TECHNICAL MANUAL

THE SMART REGULATOR

Contents

Why Have a SMARTRegulator?
Design Philosophy
Hardware Description

Charge Switch
Load Switch
Controller
Indicators

Control Software Description
Operation
Charge Control

Boost Mode
Sulphation Test
Return to Boost Mode
Float Mode
Program Selection

Load Control

Low Voltage Cut Out
Waste Energy Use
Use as a Shunt Regulator

Temperature Compensation

Installing a Temperature Sensor

Installation

Mounting and Orientation
Environmental Protection
Remote Location
Manual Boost
Start Up
Reset

Test Programs

Simple Test Program
Complex Test Program
Test Sequence and Description

Fault Tracing

Simple test program
Complex test program

Why have a SMART regulator?

This simple picture is complicated by the limitations of the battery. If you leave the photovoltaic panel connected to a battery, the battery will slowly charge up. If you are putting in more energy than you are taking out, then the battery will eventually become fully charged. If this continues, the battery will become overcharged and be damaged by corrosion of it's plates and loss of electrolyte. To avoid this, some sort of charge regulator is used to stop the charging when the battery is full.

Charge regulators
A charge regulator is an electronic switch which senses when the battery is full and either disconnects the panel from the battery or diverts the energy away from it. Most modern regulators are of the type which disconnect the panel with a switch in series with it. These are called series regulators.

Almost all regulators use the battery's terminal voltage to determine how fully charged it is. As a battery gets charged up, it's terminal voltage increases. Single stage regulators simply allow the battery voltage to increase until it reaches a set limit (in the range 14.5-15 Volts for a 12V lead acid battery) and then turn the charge current off until the battery voltage drops below a lower limit. The charge current is then turned back on and the cycle repeats.

Two stage regulators improve on this by allowing the voltage to rise to a high voltage initially (typically 15V) and then change to a second mode which maintains the battery voltage at a lower level. The first stage is referred to as a boost charge and the second as floating the battery.

Simple regulators
Both these single and two stage regulators use only the immediate battery voltage to determine battery state of charge. They make a decision about the state of charge based on only one piece of information - the battery voltage at the time. Because they rely on this scanty information, they are easily fooled and will make the wrong decision in various circumstances. They are really quite simple regulators, having little built in intelligence.

Problems with simple regulators
Based on our experience manufacturing these regulators over the last ten years, we have observed that simple regulators tend to make the following mistakes:

• They undercharge batteries with high internal resistance (due to sulphation or bad connections).
• They can boost charge a battery too often if an occasional heavy load is applied.
• Under some conditions the charge current can oscillate rapidly on and off causing annoying flickering if any lighting or other voltage sensitive equipment is connected.
• This fast switching can cause annoying radio interference.
• They waste energy in the afternoon when the batteries are full.
• Voltage transients due to inverters or large inductive loads such as motors can cause incorrect changes in regulator state.

Smart regulators
Imagine a human battery technician was controlling the charging. An experienced operator would not make a decision about battery condition based on only one piece of information (ie. the current battery voltage). The operator would ask such questions as `how much was the battery used last night', `how quickly did the battery voltage rise',`how much charging has taken place today',`are the users on holiday',`how many days since I last did a boost charge' and so on. Having put all the information together, an intelligent decision could then be made. Can we distill the intelligence of a good operator and build it into a regulator? This is what we have tried to do with our Smart regulator range.

Design Philosophy

We decided to base the design around a small microcontroller integrated circuit. This single chip microprocessor controls all the operations of the regulator. It allows all the pieces of information about the battery state and charge history to be easily gathered, stored and interpreted. The controller is programmed with a sophisticated set of rules to help it to decide what to do.

Hardware Description

The charge switch
One switch is between the charge source and the battery. This is called the `charge switch' and is used to turn the charge current on or off. It is implemented using power mosfet devices in a configuration which will block current flow in both directions when the switch is off. Both the charge switch and the load switch do their switching on the negative side of the circuit. This means that the regulator is suitable for positive ground systems. It can also usually be used with negative ground systems as well, but this should be checked. (If the battery negative is grounded, there is usually no need to ground the solar array negative.

