels in parallel or all in series. The individual switches would only have to be rated at the maximum voltage and current of each panel and the high voltage 'stacking' problem could be overcome with opto-couplers at the input of each switch. The switches would be wired at each end of each panel except the two outermost connections which would be permanently wired. In one switch position, the negative side of the PVs all connect to a common rail, in the other position, they connect to the positive end of the next panel in line. A similar arrangement would be used at the other end of the panel. Of course, it could be done with pairs or groups of panels to achieve the optimum output voltage for grid-tied or off-grid.

Brian.

 

[hughmanoid]

No batteries. There are 40 panels that I can't rewire unless / until I have to reshingle the roof they are mounted on.
There are 4 banks of 10 panels in series that feed a DC disconnect with 20 amp fuses. Each panel is rated around 200W.
The idea is to have access to a fraction of the 7 KW array if the power goes out for a long period of time (days) and not have to run a generator to keep the freezer or refrigerator going when the sun is up.

---------- Post added at 10:09 ---------- Previous post was at 09:44 ----------

BASICALLY I WANT TO RUN A SERIES PASS OR BUCK REGULATOR IN DROPOUT MODE MOST OF THE TIME !
THE REGULATOR WOULD ONLY REGULATE UNDER PERIODS OF LOW / NO LOAD WHEN THE PANELS EXCEED 400VDC.
MOST CLAMP OR OVERVOLTAGE SOLUTIONS SEEM TO FOCUS ON BRIEF POWER SPIKES OR SURGES.
SORRY FOR SHOUTING, but hopefully anyone else that skims this thread will see it better.

I probably can't tamper with the existing panels without voiding the warranty. The existing grid tie system was designed to reduce input current losses by running the inverter SMA Sunny Boy SB7000 at a higher inout volatge plus the physical arrangemnt of panels made 4 strings of 10 easier / less costly than 5 strings of 8.

I have googled myself silly trying to find a solution.
I hope to get anywhere from 2 to 5 amps from the panels and don't mind burning off a couple hundred watts using diodes, resistors, zeners , toaster ovens or whatever since this is only for emergencie or the apocalypse.
I am sure I can find a way to bypass the sacrificial load in stages to minimize heat if the voltage is under 400 volts.

Thanks for your input.

Brian E.

 

[BradtheRad]

Since your panels are wired for 400 or so VDC, the circuit below is the most convenient way to power an AC appliance from the array... PROVIDED you can access a center tap between the panels somewhere. You only need to attach a single 14 gauge wire to one panel.

This may or may not be possible as you point out. It's only a remote chance that the conversion equipment already has attached this. It depends on whether it draws on the entire 400-450VDC output, or creates a split-phase setup, etc.

The mosfets/transistors are energized one at a time. (The diagram shows the control pulses which come from the control module whose details are not shown.)

As a result the refrigerator receives square waves or 'modified' sine waves. You'll have to experiment to find out what duty cycle runs the appliance best. The mosfets must never be energized simultaneously (for obvious reasons).

 

[hughmanoid]

Excellent idea but from what I have read up on the existing inverter, there is not center tap.

I plan to build three 'drop modules' that each have a power fet in parallel with 3 50W 24V zeners.
Voltage goes to the VICOR VI-M63 DC-DC (200-400) to 24 converter. Then to 3 drop modules
in series and then to ground. I can use a comparator and possibly optoisolators to successively
turn on the fets in the drop modules as the voltage drops.
When the volatge is below 375VDC, all modules would be 'on' (minimal voltage drop).
As voltage goes up, modules are turned 'off' (zener clamps voltage drop to 24V for each stage).
When input Voltage approaches 450 , all three fets are off and 72 VDC is dropped between panel
ground and Vicor DC-DC converters ground.
If the Vicor input hits 390VDC or more, I will fire a crowbar and blow a fuse.

I could also stack the drop modules on the high side.
I don't plan on using gate drivers since the VICOR module is not happy with large Dv/Dt changes.
Hopefully the drop modules will gracefully work in succession and not oscillate at the transition points.
At noon, the existing inverter is running around 380VDC. More power if the temperature drops.
If I disconnect the AC, the inverter draws no load and the panels hit 450 or so.
The calculations for maximum panel voltage are based on maximum insolation and record low temperature.
As if it is going to freeze at noon in the middle of summer.

Still open to elegant solutions from any one else. Thanks !
I may put two series 12VDC fans in place of one FET in the bottom drop module.
That gives me continuous cooling and power for logic, leds etc...
If I find a bunch of high current / high power 12V zeners cheap, I'll parallel the fans.

The zeners voltage drop goes up in response to current and or heat so they can be
parallel and share the current without one going into runaway.

 

[BradtheRad]

Your approach can work. It has the advantage of using less expensive components (rather than a bulky coil). Your eggs are not all in one basket (as they are with one big coil).

And if something goes wrong you have replacement components, and you will know how to fix it. This is crucial if we have societal meltdown.

* One technique not mentioned yet is switched capacitors. It will not necessarily be easier to construct than the voltage drop method you have discussed. However it has efficiency on its side (and elegance). Connect a stack of capacitors in series. Charge them quickly to 400V. Then switch them so they are in parallel. This divides the initial charge by the number of capacitors.

