impose current ceiling

[flippineck]

I have a supply connected to a load. The load is a constant purely resistive load. The supply voltage fluctuates fairly wildly and is capable of supplying a lot more current into the load, than the load is capable of surviving.

I could add a fuse in series and that would impose a specific limit on the current which was allowed to be supplied to the load. But, it would involve the cessation of operation when the fuse blew.

I want a component like a fuse that will sit there passing current up to a certain value, but once that value is reached, it maybe increases it's resistance to keep the current down. Impose a current ceiling rather than just blow.

I heard of using NTC thermistors to limit transient high currents but the transients I'm dealing with last minutes, maybe hours rather than milliseconds or seconds.

It's solar panels feeding into a charge controller - most of the time here in the UK it stays averagely grey but you get these odd half days where the sun randomly has a party, last time this happened my solar system, which had been working fine for weeks under average / usual weather, suddenly melted in a heap on the floor.

I'd rather just miss out on the best of the very rare days of extreme power availability, than spend orders of magnitude more on charge controllers, or have to keep replacing fuses.

I guess failing a completely automatic option, a manually resettable breaker of some kind would be better than nothing. I just wondered if there was a simple 2-wire component that would sit there passively and hold passing excess currents down, whilst not impeding currents below the ceiling.

I saw these https://en.wikipedia.org/wiki/Resettable_fuse but I wondered if I'd be able to get them in a sufficient power dissipation, and whether the perhaps not insignificant resistance below the threshold current would cause efficiency problems under normal operation?

I saw a jfet used as a current limiter by connecting it's source back to it's gate (?) somewhere.. does that sound feasable

I have solar panels capable of producing, under the sunniest of conditions, maybe 900W; whereas the controller's PV input is only rated at 480W. 95% of the time the panels only produce two, maybe three hundred watts but then you get these very occasional spikes right up toward the 900's.

So I guess I need something capable of potentially 'throwing away' 400-odd Watts under the brightest spells

Thanks for any thoughts.

 

[hevans1944]

Maybe you could modulate the current with a MOSFET in series with the solar panel. Full current when the MOSFET is on, no current when the MOSFET is off. Vary the pulse width to accommodate the sunshine. Initially, turn the MOSFET on and sense the current with an Allegro current sensor. Start modulating when the current rises above a certain threshold. Or just turn the MOSFET off when that happens, but it seems a shame to waste good sunshine when all you need to do is tame the current a bit. And you would need some way to reset the MOSFET on again if you opt for on/off control (like a circuit breaker or fuse). Perhaps a filter capacitor of a few dozen microfarads on the MOSFET output (after the Allegro current sensor) would allow PWM all the time without problems. I assume it was the charge controller that melted. You just need to feed it a proper diet (or starve it) when the solar panel output exceeds its capacity and/or the battery is fully charged.

So I guess I need something capable of potentially 'throwing away' 400-odd Watts under the brightest spells

Never throw away electrical power generated from sunlight! It's too damned expensive. Instead "tame" your solar panel output.

 

[(*steve*)]

You could also place a relay across a solar panel. when the current exceeds a certain amount, short out a panel (or another panel). When it drops below some value un-short the panel.

Hop's idea, combined with an inductor, capacitor, and diode becomes a buck SMPS.

 

[hevans1944]

... Hop's idea, combined with an inductor, capacitor, and diode becomes a buck SMPS.

Well, relays that short out some panels could work too, but what a waste of good sunshine. I didn't want to mention that there would have to be some extra circuitry associated with the MOSFET switch and Allegro current sensor, for fear of scaring off the OP.

Can't design anything now anyway because we don't know what the charge controller looks like or even how big a battery it charges. Probably don't need a buck converter because the solar array output voltage is probably already matched to the charge controller input.

I am having trouble understanding why the charge controller didn't just shut down instead of melting. What happens when the battery is fully charged and the charge controller doesn't need to provide any more charge? The open-circuit (unloaded) voltage of a solar panel is not substantially different than its voltage under rated load, is it? Why is it even necessary to "dump" excess power capability? Why can't you just not use it? What's really going on here?