The use of power mosfets avoids the problems with relay mechanical and corrosive failure. The charge switch is protected from large reverse current flows due to charge side short circuits. If the reverse current exceeds a safe value, the protection mechanism operates to turn the switch off quickly. Once turned off, the switch will not reconnect until the charge source voltage rises to slightly above the battery voltage. This will not happen until the short circuit has been removed.

The protection mechanism also allows the charge switch to turn off at night to block any reverse drain through the panels. This is done by periodically turning the charge switch off briefly. If the open circuit voltage of the charge source is less than the battery voltage, as it is at night, then the switch will not reconnect until morning. A reverse blocking diode will not be needed in most cases.

The load switch
The other switch is connected on one side to battery negative. It can be used to switch a load connected to the battery or to switch a relay coil if larger currents are needed. The switch is implemented using a type of power mosfet called a `Topfet' which is protected against excessive current flow, overheating and high voltage transients (e.g. switching spikes). If the protection is triggered, the switch stays off until it is reset periodically by the controller.

Note: the regulator gets it's operating power from the battery connection NOT the solar connection. Hence, if you only connect a solar panel to the regulator and no battery, it will not regulate the output of the solar panel.

The controller
The device used as a controller is a single chip processor from the PIC series by Microchip Technology Inc. It is used to measure the battery voltage, the voltage across the charge switch, the temperature sensor, the B- sense line and the state of the program selection switches and the adjustment trimpots. These measurements are stored inside the controller and used to determine the correct state for the switches and indicators.

The indicators
Four indicator leds are provided. Two of these indicate the boost or float states. The load on led indicates that the load control switch is turned on. The charging led comes on when enough charge current is actually flowing to exceed the threshold. This can remove the need for a current meter in some installations.

To enhance visibility under high ambient light conditions, the leds are recessed in a black surround and are non diffused high brightness type.

The components are mounted on a circuit board, covered with a protective conformal coating and attached to a black anodised aluminium heatsink. A plastic clip on cover is placed over the circuit board. All units are given a 48 hour burn in, then retested to improve reliability by reducing the `infant mortality' failure rate. Quality certified suppliers and assembly services are used and all items carry serial numbers and are traceable.

Control Software Description

The controller reads each of the voltage inputs, linearises and corrects them relative to a reference voltage. This cancels out many sources of measurement error.

The battery voltage is formed by measuring the B+ voltage and then correcting this with the B- sense voltage and the temperature sensor reading. Both the B- and the Temperature sense inputs are ignored if nothing is connected to them.

Once the battery voltage is formed, it is fed to a digital averaging filter with a long time constant. (about 3 minutes) The effect of this filter is to smooth out short term changes in the battery voltage and allow the control algorithm to respond to the underlying trend. This is an important part of the regulator operation. It has the effect of introducing a time delay of approximately 5 minutes between a battery condition becoming true and the regulator responding to it. However, large changes will be responded to faster than small changes. The filter allows the regulator to ignore sudden changes in battery voltage due to changes in load or charge currents. The user needs to be aware of this slowed response. Some users have assumed the device was faulty because they did not allow it enough time to respond. A further consequence of this delay is that testing of the regulator by observing it cycle through all it's states is not possible in a short time.

To make testing possible, two special test programs have been built in. These allow the regulator to be fully tested in a short period of time. They also allow users to verify correct operation of their systems. There is a simple test mode designed for the end user and a complex one designed primarily for our factory testing.

To prevent the control program crashing due to noise, watchdog timer protection is employed. This will force a reset if the processor is not following the correct sequence for more than one second. The controller has an internal clock which syncronises with the start of charge each day. To aid in determining the battery state of charge, the controller records four pieces of data. These are:

• the overnight low battery voltage
• the maximum battery voltage in the early hours of the morning
• the charge amp hours accumulated since sunrise
• the number of days since the last boost cycle.

These are used, along with the current battery voltage and charge current, to sort out the correct course of action.

From these pieces of information it is possible to sort out the various battery conditions encountered and to make the correct response. This is how it tackles the problems listed earlier.

Battery Undercharge
If the battery has been left in a low state of charge for more than a month or so, it will begin to suffer sulphation which causes it's internal resistance to rise. When you begin to charge a battery in this condition, it's terminal voltage rises rapidly to a voltage that would indicate full charge if the battery were in good condition. This fools most regulators into thinking that the battery is full and so they stop charge too early. This can also occur if there is a loose connection in the wiring between the regulator and the battery.

