95 % efficiency SMPS

[Thunderchild]

I think the knack is going to be to get a SMPS regulator to have feedback from the input rather than ouput as the "logic" is reversed, if the feedback pin detects a large volatge it will restrict the output and so the input will go higher (straight up to max) if the input in under voltage it will just increase output and bring it down firther. I suppose some extra clever circuitry could reverse the logic..

 

[dougy83]

I doubt the circuitry needs to be clever. I would think an inverting op-amp buffer referenced close to the SMPS reference voltage (the voltage expected to be on the FB pin when regulation is stable) would suffice.

 

[Thunderchild]

I doubt the circuitry needs to be clever. I would think an inverting op-amp buffer referenced close to the SMPS reference voltage (the voltage expected to be on the FB pin when regulation is stable) would suffice.

you merely prove my point, what all the mystery and cost that surrounds a MPPT supposed to be about, it looks like after putting our heads together with some imagination we have figure out what these things do or rather how the same effect can be produced (I put it like this to help not damage the MPPT manufacturers pride)

So steps are:

1) know your panels nominal voltage
2) build a smps thats is as efficient as possible
3) re engineer the feedback circuit so that the reference it taken from the input instaed of the output.

dougy any suggestions on how to do this ? (step 3), perhaps step down the Vin with a divider to something like what the feedback pin normally gets and then apply it to the feedback pin via an inverting op amp ?

 

[Sceadwian]

dougy83, notice the 98% efficiency is only in a simulation.

 

[smanches]

you merely prove my point, what all the mystery and cost that surrounds a MPPT supposed to be about...

It's because there is no single optimum operating voltage on a panel. Maybe in test conditions where you have perfect light during the entire test. But in the real world the amount of light hitting the panel changes throughout the day. This changes the voltage-current curve, which means the "optimal" voltage changes as well. That's what an MPPT is for.

 

[Thunderchild]

It's because there is no single optimum operating voltage on a panel. Maybe in test conditions where you have perfect light during the entire test. But in the real world the amount of light hitting the panel changes throughout the day. This changes the voltage-current curve, which means the "optimal" voltage changes as well. That's what an MPPT is for.

well in that case there is more to it. However if your gap is between 12.5 and 17.5 volts thats not a lot to play with anyhow. you might as well track the battery's charge and let the smps take what it can from the panel, if you set it up (knowing your batteries internal resistance) so that the charge voltage will be such that it's optimum for the max charge current you will acheive from the solar panel then your probably not far off getting as best as you can from the system

 

[dougy83]

Yes there is more to MPPT (maximum power point tracker) - it tracks the optimum power, thereby effectively changing your 17.5V reference.

I thought you only asked how to keep the panel voltage at 17.5V, which even by itself, will be an improvement on directly connecting the panels to a 12V battery. I would think that the MPP would be somewhere around 17.5V in other lighting conditions anyway.

 

[Thunderchild]

well then we are back to square one again, keep the panel around 17.5 volts by using the input for feedback instead of the output, the panel's natural voltage is 17.5 because that is just how many cells there are (36 cells X a little less that 0.5 V). so basically its a case of transforming 17.5 into 12.5-13.8 V and turning the difference into more current.

now won't a BJT have more loses than a FET ? for some reason the LT1074 (5 A output) uses an internal BJT, I suppose I could use the chips output to drive 2 or more fets in paralel and so keep losses to a minimum that way. although I'd probably be better off just using a proper control chip.

 

[dougy83]

well then we are back to square one again, keep the panel around 17.5 volts by using the input for feedback instead of the output, the panel's natural voltage is 17.5 because that is just how many cells there are (36 cells X a little less that 0.5 V).

The nominal voltage is usually printed on the back of the panel and may vary for different panels.

now won't a BJT have more loses than a FET ? for some reason the LT1074 (5 A output) uses an internal BJT, I suppose I could use the chips output to drive 2 or more fets in paralel and so keep losses to a minimum that way. although I'd probably be better off just using a proper control chip.