More information, @flippineck, on what melted and why you think it melted. I suspect there is more to this "problem" than you have told us so far. What is the open-circuit voltage output of your solar array under best sunlight conditions? How much current can the array provide under the same conditions, and what is the terminal voltage under that load? Is it possible that the charge controller was trying to output more current than it was rated for? What is the voltage and ampere-hour capacity of the batteries connected to the charge controller? What is the average load and how much does it vary over a 24-hour period? Does the battery reach full charge during the day when the solar array is actively providing power? Is this system grid-tied so you can sell power back to the grid? Why do commercial installations with hundreds of kilowatts capacity not have this problem of "having too much current output"?

Methinks the problem is in your charge controller, not the solar panels. The MOSFET switch solution (or Steve's relay solution) may allow you "re-size" your solar panels to meet the limitations of the charge controller, but a more sophisticated charge controller would be a "better" solution IMO.

 

[flippineck]

Thanks guys I appreciate your attention on this. I still have the old melted mess of a charge controller in the shed (understandably the shop wasn't interested in a refund seeing as it was myself that messed up the maths) so I can dig up all the specs etc. My panels, on thinking about it, are a motley bunch of slightly differing specs. All the same basic size, electrical arrangement etc however being bought a couple of years apart, 3 of them are 100W panels and 3 of them are 150W panels (I think) with a few volts and amps difference between their specs.

I'll get all the parameters together & provide a full system diagram. Probably take a day or 2. Thanks.

 

[Harald Kapp]

I think you need an electronic current limiter.

 

[flippineck]

Looking at the characteristics graphs, that sort of thing looks ideal.

I read down at the bottom of your link.. "The entire current-limiter circuit can be packaged and used as a three-terminal device (Fig. 5)."

It'd sure be nice to find exactly such a pre-made current regulator / limiter in a simple transistor - like package. The low voltage drop sounds handy.

------------

When I blew my old controller up, the set up was as follows. Not the actual seller I used, just for info:

controller: http://www.ebay.co.uk/itm/121995813091 (tracer 3215RN)
max PV input power 390W when system running batteries at 12V

3 x 80 watt panels in series:

• Maximum Power Voltage (Vmp): 18.06V
• Maximum Power Current (Imp): 4.43A
• Open Circuit Voltage (Voc): 21.6 V
• Short Circuit Current (Isc): 4.89A

- in parallel with -

3 x 150 watt panels in series:

• Maximum Power Voltage (Vmp): 18.5V
• Maximum Power Current (Imp): 8.11A
• Open Circuit Voltage (Voc): 22.2 V
• Short Circuit Current (Isc): 8.92A

So I guess that works out at somewhere around 55V @ 12.5A i.e. 687.5W

no wonder 390W controller melted under 687.5W when the sun shone hard (it was the array feed lines that melted IIRC, in addition to the controller just going point blank u/s)

I'm guesstimating I could have done with a 7 amp fuse in the array line? or a limiter that'd keep things down to that level in an 'auto resettable' manner



----------

Right now (lashed 'em up today to get at least something up and running) I just have two of the 80 watt panels in parallel so the maximum I'm likely to put into the controller would probably be about 9A at 18V (162W) or perhaps a tad higher under really bright sun. The cheap chinese controller I'm currently using claims to have a maximum current handling capacity of 30A but whether that relates to panel input current, battery charging current or supply-to-load current or the number of sheep I'm not sure. I couldn't find an explicit figure in the manual for max PV input power.

I didn't actually get it from this particular ebay seller but it's the '30A' version of this: http://www.ebay.co.uk/itm/281391828746

I'm feeding two 110Ah 12V lead acid leisure batteries in parallel. I've just stuck a 5A fuse in series with the array for now.

------------

Planning ultimately on replacing the cheap chinese controller with something more substantial but still probably with overspecified panels in order to make the most of the dull sun that makes up 95% of the UK weather. So still looking for a way to handle these occasional 'spikes' of power (yeah, seems criminal to just 'throw away' good energy but, it never lasts for long. generally only get 2 or 3 days a year at best of really brutal hot sun here. The rest is trickle charge material, and I'm hoping to build a system that will at least gather sufficient power to run the lights in winter. I've a big diesel generator for the kettle and the cooker etc.

