SC1 - a Shareware Controller

The design, schematics and PCB layot along with some documention is available here for the interested. I have some board available at production cost. This is by no means any business, all information on this site is provided purely in the spirit of sharing what I've done, an the limited supply of boards that I have is made available only in the same spirit.

The schematics and PCB layout were created by me and you should remember that I have no formal qualifications for the job, so if you embark on utilizing, in anyway, any part of this design, you are totally on your own. Absolutely no guarantees, except that I can vouche for having tried to provide true story here. Having said that, I must say that I'm reasonably satisfied myself with the overall board design, schematics and PCB layout.I think it might be used as basis for e.g. a university level embedded system project.

As the schematics and layout are available , you can and should use your own judgement to see if they are worth anything to you. This is definatelly not a project for beginners and newcomers. The style of writing in this page is aimed to experienced hobbyist, pros and experts should look elsewhere and not spoil their day with this kind elementary stuff ;-)

Note that to utilize this board potentially quite a lot of software is required; this page mainly deals with the hardware aspects of SC-1. Maybe some day I will write about the software and maybe even make some software available. At very least I try to provide enough information to bring the tools and board alive with GNU tools. Someday.

I touch a lot of subjects here ranging from DC-servo motors to IrDA communication and quadrature sensors. While briefly introducing the concepts this page does not try to provide tutorial on those, for all of these subjects there is a lot of good, even excellent tutorials on the web. Not to mention that I'm no authority on any of these, just standing on the shoulders of giants and paroting stuff.... ;-)

Get the Documentation

Schematics as a PDF-file

PCB layout as a PDF-file

I/O-Cross Reference by MCU pin

Connector Cross Reference by con/pin

Bill of Materials

Shopping List

Power Budget

Links to Datasheets and Stuff

How this Board Came to Being

Of course there are a lot of microcontroller boards around, so who needs one more? I had a look around, but did not like any that I found. Some were too pricy, some did not have the features I wanted. Besides, designing is half the fun the other part being construction. Anyway, sane or not, I deciced that I would have a go at designing a kind of hobby board with some interesting I/O and manufacture some PCBs. The result was SC-1 and this page.

Dual Personality?

System Block Diagram

What could be done with the board

Robotics with RC-Servos

Motors, motors, motors

Back to the 70's

MP3 player

Remote control your remote control

Remote control that old amp you built

GSM Micromouse

a Weather Station

Digitals scope or data logger

The above are just some of the possibilites that I had in mind when designing this board. All you need is some imagination and of course it helps if your know your digital circuits and can do a litle programming. Software is actually what this board is all about.

Walk Through the Design

The Power Supply

The board is designed to be powered from 9 - 36 V DC supply. The higher the voltage the more losses you will have on the linear regulators and thus more heat. As there are no heat sinks this puts a limit on either current consumption or supply voltage. Maximum input voltage of 36 V DC allows the use of simple rectified 24 V AC with sufficient filter capacitors (mind you if the line voltage is 10% above nominal you may be stretching your luck as some components might be over max spec). The minimum operating voltage of 9 V DC allows the board to be run from single 9 V block or 12 V car battery.

A word about the linear regulators. I chose to have them all of type LM317 which is adjustable, so that I need not to purchase three varieties. More components (two resistors) but they also have the nice feature that they can be used up 40 V input voltages, more than most fixed voltage regulators. In theory they can put out .5 Amps but the chosen D-pack (TO-252)package (without heat sink) limits the max power dissipation to about W at room temperature. So with 34 V input voltage you can only draw about 30 mAmps from the 5V line, hardly enough to power the MCU.

5 Volt regulators

3.3 Volt regulator

12 Volt regulator

MCU

Internal Flash

MCU operatin modes

External Memory

Solenoid Drivers

The outputs are protected against inductive spikes and short circuit (the chip was origanlly designed for automotive applications where ruggedness is vital). The short circuits protection turns of an individual output in case of a short. However, it is possible to destroy these chips with thermal overloading if you keep turning on the shorted output. Even though it turns off automatically it may eventually generate enough heat to destroy the chip. Keep this in mind when developing the software.

