upgrading a diversion charge controller

[fabieville]

Is it possible to add a float charging/trickle charging system to a diversion charge controller? Meaning that when the diversion charge controller senses the battery reach a certain voltage and then diverts the incoming energy to dump load there will still be a float service/trickle charging still going to the battery just to maintain the battery.
Why i am asking this is because i realize that diversion charge controller doesn't give the battery a good charge like how PWM charge controllers does and the reason for this is that diversion chargers just senses when the battery as reach a preset voltage which doesn't really mean that the battery is fully charge as yet then it just sends incoming energy to dump load whereas the PWM chargers senses when the current been drawn by the battery drops to a very low current like less than 1 amp  then it switches to float mode to top of the battery.

So would it be possible to add this function to a diversion charge controller making it still behave like a diversion controller but charge the battery more efficiently with the aid of a float charging service or converting the diversion charge controller that it doesn't send the incoming power to dump load until the current been drawn by the battery falls to less than 1amp which would show that the battery doesn't need anymore current?

[Flux]

You could make a 2 stage diversion control where you charge initially to a higher voltage for a given time then it could drop to the lower state to a float value. These multistage charge schemes work well when charging from the mains but with wind or solar you have to charge with whatever current is available. On a good sunny day or a good windy day there is no big problem but on many days the charge comes and goes and deciding how long it needs in the fast charge state before it drops to float is a challenge.

There are many approaches to this, some better than others but normally unless the system has long idle periods ( when you are away or something) it is normally sufficient to use a compromise voltage that gets the things up fairly well but doesn't boil things dry when left floating. For flooded cells you then give an occasional gassing charge at a higher voltage to equalise things and get then back to full charge.

So much depends on your daily routine the availability of sun or wind and the periods of no use.

I can't see the relevance of pwm here, you can make a linear diversion control or in most cases it is pwm anyway with a diversion resistor being switched on and off in a pwm sequence such that the input current balances any load and maintains a set voltage.

Flux

[Flux]

Having read your post again I suspect you have not understood what a diversion controller does. When the battery reaches a certain set voltage the diversion operates to maintain that voltage constant. It will divert just enough current to prevent the voltage rising, if you have some load it will let the charging current supply that load plus any extra current the battery requires to maintain constant voltage.

It never removes the charge and the battery is floating at the set point, it will draw any final charging current plus any self discharge current . This small maintaining current will not be diverted.

I thought you were wanting to have a higher initial diversion voltage for a bit faster charge and then drop to a lower float value to reduce water loss.

You haven't been looking at that stupid Oakley controller have you? that thing removes the charge and tries to divert current to a fixed resistor, bad for solar and crazy for wind.

Flux

[fabieville]

Having read your post again I suspect you have not understood what a diversion controller does. When the battery reaches a certain set voltage the diversion operates to maintain that voltage constant. It will divert just enough current to prevent the voltage rising, if you have some load it will let the charging current supply that load plus any extra current the battery requires to maintain constant voltage.

It never removes the charge and the battery is floating at the set point, it will draw any final charging current plus any self discharge current . This small maintaining current will not be diverted.

I thought you were wanting to have a higher initial diversion voltage for a bit faster charge and then drop to a lower float value to reduce water loss.

You haven't been looking at that stupid Oakley controller have you? that thing removes the charge and tries to divert current to a fixed resistor, bad for solar and crazy for wind.

Flux

In regards to your response the diversion charge controller that i has it doesn't bleed off the excess current to dump load while maintaining the battery at a set voltage, instead it monitor the battery voltage and as soon as it reaches a preset voltage for eg. 14.8V then it just diverts all the incoming power to dump load and then whenever the battery drain come down to a next preset voltage then the incoming power from solar/wind is then reconnected to the battery.

I prefer the one that you posted in your message as your one would definately keep the battery at a float charge but how could i convert mines to operate like yours or could you post the schematic for it please.

[ghurd]

I prefer the one that you posted in your message as your one would definately keep the battery at a float charge but how could i convert mines to operate like yours or could you post the schematic for it please.

Which one did he post?
The only one I see is "that stupid Oakley".

All real dump load controllers (even the ghurd controller) work like how he described.
G-

[fabieville]

I prefer the one that you posted in your message as your one would definately keep the battery at a float charge but how could i convert mines to operate like yours or could you post the schematic for it please.

Which one did he post?
The only one I see is "that stupid Oakley".

All real dump load controllers (even the ghurd controller) work like how he described.
G-

I am talking the ghurd controller or any other one that will work like how he described. Please send me a link with a schematic.

[ghurd]

That is exactly why I prefer the term "Dump Load Controller" to "Diversion Controller".

"Dumping" surplus power from the battery is usually understood.