That's the theory anyway. It has problems of its own. It will require massive capacitors. One bad capacitor can mess things up.

I plan to build three 'drop modules' that each have a power fet in parallel with 3 50W 24V zeners.

As you may know, it is problematic to parallel heavy loads. One of the devices can have slightly different characteristics. As a result it could hog excessive current. It could burn up prematurely. Thus it is wise to install a load-balancing resistor at each dropping device. These can be very low ohms.

* It appears you wish to find a way to make a voltage regulator, compatible with the high V source. There are mosfets/transistors rated for that high a voltage, and which could endure the heat. It would require a bias circuit which likewise is compatible with high V. It could be made to automatically fine-tune itself so that the load always gets 13 V. You might even get by with just one large dropping device, rather than having to parallel zeners, etc.

This link (tpub.com website) may provide ideas:

https://www.tpub.com/neets/book7/27k.htm

Re-reading your OP, this looks the same as your Idea #3. Please disregard this if you've already decided it's infeasible.

 

[hughmanoid]

I have built several buck or boost simple switchers using ICs with integrated FET switches but havent come across
any that will run the logic side at a reasonable voltage but switch up to half a KV and use resistor divider
to a feedback pin. CAN YOU SUGGEST ANY ?
The closest I have come seem to be specialized ICs for surge suppression or require funky transfomers.

I thought of the capacitor series parallel thing awhile back with a big wheel of caps and some goofy arangement of wires
that when spun around, alternated from series to parallel to charge another massive cap.
Or a kind of loom looking thing with a paddle that bounces back and forth between two boards that alternates the + and - of each cap between two rows of contacts (one side series, the other parallel). Sort of a cross between steam punk and Gilligan's Island. My experience with capacitive voltage multipliers (Cockroft Walton) et al have shown to have severely limited current and big caps are pricey. I have one toroid the size of my fist but down know the current, voltage and henrys may be around 40 ?
I even thought of getting a big honkin surplus industrial 120 240 480 transformer and experiment with switching one side wired for 480 to the secondary wired for 120 and pray. They are out there but not cheap and probably heavy.

I have several N and P channel Fets 600 to 1200 Volts and several amps to 40 amps in TO-247 packages but don't have any experience blowing them up in order to understand power dissipation and the data sheets.
Also have a few IGBTs around 500 to 600 V and single and double digit amps.
I have some large older heatsinks drilled for TO-3 and medium to small ones for TO-220.

The series pass worries me because the device may not saturatE and drop double digits of volts at several amps.
I UNDERSTAND THAT SEVERAL IDENTICAL FETS CAN BE PARALLELD FOR SWITCHERS.
CAN SEVERAL FETS IN PARALLEL SHARE NICE WHEN NOT IN SATURATION USED AS A SERIES PASS REG ?

Time to go out to the garage and bolt some stuff together....

 

[BradtheRad]

It's apparent you have a wide background of experience which you can bring to bear on the present challenge. Also a grasp of the problems that will pop up.

Here's a few more concepts...

* Drop voltage through an electric water heating element. (Just like using up excess power from a water or wind powered generator.) Electric water heater elements are available in values from 1000 to 4500 watts at 230V. A typical water heater has 2 such elements. A 1000 W is about 50 ohms resistance. A 4500 W is 12 ohms resistance.

I don't know how similar this is to using light bulbs (your 'bad' idea #1). Maybe it will work since the element is not incandescent.

* A refrigerator may need 3x surge current each time the motor starts. If your inverter operates on 12V, the current draw could surge to 100 A at such times. Your panels may not be able to provide this much instantaneously.

* You must adjust the voltage drop units so they are able to respond quickly, both to a sudden drop or increase in transmitted power. If they are slow, the inverter may cease to operate due to detecting an undervoltage or overvoltage condition.

 

[hughmanoid]

I think I've finally come up with a self regulating system.
Instead of a series of 24V 50W zeners that are each bypassed by a fet,
a cascade of series pass regulators may be simpler.

The zener drop module idea got complicated by having to start with all the fets on.
As the voltage increased, the fets would turn off successively which meant having an
another fet to invert each stage. Biasing several stages got confusing.

So I plan to test how a series pass / source follower works in drop out mode.
I can calculate the power dissipation for each stage and easily change the
zener voltage reference/clamp for each gate since they will only need a few mA.
One concern is using smaller TO220 or TO247s, I may have to consider that at 5 amps
and 10 volts, thats 50 watts per device. That means adding more stages to compensate
for lower voltages per stage.

I also ordered a hefty IGBT module (2 in a half bridge) that may be easier to work with.
I believe they are designed more for hard switching (full on or full off) and may not like
being in or around the linear region. Since there are two in series, I can split the V drop between them.
MG100Q2YS1 absolute maximum ratings: (1)collector-emitter voltage:1200V; (2)gate-emitter voltage:±20V; (3)collector current:IC:100A, ICP:200A; (4)forward current:IF:100A, IFM:200A; (5)collector power dissipation:800W; (6)junction temperature:150℃; (7)storage temperature range:-40℃ to +125℃; (8)isolation voltage:2500V

I may even try using it as PWM driven switcher instead.

Brian