The chip has some built in diagnostic capacity in that if an output is in the off state you can check weather a load is present or not. And if an output is in on state you can check if it is shorted or not.

Despite of the immediate and fast over current protection it is possible to drive also loads with high initial (inrush) current, such as incandescant lamps. For this purpose you need to install the jumpers (J12,J13) to place the drivers in linear current limit mode.

In theory these drivers can drive 3 amp from each output pin. The Rds_on is 0.45 Ohms per output so the power dissipation is 4 W at max current which is quite a lot of heat, but you can parallel the outputs at will, so you can get this down to 0.5 W. With a maximum usable supply voltage of 50 Volts this gives 150 W of controllable power. With two chips that is 300 Watts, not much for disco but probably enough to discofy your T-bird not to mention controlling a dozen or so magnetic valves for automatic watering of flowers or a drink dispencer! Actually the board layout does not support this much current but you can augment the board with external wiring. Check the layout before attempting high currents.

Analog Inputs

These signals can also be used as general purpose digital I/O pins.

Inputs with 4-20 mA Loop Cmpatibility

The 4-20mA loop, for the uninitiated, is a simple two wire scheme which provides power to a sensor and the signal from the sensor over just two wires. The scheme is that an unregulated voltage is alway provided between the two wires and the sensors can use this voltage to power its internal electronics or any local circuits it needs. The signal is transmitted by modulating the sensors current consumption proportionally. The great advantage of using current as the signal carrier is that it is immune to electrical (capacitive or inductive overhearing) interference, because the current modulator at the sensor side will compensate for any noise. As (almost) any sensor is bound to use atleast some current, 4 mA has been defined as the zero reference and 20mA as the max scale for the current loop interface.

To use these inputs the sensor (or whatever) is connected between the input and a supply voltage. Any current consumed by the sensor is visible as voltage loss over the shunt resistor (R90,R91) and can be digitized with the MCU analog inputs. This simplistic implementation (on this board) sacrificies 20% of the AD converter range because the zero point (4 mA) is not compensated for in anyway. But the simplicity is worth that I thought. And if you do not mount the shunt resistors you can use these in the same manner as the other analog inputs (and also as digital I/O). Also these inputs are protected by a diode network (IC24).

Analog Outputs

Either buffered voltage mode or 4-20 mA current mode is available, configurable with jumpers (J4,J5).

In the voltage mode (jumper between pins & 2) the output voltage depends on the voltage divider (R86/R89 and R88/R87) and is of course limited by the installed opamp (IC21) output capacities and the 12 Volt opamp supply. To use this mode the load must be connected between ground and the output.

In the current mode (jumper between pins 2&3) a transistor (Q1,Q2) functions as an open collector controlled-current sink. The current is controllable between 0-20 mAmps making it compatible with the 4-20 mA loop spec (if there is one, it is such an ancient thing). To use this output mode the load needs to be connected between a supply voltage and the output.

RS232C Port

IrDA Port

IR-remote control

My bright idea was that to be able to send IR remote control commands to your VCR or TV would open up interesting possibilites. And to be able to receive command from standard remote conrols would make it possible to command the board with an off-the-selv remote controller (like the one that was left over from that TV set that was dropped out of the window when the license inspector knocked on you doormitory door).

You have the choise of two outputs and two inputs to drive/read the transceiver. The reason is that to drive the transmitter IR-LED you probably would like to use one of the TPU (Timer Pulse Unit) outputs in the MCU, which means the pin in question cannot be used for anything else. So you need to sacrify something, basically either one of the half-bridge drives or one of the (possible) RC-servo outputs. I couldn't decive which, so I put in a jumper header (J6). Now you can choose.

The same reasoning applies to reading in the IR-opto transistor, you would probably want to use the pulse counting facilites of the TPU to receive the data but do you want to sacrifice one of the quadrature inputs or one of the half bridges. As answer dependce on what you want to do I put in another jumper header (J1).

I have to confess, I'm a bit out on a limb here as I haven't got any experience with either IrDA or remote control IR application in general. Just thought that might interesting to experiment with those too.