The LT1074 has a darlington output which drops a fair bit of voltage. Note that the efficiency isn't even 90% in the datasheet. You will be better to go for a controller chip with external FETs & synchronous rectifier drive. If you can get the whole thing integrated, that may be even simpler; obviously you won't be able to tweak as many components & you're limited by the IC choices.

 

[ronsimpson]

It looks like we can get close to MPPT by just watching input voltage.

I was just looking at the data on a solar panel. It heats up to 50F above ambient in bright sun light. This causes the MPPT point to move 20% in voltage.

The last MPPT controller I worked on hunted for the sweet point 100 times a second in 0.1% steps. It would hold the duty cycle at constant for one period of time and then measure the input power. Then it adjusted the duty cycle by a small step, waited one time period and measured power again. It the power went up then it continued in that direction. It the power went down it reversed direction and continued hunting. It used a PIC18Fxxxx as the PWM and had 1024 steps in duty cycle.

 

[Thunderchild]

The nominal voltage is usually printed on the back of the panel and may vary for different panels.

The LT1074 has a darlington output which drops a fair bit of voltage. Note that the efficiency isn't even 90% in the datasheet. You will be better to go for a controller chip with external FETs & synchronous rectifier drive. If you can get the whole thing integrated, that may be even simpler; obviously you won't be able to tweak as many components & you're limited by the IC choices.

yes the LT1074 has an efficiency of 80-85 %, basically what little I was gaining was being lost in the SMPS, only advantage was it stoped my battery charging at 16 V

I've ordered a standard charging controller for now as I don't want to ruin the battery, it says that it works on PWM could that be a round about way of saying SMPS ? well I'll find out when I check input current versus output

 

[Ubergeek63]

MPPT ... start by ditching everything you think you know about battery chargers and switchers

once you have the right inductor, power FET, and synchronous rectifier (another power FET) controled by a uP

At a constant PWM the battery will determine the "output" voltage and the resultant current times the current battery voltage is the power delivered to the battery. Add to that the regulator losses and you get the power drawn from the solar array.

since for small variations in current the battery voltage is essentially constant, if an increase in PWM results in a decrease in battery current then you are going the wrong way.

Dan

 

[Thunderchild]

so your basically saying that by varying the pulse widths you vary the power thats been drawn and having found the highest current output point you stick to it

 

[ronsimpson]

Yes. In a high-end system there are two functions.
1) Get the most out of the panel. Find and stay at the sweet spot.
2) Charge the battery in the best way. Limit current, voltage and battery temperature.

 

[Thunderchild]

I wonder if there is an analog way of varying the PWM to suit the best power.

I expect the charge current is detected with a damn small serieas resistor, where can i find a real low value resistor ? or maybe my own wirewound would do after all its not about a specific value and feedback voltage, the circuit should automatically home on the pulse width that produces the largest output.

 

[Ubergeek63]

you have to continuously correct for it therefore it is not a matter of "sticking with it" since environmental factors (light level and temperature) are constantly changing.

you need to continuously dither the PWM while monitoring the charge current to determine which way to change the PWM. This is the maximum battery charge rate. When the full charge voltage is reached the battery charge rules take control.

As for current sense resistors, they are easy to find:

Digi-Key - PWR4412-2SDR0100F-ND (Bourns Inc. - PWR4412-2SDR0100F)

for instance.

For something a little more elaborate:

Digi-Key - 620-1192-1-ND (Allegro Microsystems Inc - ACS713ELCTR-20A-T)

is an integrated current sensor that would eliminate the amplifiers needed before the uP A/D.

Dan

 

[Thunderchild]

well it sounds simple enough at least if you can program pics and the like I expect I need to learn to do that and then I'm laughing...

 

[Ubergeek63]

it cn be done in anlog but it is trickier... any micro will work... pics just seem to have a fanatic following

Dan

 

[Thunderchild]

well i already have the pickit2 programmer

 

[Ubergeek63]

it happens if more people knew how simple programmers were for the chips these days they would not be so reliant on opportunistic jerks reaching for wallets.

the fact of the matter is most micros these days only need RS232 translator chips and PC software to reflash the code as often as you like. They call it ISP short for In System Programming - solder the chip down and bring 5 pins to a connector. We do not even use a connector in our designs, just a card edge pattern.

Dan