250W boost converter controlled by a micro controller

[sonnytiger]

This is for a senior  project in college, me and my partner need to connect a single 250W panel to a run of the mill water heater. The obvious solution is to just connect it directly but this is far from ideal as the maximum power capabilities of the panel will be unused (poor impedance matching). I was thinking of using a boost converter and a micro-controller, with the micro controlling the boost converter to get the most power out of the panel. I am not really sure how to go about doing this as I am in electrical engineering and don't have much in the way of experience in power electronics. Any help is greatly appreciated, if I missed anything just let me know.

Thanks!

[electronupdate]

Your post is a bit unclear.  Could you please provide more details.  Is the "250W panel" a solar panel?

It would help if you could write down the full requirements of the problem.  Can you describe the panels output voltage and current?  Can you describe the "run of the mill water heater"?  AC? DC? 120VAC?   When the panel creates more energy that that which can be stored by the water heater where does it go?

[Marco]

When the panel creates more energy that that which can be stored by the water heater where does it go?

Heat in the panel.

[sonnytiger]

Excess energy (unlikely to happen) will just go unused. We have been given no information on the panel but it will be under 50V, likely in the 20 to 30V range, so 10A max is what I would design for. We are applying the boosted voltage directly to the heating element so it is just going to be a resistive load. so DC, and the output voltage needs to be varied according to input from a micro-controller (or similar device)

Hopefully I cleared some things up, sorry about that.

[NiHaoMike]

Use a push pull configuration with the output winding directly connected to the heater. Use a dual MOSFET driver chip and a microcontroller with a built in SMPS control block to simplify the design or use a common push pull SMPS driver (such as the TL494) and a low pass filter to drive it from a regular microcontroller PWM to keep things simple.

For real extra credit, implement Shannon Liu Quadrature Drive to drive a refrigerator compressor that is used as a heat pump. (Not as far fetched as you think - the GE Geospring uses a refrigerator compressor.)

[Richard Head]

Watch out for flux walking with a voltage mode control push-pull design .

Use a push pull configuration with the output winding directly connected to the heater. Use a dual MOSFET driver chip and a microcontroller with a built in SMPS control block to simplify the design or use a common push pull SMPS driver (such as the TL494) and a low pass filter to drive it from a regular microcontroller PWM to keep things simple.

[sonnytiger]

Wow that sounds like it would work but a lot of that is over my head haha, I will search around on Digikey for a chip that matches your specifications. That sounds cool but the the purpose of our project is to simplify as much as possible the transference of sun energy to hot water, so a compressor and plumbing would be out of the question.

[sonnytiger]

For further clarification, I am going to vary the output voltage of the converter to match the power being delivered by the panel. For example, at 250W around 70V is needed. But if the panel was only supplying 100W then I would reduce the voltage down to ~44V. I am not sure how I will tell what the peak power the panel is putting out at a given incident sun level but I can try and figure that out.

I am also currently looking at the TL494 chip, I think I will try and use it, any tips on trying to use it?

Thanks!

[dannyf]

Freescale is the king in this segment of the market, for good reasons.

[muvideo]

If I understand, you will implement an MPPT algorithm?
That stands for maximum power point tracking.
There some appnotes for that around, I remember on
form ST an perhaps from NXP.
Probably your Solar Panel will be a 72Cell unit, that
will work around 33V most of the time, but the best voltage
will depend on temperature and irradiation.

As for the topology, since you dont need isolation,
and the voltages are low, probably a simple boost
converter will be easy enough to control using a
microcontroller, and you dont need fancy level shifters:
a mosfet, a mosfet driver (or two bjt emitter follower),
two (big) rectifiers, a (big) inductor, and some capacitors.
The "hard" part will be the calculation of the right inductor
size vs ripple current and pwm frequency. There are lot
of online calculators and resource.
For the feedback the simplest thing will be to measure the
current trough the load.
For protection you need fast overvoltage and overcurrent,
the first will stop the mosfet in case of output overvoltage,
the second in case of overcurrent trough the mosfet ( using
a resistor between the source of the mosfet and ground).
A buck converter can be also used, but you either will have
an high-side switch or your load will be floating vs ground.

[NiHaoMike]

Watch out for flux walking with a voltage mode control push-pull design .

