designing a charge controller for hybrid system

[Warpspeed]

Wire up all the power components, mosfet, choke, flywheel diode, input and output electrolytic capacitors to each other directly, with very short lengths of nice thick wire.

You can still run your gate driver and microcontroller on the breadboard.
That should make a very big difference to how it all works.

 

[diksha.k]

Wire up all the power components, mosfet, choke, flywheel diode, input and output electrolytic capacitors to each other directly, with very short lengths of nice thick wire.

You can still run your gate driver and microcontroller on the breadboard.
That should make a very big difference to how it all works.

But, I can assume that this convereter is working well.

 

[Warpspeed]

But, I can assume that this convereter is working well.

Its not working well if it still drops two volts flat out.

With a 17 milliohm mosfet, and no more than about about five amps from the panel, there should be no more than 0.1 volts absolute maximum.
You have about twenty times that.

 

[BradtheRad]

I agree with Tony. Mosfets frequently seem to need a lot of coaxing before they turn on completely.

You might want to do simple tests on it, to see how high the gate voltage must be, to obtain desired current through the body of the device. That's how you'll know you reduced 'On' resistance to a minimum. Hold it with your fingers so you know if it's burning up.

 

[diksha.k]

You might want to do simple tests on it, to see how high the gate voltage must be, to obtain desired current through the body of the device. That's how you'll know you reduced 'On' resistance to a minimum. Hold it with your fingers so you know if it's burning up.

Ya MOSFETS were getting warmer

- - - Updated - - -

Another thing i want to clarify is:
I am using a potential divider across the solar panel for voltage sensing.
The resistance values are:
upper arm:99.9K(connected to positive of panel),lower arm=21.8K(connected to negative of panel) connected across panel.
then voltage b/w resistors is given to analog pin of the uc.
As there's 10bit adc within
the procedure to calculate voltage across panel is:
voltage across 21.8k=[(adc value)/1023]*5V.
voltage across panel= [(voltage across 21.8k)*(21.8k+99.9k)]/21.8K.
But whatever voltage i am getting as sensed by the uc is 2V less than expected.
i.e say if true voltage across panel=16V it senses it as 18V.

 

[Warpspeed]

voltage across panel= [(voltage across 21.8k)*(21.8k+99.9k)]/21.8K.

That should be = voltage across 21.8K multiplied by 21.8K / (21.8K + 99.9K)

If the mosfet is the only thing getting noticeably warm, then that must be what is causing most of the voltage drop. It can only really be as Brad says, insufficient gate drive voltage.

Try it again with just a 9v or 12v battery connected directly between gate and source, and see if that reduces the voltage drop.

 

[diksha.k]

That should be = voltage across 21.8K multiplied by 21.8K / (21.8K + 99.9K)

No :as per this image :**broken link removed**

here vout=voltage across 21.8K and vin is panel voltage(as per diagram above). So to get panel voltage what i have mentioned in right.

If the mosfet is the only thing getting noticeably warm, then that must be what is causing most of the voltage drop. It can only really be as Brad says, insufficient gate drive voltage.
Try it again with just a 9v or 12v battery connected directly between gate and source, and see if that reduces the voltage drop.

I tried using a 9V battery it worked as with the same condition no improvement then i tried with 18V that was satisfactory with voltage across battery rising to 14.40V and panel voltage=14.60V.
So as my driver circuit produces vgs=8.54V. I guess 8.54V is insuffecient

- - - Updated - - -

Datasheet suggests 10V is vgs during on state

 

[Warpspeed]

Look at it again, its bottom/total which is what I said.

What you said in #85:
total/bottom 22.8 + 99.9 / 22.8

voltage across panel= [(voltage across 21.8k)*(21.8k+99.9k)]/21.8K.

Anyhow, you have found the problem.
Most mosfets turn on reasonably well with 9v, but not all.
This one obviously requires a much higher gate voltage.

I only suggested 9v battery because its something almost everyone already has.