Dozen or so Half Bridge Drivers

The twelve half bridges are made of two N-channel MOS-FET swithces (Fairchild NDS 9956A) each (Q5A/B-Q16A/B). These come conveniently in a small surface mount (SO-8)packages. The switches can handle 30 Volts at 3.7 Amps with Rds_on of meager 0.08 Ohms. This means about 1 Watt heat dissipation at max current! No need for heat sink! The half bridges are fed directly by the DC input of the board, so what ever riple voltage you have there is directly visible at the outputs of the drivers.

Note that the 30 V max rating means that your power supply must not put out more than that, so rectifying a 24 volt winding and filtering it is out with these FETs installed. If you need that you could try FDS3890 devices that have even more impressive specs, but I have no experience with them. They should handle 80 Volts at 4.7 Amps with Rds_on of only 0.044 Ohms, yes that's right, 44 mOhms.

With N-channel MOS-FETs driving the upper side switch is always a bit complicated, but the HIP4086 (IC12-IC15) chip makes it easy, it has built in charge pumps to generate the above-supply voltage levels necessary to drive the high side gate and also provide circuitry to prevent accidental (or intentional for that matter) shoot through, which is nice, especially in a software driven system.

In order to protect the MOS-FET and maybe even more importantly to protect the PCB board an over current protection circuit is implemented by (R112,R103,C36,R95,Q17,R99,C43,R104 and Q21 for motor drive 1, see schematics for the other motor drives). Quite a lot of component (even if cheap) for just protection, but given the nature of software I think they are worth every cent. But if you think you can lead a meaningfull life withtout them, just do not install them, just replace the current measurement resistors (R112,R113,R114,R115) with pieces of wire.

Over current triggers a flip-flop (IC9/A) which disables all of the half bridge drives, loose one motor, loose all. Software can then reset the flip-flop when approriate. A LED (D2) is provided to visually allert the debugger (that's me) of the problem. The over current signal is also connected to the a MCU interrupt line to alert the software, the flip-flop is there realy only to provide fast and software independent action. Again if you do not need the over current protection, you can leave some parts out (IC9,R7,R16,D2).

Note by, my main focus here has been the 3-phase stepper motors as you can read from the schematics as the over current protection is common to 3-half bridges. However, this should not make much difference if two phase motors are used.

So why not a ready made H-bridge? Well, good question. First of all the price of this solution (to me atleast) was atractive, the HIP4086 should be available for less than 5 EUROs and the FETs at less than 80 cents a package. The last time I checked I could not beat this with an integrated solution (i.e. one chip). Also most of the one-chip solutions require heat sinks as they are old designs with inferior MOS-FETs. And the flexibility of a semi discrete solution is that you can use better (or worse) FETs as they come available.

Driving Stepper Motors with the Half Bridges

First of all there is no sequencer, instead it is all done in software in the MCU using the TPU, which can put out 15 PWM (Pulse Width Modulation) signals.

Secondly, there is no current chopping. In motor, magnetic force is generated by current which in turn is generated by applying a voltage to the coils of the motor. Now if you drive the motor slowly (I'm talking about steppers here) you apply the nominal voltage to the coils and get the nominal current flowing as a result. At higher speed, however, the short pulses of the nominal voltage are not enough to rise the current in the coil, because of the coil inductance, to the required level. So the normal practice is to drive motors from considerably higher than nominal voltage and have a current measurement circuit that turns off the voltage as soon as the desired current is reached. This prevents the motor from burning with under slow speed conditions and still enable enough current and hence torque at high speeds.

But my plan is to do this open loop ie no current feed back. With PWM control of the half bridge drives it is possible to generate a virtual or average voltage (between ground and V_DC) at any of the outputs (J19/11&12&13,J19/8&9&19,J19/5&6&7,J19/1&2&3) at high (say 40 kHz) frequency. As the motor characteristics (coil inductance) is known it is then possible to predict, ie calculate, how large a voltage needs to be applied to the coils depending on the speed. It should even be possible to do so called micro stepping with this technique.

To drive stepper motors you connect each coil end to one half bridge. This also works for permanent magnet DC-motors. As stated you could control the direction and speed of up to six motors with this (or with a suitably stiff midle tapped power supply you could do it even for twice that many motors!)

A bit unconventional approach, sadly not (totally) original to me.