What allows us to distinguish a fully charged good battery from a fairly flat sulphated battery? One factor we can consider is the rate of rise of the battery voltage. If it drops to a low voltage overnight or under load and then rises very rapidly once charging starts, we would be suspicious. By keeping track of the maximum voltage in the early hours of the morning (often a good approximation to the no load voltage), the lowest average voltage after charging has ceased for the day and the rate of rise of the voltage under charging, we can put together enough information to determine if the battery is full or not. If not, the charge time is extended. regardless of voltage.

Too frequent Boost
The flip side of the problem above is when the load is low compared to the available charge eg. when you are away on holidays. Under these circumstances a boost charge each day is undesirable. By setting the battery voltage at which the regulator begins a boost cycle to be low at first and then raising it a little each day, the boost frequency can be adapted automatically to the use pattern. (see below)

Rapid switching interference
After charging, we try to keep the battery voltage floating in a given range. To do this, the charging current is switched on for a short burst and then switched off again repeatedly. If something that is voltage sensitive such as a light is switched on during this time, it appears to flicker due to this switching. The switching on and off can cause audio interference to AM radio sets in weak signal areas. To stop this, a minimum off time is enforced. This spaces the changes so far apart that they are a lot less noticeable.

Line transients
Ripple or spikes on the battery voltage (often from inverters) can sometimes cause problems. The smart regulator averages a large number of voltage readings to get an average voltage and makes it's decisions on this. The spikes or ripples are smoothed out in the averaging process.

Operation

Charge Control

Boost mode
The maximum boost voltage depends on six variables:

• the minimum overnight voltage, Bmin
• the maximum voltage over one hour a few hours before dawn, Bmax
• the number of amp hours of charge so far that day, AH
• the charge current, I
• the boost maximum adjustment trimpot, Badj
• the program selected.

If the battery voltage exceeds this test voltage, then a set of tests are performed which allow a full battery to be treated differently to a sulphated battery.

If a significant amount of charge (approximately 5-10% of capacity) has been put into the battery since charging began that day, then the controller will change into the float mode. If not, then Bmin and Bmax are examined to determine if the quick rise in voltage is because the battery was already fully charged or because it is sulphated up.

The Sulphation test
If Bmax, which is a fair approximation to the battery open circuit voltage, is over 2.08 V/cell and Bmin, the minimum voltage overnight is above 1.95V/cell, then the battery is considered to probably be fully charged and the float mode is entered.

If not, then the battery is considered to be probably sulphated. This is based on 3 pieces of evidence characteristic of sulphation:

• low open circuit voltage
• voltage drops substantially under load
• fast voltage rise under charge.

Once the batttery voltage is greater than the set point, then the rate of rise is tseted. If more than 10Ah has been put into the battery that day then the controller will switch to float. If not, then the battery voltage is considered to have risen too fast and then battery condition tests are performed to determine if the charge time needs to be extended. If Bmax was less than 2.08 V/cell then it should have taken more than 10Ah to have reached the cut off and so charge is extended. To indicate that charge extension is occuring, the float light is flashed once per second. If Bmax is >>2.08, but Bmin<<1.95, then charge is extended because this indicates a possible high resistance condition.(Sulphation)

If the battery voltage rises to greater than boost max +0.13 V/cell the controller goes straight into float mode. This prevents charge from being extended too far.

Return to Boost Mode
The boost cycle frequency should depend on how much the battery is being used. If it is being used heavily, the boost should be frequent and if it is not being used, then boost should only occur occasionally. This is achieved by forcing the controller to return to the boost mode if the battery voltage falls below the boost return voltage.

The boost return voltage is set at 2.05 + BD/240 V/cell where BD is the number of days since the last boost cycle was done. What this achieves is to slowly raise the boost return voltage each day. On the first day, the battery must undergo significant discharge before it will return to boost, ie. it must be being used heavily. Each day it will require lighter and lighter loads to trigger the boost return until finally the boost return will rise above the no load voltage of a full battery forcing it to have a quick boost charge. This stirs up the electrolyte and helps to prevent stratification. The maximum time between boost cycles is roughly 10-14 days.