 

[diksha.k]

ya thanks,
Say for 100% duty cycle the voltage as i said for complete turn on of mosfet battery voltage is raised to 14.4V and panel voltage droped to 14.6V.
But,
simultaneously there will be sensing of battery voltage as well to operate it in safe operating area(soa).
But when for a ""moment"" of 100% duty cycle the battery voltage has raised to 14.4V implies conntroller will sense it has exceeded the soa(as 13.5V say is max voltage to charge battery to 70% of soc) which will make the controller to turn of or go in constant voltage mode but, as soon as controller goes in cv mode battery voltage drops to discharged state voltage so again controller switches to mppt mode. Wont this action be to & fro leaving the battery uncharged.

 

[Warpspeed]

That can never happen, the system will never see 100% duty cycle in normal operation, because the solar input voltage will be regulated to the MPPT voltage.

If the duty cycle tries to increase higher than that, it will pull the solar input voltage down below the MPPT voltage. The control system will then back off the duty cycle to keep the solar panel voltage correct.

When the battery reaches the 14.4v set point, charging current will slowly be reduced to hold it at exactly 14.4v .

The control system never just instantly switches on and off between 100% and 0%.

First it holds the solar panel voltage constant, with whatever solar input is available whenever the battery is below full charge.

Then it holds the battery voltage constant when its fully charged, and allows the solar voltage to rise above the MPPT voltage.

Its either doing one or the other, and it slides gracefully from one mode to the other mode.

 

[diksha.k]

ok,
but, what i observed is this right?
that is: when duty cycle=100% panel voltage=14.6V and battery voltage=14.4V though battery was not charged say around 11.9V before i hooked it up for 100% duty cycle(for mosfet completely turning on ).

 

[Warpspeed]

Its certainly possible with a small battery that is pretty deeply discharged to pull the voltage right up with sufficient charging current.

What should happen when its all set up properly, is that the voltage at the solar panel should sit at the set MPP value. The battery voltage may rise up to 14.4v pretty quickly, if the battery is pretty flat and the internal resistance fairly high if the battery has not been recharged for a long time..

The controller will limit the battery voltage to 14.4v and the solar panel voltage will probably also be pretty high to start with.

The battery terminal voltage should fall as it gains some initial charge, panel voltage should fall to MPP and bulk charging should then go on for quite some time.

 

[diksha.k]

The controller will limit the battery voltage to 14.4v and the solar panel voltage will probably also be pretty high to start with.

Yes still i haven't performed a complete control action that is maintaining the vmpp as well as battery voltage not exceeding 14.4V (as per soa). But, as i found that @ 100% duty cycle voltage across battery raised instantly to 14.4V then, as per my algorithm 14.4V almost at edge 80-90% of soc hence it should terminate charging and begin with constant voltage mode that is what made me ask you this question.

 

[Warpspeed]

Testing at 100% is interesting, because it provides a pretty good test of controller losses, and it certainly showed up the insufficient gate drive voltage problem.

But it will never run at 100% once you get your control algorithm working under real operating conditions. It will definitely go to 0% duty cycle at night and get reasonably high during a blue sky day, but never reach 100% duty cycle.

Battery condition and operating conditions are a study in themselves !

But once this system is up and running, (and given time for the battery to settle down) it will be fascinating to watch how it self adjusts to changing conditions.

All it needs is a control algorithm to maintain constant solar panel voltage, over ridden by maximum allowable battery voltage. Once you have that, it will pretty much self adjust and look after itself.

 

[diksha.k]

busy with my examinations.
about wind turbine:
As pmdc generator is used in the wind mill i remember you mentioning about using anemometer to find wind speed then, either by experimentation or if having a look up table, controller has to decide what power the dc gen in giving out.
But, if i go other way round.
Say,i know no of poles used in dc generator, kind of winding used for armature(lap or wave) hence descide no of parallel paths and no of conductors used.
Now i am mentioning this because:
say oc voltage generated can be known by voltage sensing:
then,
as per faradys law:
e=d(phi)/dt
but, d(phi)=speed(N)*(flux per pole(phi))*(no of poles(P))
dt=time per revolution=60/N
Z=total no of conductors of armature ,A=no of parallel paths of armature
therefore :
e=[N*(phi)*P*Z]/(60*A)
As everything is known except N it can be found out and so, looking at look up table or generator characteristics i can find what may be the power developed.