 

[flippineck]

Double post sorry, ran out of editing time. Just a quickie - I wondered if I could use a filament lamp as a current limiter? It would have to have quite a pronounced bend / knee to the IV curve to allow sufficiently free current below the threshold without much dissipation. I don't know if filament lamps with that sort of characteristic are easy to find

 

[hevans1944]

Automobile brake and turn-signal lamps have heavy filaments that require substantial current to "light up" unless they are LED lamps. Try some, perhaps two or more in parallel, to see it that works.

 

[flippineck]

Still in the process of trying the bulbs. Blown a few, need to buy more

Anyone any knowledge / experience of this sort of thing: https://en.wikipedia.org/wiki/Constant-current_diode

(would this be as simple a solution as splicing one two-lead component into the solar panel feed line?)

They look like they might only be available in very low current ratings though, I'm looking for current limits in the single-digit Amps range rather than the low milliamps

 

[Chemelec]

Why Not Split up your Panels, Make Two Separate Systems.
Two Sets of 3 Panels with one controller on Each one.

3, 150 Watt panels on one.
3, 100 watt panels on the other one.

 

[flippineck]

Good idea. The lower power panels were the ones I bought earlier, the higher powered ones were bought a couple of years later. In the intervening years the output of the same physical size panel had risen, panels of the same size in the lower rating seemed to have become unavailable.

Right now I'm only using the three lower powered panels to feed the small 'sun YOBA' brand controller (it describes itself as employing MPPT technology but I doubt it does, at least not in the usually accepted sense). I've discovered the input parameters for this controller are a solar array max voltage 48V, max power 380W. I'm charging two 110Ah lead acid deep cycle batteries and running eight or so 10W 240V LED lamps off them sporadically via a 2000W 12VDC-240VAC inverter.

Still want to investigate current limiting for when I add many more panels, beefier controller etc. Like I say, the really sunny weather is very hit and miss in UK and a large number of panels is useful to wring out useable power from the mainly dull days. Just gets a bit scary on the annual day the sun shines


Looking at these links, it seems high power Current Limiting Diodes / CLD's, might exist? Can anyone direct me to any actual examples in UK-deliverable parts catalogues?

http://www.scientific.net/MSF.615-617.911

https://www.researchgate.net/public...urrent_limiting_FETs_CLFs_for_DC_applications

 

[(*steve*)]

Current limiting can cause you to need to dissipate a lot of heat.

When I first read the title of your thread, a constant current circuit seemed to be the answer, but after I read what you actually needed to do this was not a solution I was going to suggest.

 

[flippineck]

Could I use something along the lines of this http://www.radio-electronics.com/in...rent_limiter/power_supply_current_limiter.php ?

Let's assume I incorporate a 15 amp fuse to provide a hard non-resettable safety limit to the current from the panels, but want to limit the output current that goes to the battery charging controller to a maximum of 7A, what sort of component values and types would be suitable. Would I need a large heatsink on the transistor and sense resistor? Would it be likely to need to be actively fan air-cooled?

Interested to hear more detail regarding what your concerns are Steve.

 

[(*steve*)]

Interested to hear more detail regarding what your concerns are Steve.

The main concern is that 50% of the power coming from your panels could be converted to heat.

That's a lot of heat.

Even worse, if the output is shorted, 100% will be converted to heat.

Whilst the characteristics of solar panels are such that you'd be very unlikely to hit the sweet spot to get maximum power dissipation, you would still be dissipating a lot of power. Most constant current circuits are designed for relatively low power use.

Another concern is that you may be operating with the device several hundred volts above ground. This makes insulation very important. Can you design something which can safely keep you away from 1000V (to give you an additional safety margin) and still be capable of transferring heat to the environment?

A good inverter will limit the maximum power it accepts in order to protect itself.

 

[flippineck]

The maximum voltage experienced by the current limiter, directly across it's inputs and as far as the input to the charge controller, at the moment, should only be between the three 80W panels maximum power voltage of 18.06V, and the 21.6 V open circuit voltage.

The maximum potential current under fault conditions of a completely shorted current limiter would be 14.67A (three 80W panels wired in parallel under full sun conditions).

The charge controller has one two terminal solar array input, and two seperate, two terminal outputs - battery bank + & -, and load output + & -.