Float Mode
In the float mode, the controller tries to keep the battery voltage up above it's rest voltage but not above the point at which gassing becomes significant.

This could be done by linear control of the current, but this requires large heatsinks. It could be done by using a fast switchmode control scheme, but this generates a lot of interference or else requires the use of bulky filters to suppress the switching noise. The technique that has been chosen uses slow speed switchmode control combined with slowed pulse rise and fall times to achieve float control without excessive heat or interference.

It works like this. When the battery voltage falls below 2.22 V/cell, the charge switch is turned on. The battery voltage will then rise if the charge current exceeds the load current. This can be a fraction of a second if the battery is well charged and there is significant charge current. When the battery voltage rises above 2.35V/cell, the charge switch is turned off again and the voltage falls. When the voltage falls below 2.22V/cell again, the switch is turned on again and the cycle repeats indefinitely.

Because the voltage rise time can be fast, the filtered battery voltage cannot be used to determine when to switch off. The instantaneous (i.e. unfiltered) battery voltage is used. This means that the charge pulse can be short. The minimum charge pulse width is about 0.6 second. Be warned, this is so short that you may miss it unless you are watching carefully.

To avoid the flicker due to voltage modulation and radio and audio interference effects, a minimum off time of about 20 seconds is enforced. The off time will vary with the rate of fall of the battery voltage. If there is a substantial load on the battery, the voltage will fall quickly and the minimum time will apply. If there is very little load on the battery, there may be 5 minutes or more between charge pulses. Because of the minimum off time, the battery voltage may fall significantly below the 2.2V/cell threshold if a significant load is applied. This broadening of the float voltage window is the trade off for reducing the rapid oscillation which would otherwise occur.

Program Selection
There are 6 different programs that the regulator can follow. These are designed to cover most of the common situations in which the regulators are used. If you have special requirements we can often change the regulator to suit them.

Choosing the correct program
For operation with lead acid batteries with liquid electrolytes (the most common sort of storage battery), choose the NORMAL program. This has all the smart features enabled and an adjustable boost maximum of 2.42-2.58V/cell. We suggest 2.5V/cell.

If sealed gel batteries or valve regulated batteries are used, many boost at all. If this is the case, use theLOW program. Boost max is adjustable 2.25-2.42V. Sulphation correction is disabled. If the Boost max setting is below 2.33 V/cell, then the boost condition will be disabled and the regulator will be controlled by the float maximum of 2.33V/cell.

If problems with inverters cutting out due to overvoltage are experienced, use the SIMPLE program.

Note: the float voltages are the same in each program.

Load Control

Low voltage cut out
If the load on the battery is connected to the load terminal, then the load can be disconnected to prevent overdischarge if the battery voltage falls below the disconnect voltage. (User adjustable from 1.82 to 2.0 V/cell)

The load will automatically reconnect when the battery voltage rises above 2.17 V/cell. There is a time delay of about 5 minutes before disconnect and reconnect will occur. This delay is achieved by using the filtered battery voltage to compare with. This means that disconnect will occur faster if the battery voltage drops significantly below the threshold. The reconnect voltage has been set high enough to ensure that some recharging of the battery is done before reconnection will automatically occur. This prevents the load oscillating on and off but does not require a long wait until the battery is fully charged before reconnection.

If the load current required is greater than the rated load current, then use the load control switch to operate a relay with an appropriate rating. This sort of load control can be very useful if used on the lighting circuits. If the user goes away and leaves a light on accidently, then the load disconnect will prevent battery damage.

Warning: do not connect an inverter to the load terminal unless it will draw less than about 80% of the rated load current. (e.g. 100W is the maximum inverter load on a PC1212). The inverter will normally connect straight to the battery and use it's own low battery protection circuitry. If the load disconnect or the load switch protection is accidently activated, the load may be reconnected by briefly disconnecting the Bat+ regulator connection and then reconnecting it.

Waste energy use
When the battery is fully charged, there is often a few hours of charging left in the day. This energy could be used for other activities such as pumping water, charging a spare battery bank or irrigation. When the charge dump mode is selected, the load switch turns on when the regulator is in float mode and is not charging the battery. This can be connected to another load directly or used to switch a relay.(see connection diagram) Once turned on, it remains on for at least 5 min to avoid rapid on/off cycling.