 

[Warpspeed]

That is only part of the problem, the great unknown will be the aerodynamic behaviour of the blades and the gearing to the alternator.

As with the solar panels, there will be an optimum loading point, where the whole system combined (blades + gearing + alternator) rises to a power peak, then rapidly falls on its face with any further loading as the blades suffer flow separation and go into turbulent stall.

In theory it should be possible to monitor the power output and constantly adjust the electrical loading up and down using a perturb and observe algorithm, to follow changing wind conditions.

In practice, gusting wind speed and stored rotor inertia will confuse the software which will be constantly chasing its own tail, and trying to correct for its own corrections. While this will work reasonably well in a very steady wind velocity, you will only get that where there are no up wind obstructions such as trees or buildings.

Of you are located in a flat desert or on the coast with a steady on shore breeze, and slowly rising land from the beach with no up wind obstructions, it all becomes much easier.

Where there are trees, buildings, or hills up wind, within a few miles, there will be turbulence and gusting, and this makes accurate load control much more difficult.

Any feedback system can only react to changes in the output, and inertia of the blades with its stored energy needs a long time constant to stabilise. But gusting wind is random in nature, and its not really possible to tune a PID loop or a perturb and observe feedback system to give anything like optimum results over wide range of rapidly fluctuating wind speeds.

A feedforward system will behave much better, and it cannot become unstable in the same way a feedback system can, if negative feedback becomes positive feedback under some conditions and it stars to surge in rotor speed.

If you use a completely separate independent fast acting anemometer to monitor the highly variable wind speed, and use that (via a lookup table) to adjust the electrical loading, the inertia of the blades will be working in your favour instead of against you. Electrical loading will respond almost instantly to sudden wind speed changes and stability of the control system is assured as nothing is fed back to create instability.

 

[BradtheRad]

Your approach is scientific in view of the fact you were not provided with a real wind turbine for testing purposes. The battery in the output, creates a different situation regarding net V, A, W.

Here is a repeat from my earlier post... When I was investigating home power setups, articles gave a frequent admonition to make use of excess generator energy. Otherwise the turbine spins out of control and becomes a danger. I did not see articles saying stop the turbine as the battery reaches end of charge. (Often I see turbines which are motionless, so it is done on purpose, perhaps being stopped manually.)

Therefore as the battery reaches end of charge, your controller must start to divert excess energy. The load can be a water heater, or light bulb, etc. Since its voltage is less than the battery's, it would gobble all the current if you were to leave it connected all the time. To make a diversion circuit will complicate your design.

 

[Warpspeed]

This is all true and very important.
First get it delivering maximum power under a wide range of wind speeds.

Then think about diverting power when it cannot be otherwise used by either the battery or normal load.

Last thing is protection against storm damage, which needs to be mechanical.
Either feather the blades, or turn the machine side on to the wind so it stops turning.
There are some pretty ingenious ideas on how best to do this.

 

[diksha.k]

Before all this i had "thought " about using ward leonard method as wind turbine that is: 3 phase induction motor will drive a dc geneartor with field seperately excited to make it behave as pmdc generator(as the actual wind mill has pmdc gen) and varying the wind speed would be varying speed of induction motor this i was going to do using static rotor resistance control.

 

[Warpspeed]

Separate field excitation is bad news for wind machines because of the power lost exciting the field winding.
A few experimenters have attempted to use automotive alternators.
They very quickly discover that the two amps or so required to excite the field winding, consumed most if not all of the generated power at low wind speeds.

Permanent magnet machines, especially those using powerful rare earth magnets are definitely the way to go.
Then use an efficient switching power supply to convert whatever voltage there is up or down to whatever is required.