I'm feeding the load output terminals of the charge controller into the inverter's 12V battery input terminals. I've witnessed the inverter does shut itself down if too much load is placed upon it, and recovers when the load is reduced subject to a certain amount of hysterisis.

The inverter's design is such that the neutral 240V output pin is internally bonded to the output earth/ground pin. The installation manual specifies that the inverter's ground output be attached to a copper earthing spike driven into the ground in field situations. Steve - Is this where your concern comes from regarding operation several hundred volts above ground? Or can you forsee a different circumstance apart from the mains generation side of things that would cause such high voltages?

I'll go back and have a good look at the Allegro current sensor - I'm thinking, this could possibly be arranged so that a big dissipation problem can be avoided - i.e. the system just disconnects the panels entirely when the current rises too high, obviating the need to actually handle any excess power at all. I guess I need in this circumstance, to ensure that natural failure modes all fall toward open circuit.

 

[(*steve*)]

For some reason I was thinking a grid connect inverter with several hundred volts at 10A or more when I was writing that. The insulation requirements in your case are far less stringent

It would be worth trying a modified version of this circuit:

Except do it with beefier transistors. For an easier time, make 3 of them, each designed to limit the current to (say) 3A and place them in parallel.

Unfortunately, as with all semiconductors, the initial failure mode is short circuit, however if you have multiple circuits you can fuse protect them. A 5A slow blow fuse for a 3A limiter would be appropriate.

The problem with using a darlington in this application is that it will develop a voltage across it of 1.2V at a minimum (plus the 0.7 across the sense resistor).

My modification would be as follows:




Q1 is an appropriately beefy and heatsinked N-Channel mosfet.
Q2 is some general purpose small signal NPN transistor
R1 could be 10k
D1 is a zener to protect the gate of the mosfet. 10V should be fine.
S1 is a 5A slow blow fuse
R2 is something like 0R220 for about 3A

Calculating the actual power dissipated is non-trivial. If you design for 80W then you'll be fine even if the inverter short circuits. The average power should be far less.

The inputs and the outputs are paralleled.

The more ways you split up the 10A, the more reliable will be the protection if one unit goes short circuit. I would suggest that you build between 3 and 5, appropriately rating the fuses.

It is perfectly reasonable to build one (set for 2A to 3A) and use it on its own for testing.

If you place a LED and a resistor across the fuse, it will indicate a failed unit (but only when all units have failed or the remaining units are actively limiting current)

 

[flippineck]

Thanks for that Steve. Going quiet while I spend some time looking at that circuit in detail, I'm very rusty so I'll be googling aplenty for a while.

Does 0R220 mean 220 ohms or 0.220 ohms

 

[(*steve*)]

The R is the decimal point, so it's 0.22 ohms.

The formula is approximately I = 0.7/R (or R = 0.7/I).

So the power dissipated in the sense resistor is approximately 0.7I. To keep things cool, choose a resistor about twice the value of the power dissipated.

In this case that would be 5W. A simple way to do this is to use ten 2R2 0.5W resistors. This is likely to be cheaper than a single 5W resistor and you can add or remove a resistor to trim the current should you want to make it marginally higher or lower.

Another bit of magic, is due to the fact that there is a constant voltage across the sense resistor, that you can add or remove as many 2R2 (that's 2.2Ω) 0.5W resistors without any fear that they will overheat (they will always dissipate 0.22W each).

Pick a mosfet that has a Vgs(max) of 15V, and has an Rds(on) of around 0R1 (100 milliOhms, 0.1Ω). Get something in a TO-220 package, preferably in a plastic package or with an isolated tab.

The power dissipation will be up to 25 * Imax (as set by the sense resistor). Check out the calculations to determine the appropriate heatsink size. Remember that the ambient temperature may well be above 25C when this is called on to dissipate the most power. The more conservative you are (e.g. 10 copies of the circuit at 1A each is far more conservative than 3 at 3.33A) the more reliable the circuit will be.

 

[flippineck]

I *think* I'm beginning to get how the circuit works - not sure about the zener diode though, how does it protect the gate of the mosfet.. with a 10k resistor upstream of it, where does the danger of 10V come from?