"Diverting" power causes confusion.
It could imply it 'diverts' the excess power, or it 'diverts' all the charging current.

Diverting all the charging current is 'stupid' at multiple levels.  And it tends to involve automotive grade 40A relays, most often in parallel for 80, 120, and 160A versions.


For links to name brand (Morningstar TS-xx, Xantrex C-xx, etc) schematics, part numbers, Gerber files, operational parameters, etc, the best bet would be to contact the company directly and ask where they posted them.

[fabieville]

i was checking out the dump load controller on http://www.ghurd.info/

Could you please tell me what is the rated amp that it can handle or what do you change so that it can handle more current?

Also if using the dump load controller with a solar panel instead of a wind turbine then you would not necessarily have to put a load on the dump load section right? would it just send the excess current to dump load as open circuit(VOC) from the panel while maintaining the battery voltage?

[ghurd]

The ghurd controller simply turns mosfets on and off very accurately.  And quickly if need be.

The amp rating is based on the quantity of mosfets, and the type of mosfets, and the type of load per mosfet. 
And the decision of ghurd controller's resistor RX.

[fabieville]

The ghurd controller simply turns mosfets on and off very accurately.  And quickly if need be.

The amp rating is based on the quantity of mosfets, and the type of mosfets, and the type of load per mosfet.  
And the decision of ghurd controller's resistor RX.

what do you mean by the type of load per mosfet or u talking about the max current coming from the solar/wind?

This is what i have understand about the dump load controller.
For eg. if i have both my wind turbine and solar feeding the  battery then both devices would be connected to the battery at all times and when the battery reaches a preset voltage the controller would then dump the excess current to dump load and if i dont have any load connected on the dump load section then the excess current would then be dissipitated as heat in the mosfets?

Am i correct on this theory?

[Flux]

A dump controller must always have its resistive load connected or it just won't dump. Mosfets are series switches and are on or off. if there is no dump resistor present then no dumping takes place and no current flows in the mosfet and no power is dissipated.

You are right in that both solar and wind charge the battery directly and you need no controller until the battery reaches something near full charge. At that point the battery can't take all the charging current without the voltage rising and it starts gassing. The dump controller maintains the battery voltage constant by dumping exactly the current needed to prevent the voltage rising.

You could build a simple linear dump load with a transistor that diverted current exactly to maintain constant volts. In this case there would be dissipation in the transistor but the diverted current would be constant dc.

For larger power linear schemes are not very practical so mosfets are used as switches to switch in a resistor chosen to be able to dump the full wind and solar charging currents. Batteries see a load switched on and off as just the mean current of the switched waveform as long as the switching is above a few 10s of Hz.

The diverted current is now not smooth dc but a series of constant current pulses. When the pwm is zero there is no current. When 100% the current is that determined by the resistor. At 50%pwm the current will seem to the battery as though it is half the current determined by the resistor. By changing the pwm you can make the dump current balance the input charging current plus what the battery can take as charge.

Although the dump load is being pulsed on and off between two fixed values the battery regards it exactly as in the case of the linear dump.

Solar series chargers work in the same way, they are series controllers but it is not a linear regulator the current is pw modulated between zero and full charging current and the pwm is adjusted to maintain the set battery charging voltage.

I hope this clears up how pwm actually works, it may not be obvious for those not familiar with it.

Flux

[fabieville]

I understand now.
But I just have a few more questions to ask. In regards to choosing the dump load it has to be a resistive load that can always handle the max current the wind/solar can provide so that the dump load controller will be able to dump the amount of current needed to dump over right?

Also in the case of a simple linear dump load controller using a transistor why wouldn't this be feasible/practical for larger power schemes, is it because the amount of current the transistor can handle would be an issue?
Couldn't you just use bigger transistors that can handle more current?
Or is because the transistor would be dissipitating too much heat? Couldn't you just add more transistors if this is the case?
Or there is another issue using simple linear dump load controllers for larger power schemes?

[Flux]

Yes you are correct, the dump resistor needs to be sized to deal with the total solar and wind input worst case.

Typically it is difficult to get transistors capable of dissipating much over 200W. There were some beasts about with thyristor type packages capable of a great deal more but they were impossibly expensive and with everything going pwm I suspect they are no longer available.

The most common linear device with a sensible dissipation is the 2N3771. You could use some of them in parallel on a big heat sink and you can further reduce the dissipation by having a resistor in the collector or emitter. Choose things so that when the transistor is at halfvolts the resistor limits the current to the dissipation limit of the transistor.

In some ways it is just as cheap to dump power into transistors as resistors, the cost is not much different at about 150W. The heat sink cost is a killer unless you get it cheap or make your own.