In the direct connection method, the load is taking energy from the battery and the solar panel is switched on now and then to replace this energy. This is best for voltage sensitive equipment. The use of a relay to divert the charge current somewhere else makes maximum use of the available energy but requires that the alternate load be suitable for direct connection to a solar panel. If the alternate load is a reserve battery bank, then some thought must be given to regulation of the charge to that battery.

Use as a shunt regulator
The load switch can be used to make the controller perform the function of a shunt regulator rather than a series regulator. It can even function as both a series and a shunt regulator at the same time.

To use it this way, the regulator is put in charge dump mode and the dump load is either connected directly to the load terminal if it is small enough or the load switch is used to switch a relay controlling the load. When the battery needs charge, the dump load is turned off, but when it is full, the dump load will be turned on periodically to dump the excess energy and maintain the float voltage on the battery.

This scheme has been used with windgenerators, water turbines and steam engine generators. It also allows a solar component to be used with these sources via the charge switch. If there is no solar input, use the simple program to avoid the possibility of false sulphation correction occuring.

Temperature Compensation

Installing A Temperature Sensor
The temperature sensor TS1 is supplied in a sealed polypropylene moulding with a 3 metre cable.

The sensor is electrically a current source which is proportional to temperature. It is electrically polarised. Care should be taken to connect it in the correct polarity. The negative wire (the one with the stripe) of the sensor connects to -. TThe cable may be extended if it is too short.

The sensor must be installed in good thermal contact with the battery case. Do not place the sensor near but out of contact with the batteries. This will give false correction because the air temperature will not be the same as the battery temperature. Good thermal contact can be achieved in a variety of ways. Here are some suggestions.

1. Glue the sensor to the battery case.
2. Wedge the sensor in the gap between two batteries in the bank with a piece of foam rubber. The foam rubber will hold it against the battery as well as sealing around it.
3. Place the sensor against the battery and cover it with a duct tape strip which goes all the way round the battery and wraps back on itself a couple of times. This will prevent it coming undone easily.
4. Sit one of the batteries on top of it. A space can be cut in the support shelf and a piece of foam rubber used to hold it against the battery.
5. Wedge between the battery and the wall using foam rubber to insulate it from the wall.

Installation

Environmental Protection
The regulator is mounted on a rugged anodised aluminum heat sink. The adjustment trimpots are sealed and the circuit board is covered with a conformal coating. The regulator should not be mounted where it will get wet. This is particularily important in marine use. It is not possible to guarantee that the coating is 100% pin hole free. Because it is not possible to stop water getting into a sealed box, The case design is open to allow condensation to escape from the circuit board area quickly.

Remote location
A separate battery negative sense terminal is provided which allows the regulator to be positioned some distance from the battery without wiring voltage drops affecting it's accuracy. The regulator can correct for up to a +/-1.2V difference between the real battery negative and the regulator B- terminal. If the correction is not needed, then leave the B- sense input unconnected.

Manual Boost
An alternative use for the B- sense input is for manually forcing the regulator into boost mode. To use this, wire a normally open push button switch between the B- sense input and the BAT- terminal. Holding the switch down for about 1 second will force the regulator into boost mode.

Start Up
The regulator wakes up in float mode with Bmax set to 2.1V/cell. This assumes a fully charged battery at startup.

Reset
The unit can be reset by putting the switches into test mode and then returning them to the desired setting.

The Test Programs

The first one is the simple test program which is described in the operating instruction sheet. This program merely switches on and off both the charge switch and the load switch which allows the user to make a quick simple check that the system is functioning correctly.

The second test program is mainly for the manufacturers internal use. It enables the correct function of all the pieces of hardware on the regulator to be verified individually and calibration to be established and then checked.

Without these test programs, proper testing of the regulator is difficult and rather time consuming. With the programs it is a relatively simple task.

Simple Test Program
The simple test program allows correct regulator (and system) operation to be verified. This makes installation and fault location much easier. When installation is complete, the installer can set the internal program switch to simple test mode (all down). This causes the regulator to exercise each part of it's own hardware in turn. This allows the user to verify correct operation of the regulator and also the system connected to it.