Use a push pull configuration with the output winding directly connected to the heater. Use a dual MOSFET driver chip and a microcontroller with a built in SMPS control block to simplify the design or use a common push pull SMPS driver (such as the TL494) and a low pass filter to drive it from a regular microcontroller PWM to keep things simple.

You'll be surprised how many cheap (and not so cheap) inverters get away with voltage mode control. It works fine as long as the duty cycle is low enough to allow the core to reset. It also helps that a solar panel has a rather limited current output so if the core starts to saturate, the voltage would drop and the MPPT will compensate by reducing the duty cycle.

The easiest way would be to select a heating element with a low enough resistance to allow the use of a simple MOSFET to perform PWM. No inductors needed and a common microcontroller with conventional PWM would work great.

Even more cost effective would be solar thermal panels, but then you'll need to run some plumbing.

[nctnico]

A good idea is to have current limiting for each PWM cycle in hardware. In one of my designs I used a microcontroller to control a PWM PSU. I've used a flipflop (to enable the MOSFET driver) which gets set by each PWM edge and reset by overcurrent.

[sonnytiger]

Maximum power point tracking is exactly what i want to do, it has to be a run of the mill electrical hot water heater, and the average is overwhelmingly is 3kW at 240V (around here at least), that's a really cool little trick there with the flip flop over current protection, I like that. I am going to concentrate on getting something running first. I will respond with more info and request for help when I go to school tomorrow and discuss this with my partner.

Again, thanks for all the help!

[NiHaoMike]

You could use the water heater for its tank only and do a solar thermal setup.

[Phoenix]

Building a 250W boost converter (that works well) is a decent challenge, with plenty of hazards along the way, how long have you got?.

I think you need to break the project down a bit more, before you go building hardware you should be able to answer most of these:

=Basic Specifications
   Solar panel - what voltage range/current range do you expect? Where will the MPP be under various operating conditions?
   Heating element - what's its ratings (power, voltage, current, resistance, any thermal coefficients)?
   Power electronics - what type of converter is appropriate for the supply and load you just spec'd? Buck, boost, buck-boost? Isolation - forward converter?

With this info you can build a prototype converter running open loop from a function generator and controlled DC source.

=Switching Converter Design Points
   MOSFET (P or N channel to make gate drive easier?)
   Make sure free wheeling diode is a fast diode (may or may not (probably not) be schottky)
   Capacitance - make sure you have some smaller (100nF~1uF) non polar capacitors in parallel with your larger electrolytics cause you will have significant current ripple
   Inductor - designing for continuous/discontinuous conduction? When you look at inductors they may have different average/peak current limit and saturation current
   Switching frequency - I would guess in the order of 100kHz
   Boost converter with no load may ramp up to high voltage (and boom)
   Don't prototype on breadboard - creates all sort of problems!

=Further Requirements
   MPPT - what algorithm (P&O, incremental conductance, ??)? What sensors/circuitry are/it required to implement this?
   Converter safety - detect over voltage/over current/MOVs? What sensors/circuitry are/is required?
   Microcontroller - safely interface to gate drive circuitry, safely interface to measurement circuitry (switching converters+EMI go hand in hand). How will you power the uC (from the panel, externally?)

With this you can interface the uC to the converter and try to make it run properly!

[NiHaoMike]

Assuming you have to use photovoltaics to operate some resistance heaters (not a very good idea except when there's no other use for the power), a 240V 3000W element would have a resistance of about 19 ohms. To get it to dissipate 250W, it would need to be supplied with about 70V. To keep things simple, you could pick a panel with a MPP voltage just above that and use a MOSFET to PWM it. You'll want a freewheel diode just to handle stray inductance but it doesn't need to be anywhere near the ratings of one used in a buck converter. You can use a low PWM frequency to reduce losses, but then you'll need to use more input capacitance so the panel doesn't see too much ripple current.

Note that most water heaters actually have two elements, of which only one normally operates. If you wired both elements in parallel, you'll only need just under 50V to achieve the 250W goal. You can even be a little fancy and use two MOSFETs with the PWM clocks 180 degrees out of phase to reduce the input ripple current.

[sonnytiger]

If what you say about the elements in parallel is true then that would be very good, we need to keep the voltage below 50V so we don't have to follow a bunch of code guidelines that would increase costs. I like the PWM thing as, thanks for the ideas!