I have used linear dumps up to about 300W, I like the simplicity and there is no radio interference that can be a problem with pwm. I have an experimental unit that I use for tests with an ancient thyristor type housing that I can load machines to over 1kW at 24v but I didn't have to buy the transistor.

You can make perfectly satisfactory linear dump loads if you want to go that way as long as your system is not too big.

I have never found mosfets satisfactory in linear mode except those special ones designed for audio use, the normal Hexfet type switching ones are very tricky to run linear and I would not even try for half the claimed dissipation.

I also have a suspicion that batteries prefer the pulsed waveform of pwm compared with the smooth dc from a linear dump. I don't believe in the desulphator stuff but there is something in the battery design that seems to work better if the current is rough.

[fabieville]

Couldn't you just have a fan cooling the transistor(s) while on a heatsink? wouldn't this eradicate most or if not all of the heat?

Not to go off topic but I always wonder how much heat does a computer power supply fan can cool down. I have a 20 amp solar pwm charge controller that i built and i have the fet and the diode on one of those heatsink that you find in a computer power supply. I am wondering suppose i am producing the 20amp would the fan be able to keep the fet and the diode cool enough and suppose i was to upgrade the circuit to about 60amp would the one fan plus that heatsink would be enough?

[Flux]

A fan will reduce the size of heat sink you need very considerably but it won't let you increase the dissipation from a transistor, the limit is the thermal transfer from chip to heat sink. To work at the makers dissipation you need the heat sink at about 25 deg and the chip will then be at near 150C.  When the fan dies the whole thing dies unless you have some built in safety devices.

The same is true of uprating any electronic devices with fan cooled heat sinks. those computer processors are very good indeed at getting the heat to the heatsink and the sink is also very good at being fan cooled. If you mount your own devices on that heat sink you will be limited by the heat transfer of your fets and diodes. The common TO220 package is not capable of dealing with much over 60W and that sets the limit per device.If you parallel lots of them and deal with the current sharing issues then you can probably get rid of a lot of heat from those fan cooled processor sinks.

Most people believe that fans are not that desirable for dump controllers, they die sooner or later. You end up with too much protection and more complexity or you risk the failure when the fan dies. Similarly there are good reasons not to use lamps as dump loads, there is the non linear resistance problem that causes surges but the other factor is that they eventually fail and you stop dumping.

flux

[fabieville]

when u parrallel more than one fet do you have to use a resistor between them to share the current equally? if you do how should i connect them and what size resistor i should use? Also how would should i insert the flywheel diode in the circuit? is it for inductive loads(fans/motors) on the dump load?

[DamonHD]

One scheme is to use a series of smaller (higher-Ohm, lower-Watts) resistors one driven by each FET separately, with no current sharing between them.

Rgds

Damon

[fabieville]

so please tell me how exactly i would connect them as well as connecting the flywheel diode in full detail please as i am kind of an amateur when it comes on to electronics. I know about it to a certain extent.

[Flux]

Dealing with switching circuits with mosfets is at the tricky end of amateur electronics. paralleling them in theory is simple but in real life it is far from simple. Even at the low frequency required for a charge controller you need to be very careful with layout, bits of wire act as inductors and you need to keep the inductive loops as small as possible. Similarly if the layout is bad your flywheel diodes will do little good.

If you are not familiar with the layouts needed you will probably be better off having one resistor for each mosfet and have a flyback diode for each fet resistor circuit.( if it is only one fan that is inductive you can drive it off one mosfet rather than the paralled bunch). You still have to drive the gates of the mosfets and you need a low impedance drive source to feed the lot but you will need small resistors to each gate and they need to be as close to the gate as possible as they mainly stop parasitic oscillation.

I can't draw circuits and post them here, I'm not that good with computers. If you follow ghurds instructions for his controller you should get some good ideas.

Flux

[ghurd]

Recommended:
Keep each fet load separate, and keep the amps per fet low.




NOT recommended:
But can be made to work if absolutely necessary.  Derate the fets even more than I derate them for seperate loads.

[fabieville]

In regards to the second diagram why isn't there a flyback diode in that diagram?
Also when you says I must derate them could I use a rating of each fet handling just 1/2 the rated max current so if the fet is rated for 10amp then I would only push a maximum of 5 amps in it just to be on the safe side?

I know that you said the second diagram is not recommended but as you said it can be made to work if absolutely necessary, so what i have ask above would it be the most efficient way to accomplish efficiency using the second diagram approach or what way I could apply the second diagram in the circuit without worrying about failure in the fets sooner or later?

And please tell me the exact value of the resistors in the second diagram that you have between each fet?

Also when using flyback diode the current handling of the diode have to match the load or can you just use any size for eg. 1amp rating?