If, some time later, the user suspects a regulator fault, the test mode can be used again to check the regulator. This makes it a lot easier to diagnose faults over the phone (one of my pet hates!). To use the test program, set all the program select switches off.(all down) The sequence of events in the test is:

Note: the charge current must be greater than the charging indicator current threshold before the charging indicator will turn on. (see specifications)

Complex Test Program
The second test program is more complex. It is mainly for the manufacturers internal use. Each piece of hardware is exercised individually and can be checked at a reference point to ensure accuracy. The overall calibration of the voltage measuring process can also be checked.

Start the complex test program:
Set switch 1 on, switches 2 & 3 off. (If the regulator was already in the simple test mode, it will finish that sequence first before starting on the complex test). Switches can be changed with the power connected.

Test sequence and description
The test sequence consists of 7 short tests and then finishes with the tests of the load and charge switches exactly as for the simple test routine. The sequence is an endless loop which will continue until the switches are changed to another setting. Putting the regulator into test mode does not destroy any of the charge history information collected by the regulator during normal operation.

The 7 short tests last for about 6 seconds with about a 2 second all leds off gap between them. The test sequence will appear to go too fast to the first time user because they are not ready for what is happening. However since this test is primarily designed for our internal testing, we run it this fast because once an operator has memorised the sequence it actually seems to be rather slow!

Test sequence
(initial setup delay of about 2 seconds. All leds off.)

1. Internal calibration correction value. This is related to the processor clock speed and is of no direct relevance to a user of the device. This will mostly show the Load led, but can also show the Float led.(2 second all off gap - load switch is turned on in the gap)

2. Calibration check. This is used to verify that the regulator is reading battery voltage accurately. The nominal battery voltage is applied between Bat+ and Bat- and the led output will show what is being read internally. (12.0 applied for a 12Vreg, 24.0 for a 24V reg). The led output is arranged so that:

   Led On	Meaning
   Load	internal reading is less than 12.0 (24.0)V.
   Boost	12.0 (24.0) volts is being read internally.
   Float	internal reading greater than 12.0 (24.0)V.

   The yellow led covers a voltage range of about 0.06 volt (0.12). Internal calibration is considered to be acceptable if the yellow led comes on somewhere in the range 12.0+/-0.07V.(24.0 +/-0.14) This is checked with a variable power supply with a coarse and fine control. The coarse setting is adjusted to 12.0, and the fine control is used to sweep over the calibration range to check the voltage at which the yellow led is on.(2 second all off gap )

3. B- sense line check. When open circuit shows as Float. Shorting B- sense line to B- will cause the Boost or Load led to show indicating that the B- sense input is working correctly. (2 second all off gap)

4. Temperature sense line check. A variable resistor is connected between T+ and T- to act as a temperature sensor (nominally about 7K with a 12V supply). Varying the resistor value should sweep the display from Load though Boost to Float, indicating correct function of the temperature sensor input.(2 sec all off gap)

5. Load disconnect trimpot check. As the load disconnect trimpot is turned from full anticlockwise in a clockwise direction, the display will show Float, then Boost at the mid point and then Load further round. Hence the trimpot operation can be verified. (2 sec all off gap)

6. Boost maximum trimpot check. Same as at point 5. (2 sec all off gap)

7. Solar - voltage sense check. Used to check the solar voltage sense amplifier. Shows Float normally. (2 sec all off gap)

8. Flash all leds on for 2 sec (2 sec all off gap)

9. Flash all leds on for 2 sec (2 sec all off gap)

10. Put Load and Float leds on. Turn charge switch on. Charge led will light if charge current above the threshold is flowing.

Fault Tracing

Simple test program
The simple test program can be used to quicly establish if the basic charge and load switches are working. This can be done over the phone and can be used to determine if the fault is in the regulator or in the rest of the system.

If it is sunny, check to see if the charge light comes on indicating that charge is flowing during step 5 of the simple program. If no charge flows, then bridge the sol- to bat- terminals, and check again, with an external meter, for charge flow. This enables the charge switch to be checked. Put a load on the load terminal such as a light, and see it it turns on and off during steps 1-4. If the load does not switch off and on, then the load switch may be faulty.

Complex test program
If you want to check the calibration of the regultor, use a variable power supply instead of the battery and use step 2 of the complex test program. For further tests, follow the details in the test description.

Home

Regulators

Developments

Specifications

Downloads

Suppliers

Resources

No Frames