[Flux]

I can't answer for ghurd but as the last diagram shows a heater it strictly wouldn't need a flywheel diode. I think it was just intended to give you an idea of the connections anyway.

Layout is critical for this parallel circuit and the de rating of the mosfets depends on how good your layout is. Unless you can design careful printed circuits with low inductance tracks and a short path for the flywheel diode you will have to do a lot of derating.

If you struggle through the International Rectifier site and look in the application sheets you will find some data on paralleling mosfets but it may be well beyond your means without an oscilloscope and a detailed knowledge of pulse circuits. The gate stopper resistors will be something in the order of 22 ohms.

There is a mistaken belief that as mofets have a positive temperature coefficient they parallel easily, This only applies at high pulse rates and with proper layout, at low frequencies you find severe limitations that are similar to second breakdown in normal transistors. In fact hexfets and similar are not at all good in the linear mode and even with pwm you will have to drive the gates very hard or you will be in the linear mode long enough for many strange things to happen.

I am not sure that even the commercial manufacturers have got all the bugs out of these controllers unless you use near non inductive resistors and short leads, some controllers have been known to do strange things with slightly unconventional load resistors. If they struggle then you are going to have to seriously derate things to even stand a chance.

The thing to look for when coosing mosfets is Rds on, this gives a far better idea of the current they will handle than the nominal rating. Even so you only benefit if you can drive them hard on quickly. Despite ridiculous claims for peak current handling the TO220 package is very limited thermally and you will do better with TO247. If you go to the makers data and look at safe operating region for dc loads you will find it is a lot less than the pulses you can handle in a high frequency switching supply. Probably safe to use half that value for one fet with decent heat sink but when you parallel things get worse quickly and more than about 3 in parallel will need a drastic derating.

Flux

[ghurd]

What Flux said.

Plus, I derate a LOT more than 10A rating to 5A.
The IRFZ44N is rated for ~50A.  I recommend 6A.
FET FDP7045L is rated for 100A, 300A surge, but do you think those tiny legs can handle 100A for long?

My friend does quite a bit of business fixing paralleled fet boards from some sawmills.  Poorest designed board anyone could make on purpose.
So, like Flux said, even the commercial manufacturers have issues with parallel fets.

The missing flyback diode is implied.  It won't hurt anything, so spend the 50 cents and use one.

Exact values depend on everything else in the system.

[joestue]

all you have to do is throw an appropriately rated capacitor on the board and then there's no derating needed (within reason) and it don't care how fast you pwm it (within reasonable gate drive limitations/switching loss.)

please note that the 20nH inductors are parasitic. you can't get rid of them. the inductance of the capacitor is included in the others.
the inductance of the resistor doesn't matter. it could be 10 millihenries if you use a 500foot roll of wire as a resistor... that's what the diode is for.
the 10uH supply side inductance is an order of magnitude estimation.
you can go here http://www.qsl.net/in3otd/ind2calc.html and calculate it yourself based on the physical layout of your battery bank, then add some.

if you keep the frequency low then the caps won't get hot.
otherwise the capacitor(s) need to be rated (for CYA liability) to the same RMS current that is flowing through the resistor(s).

[fabieville]

In regards to the flyback diode which is a very fast recovery diode that is used as a catch or freewheeling diode. Does it have to be a low voltage drop/Schottky diode even thou it is parrallel across the output instead of connecting in series on the positive rail going to the load which in that case you would add a low voltage drop one?

I found one that stated that it is a SUPER-FAST GLASS PASSIVATED RECTIFIER and the details was that it is super-Fast Switching for High Efficiency so I am guessing that this one would be appropriate. Here is the part number so u can check it out if u like and tell me what you think. However it did not stated low voltage drop in the detail thou.

Part Number: ER1600CT

http://www.datasheetcatalog.com/datasheets_pdf/E/R/1/6/ER1606CT.shtml

[Flux]

Volt drop is no big issue but it must be fast or hyperfast. It happens that schottky is fast and works in a different way that is advantageous. If you are within the voltage rating of schottky then they are fine. Otherwise hyperfast is your choice. Conventional diodes will do absolutely nothing to protect a mosfet. They are ok for energy recovery as freewheel diodes on power circuits with thyristors with large inductive loads.

Flux

[kevbo]

The OP is making a common mistake:  Float charging, and trickle charging are two different things.  Float charging holds the battery at a constant voltage regardless of current.  Trickle charging provides a constant (small) current regardless of voltage.  They are similar, in that they are two different ways to keep a battery "topped up".

Different battery chemistries work better with one or the other.  Lead Acid works best with float charging.  NiCads and NiMH work better with trickle charging.  Using the wrong technique can lead to problems ranging from shortened battery life to catastrophic things like explosions or fires.