Quadrature Inputs, or DC-servos anyone?

A DC servo, for those not in the know, is a (usually) permanent magnet DC-motor, with a positional or speed sensors and control system that uses the sensor information to precisely control the speed or angular position of the motor. These have many same applications as stepper motors but have inherently more torque and speed available because, unlike stepper motors, they are operated in closed loop fashion and can never loose steps (they sometimes may lag a bit or over-shoot the target or oscillate about the target a bit, but then there is no such thing as free lunch).

A common way to get both (angular) positional information and speed information is to mount a pulse sensor on the motor saft (or gear box output saft, there are pro's and con's here). Basically they are opto fork and interrupter discs, you probably have two of them inside your mouse! Usually they have two opto sensors for each disc so placed that there is a so called 90 degree phase shift between them. In practice this means that if the spokes (or holes) in the interupter disc are 1 mm appart then the sensors are 0.5 mm appart. This means that a) the two signal never change simultaneously and b) if you look at the other signal at each rising (or falling) edge of the other signal it will tell you the direction of rotation! What a clever piece of hardware. Agilent makes a range of these, under the type code HEDS. They are not cheap but then they are of the high precision required for this application +500 pulses per revolution.

So anyway, the MCU has two quadrature inputs (ie two times two inputs) that are internally configurable to count these pulses so that both position and speed information can be recovered. All you need then is some software that implements the classical PID (Proportional, Integrating, Derivative) control system ie if the position is laging behind, increase speed (voltage) and vice versa. But that's worth a book all by itself.

These quadrature inputs are wired to the connectors (J8/7,11,15,3) in case you or I want to experiment with this topic.

Reference Inputs

For this purpose the board has inputs for four reference sensors, one for each motor (if you are using 3-phase steppers). My idea is to use the kind of reference sensors that change state in the midle of the most usefull travel of the movement. In this fashion the state of the signal directly tells you which way to drive the motor if and when you want to find the actual reference position and when the signal changes you know that this is it! There is no need for limit switches (unless you distrust the software ;-)

The inputs (J19-15,J19-24,J15-21,J19-18) are compatible with simple switches, HALL sensors (such as Allegro UGN 3123) or common anode/emitter type opto forks (such as Alep OJ 2020). Each input is provided with a pull up so that the sensor just needs to be of the closing contact type (open collector).

Power is also provided for each sensor. As opto sensor usually require a current limiting resistor these are incorporated into the power feed for the sensor (R119-R122). Moreover, as opto fork and to lesser extent HALL sensors as well are current hungry devices the power feed for the sensors is under software control (Q3,Q4,R37,R38). These transistors (same as the DA current mode sink transistors Q1 and Q2) are not surface mount devices so that they can better handle the current and the heat. The optos can take up to 50 mAmps each so the transistor need to handle up to 500 mAmps peak. Note that many common small signal switch transistor cannot handle that, especially not continuoysly, so take care when writing the software.

Again, for those that do not know, a HALL sensor can be thought as open collector transistor whose base current is induced with a magnet! In otherwords in presence of magnetism the transisto starts to conduct. In actuality internalay there is HALL-effect sensor and some smart sensing and amplificatio logic, but who cares. We nice in may application as it provides for non contact approximity sensing, sometimes inherent magnetism in parts is enough and no external magnets are needed, though usualy they are used. Mostly used as on/off ie logic level input device but there are varieties that can be used to also indicate the strength of the magnetic field.

Touch Probe Input

Speed Control Potentiometer Input

LCD -interface

Hex Keyboard Input

Ethernet Interface

Again, if you do not need the Ethernet you can save some money by not installing the Ethernet stuff (IC18,X2,C79,R116,R117,R124,R125,D7,R66,R69,R71,R70,R68,R67,T1,C63,C64,C77,C78). The Ethernet chip is probably the most expensive component on the board, in the order of 15 EUROs + the transformer 5 EUROs. Which is not much.

Connections (not by James Burke)

I/O Connectors J8 & J10

The pins are arrange into 'cluster' as illustrated below.

This makes for convenient wiring, just splice the ribbon cable to threads of two to four wires and connect them. No need to connect more that one device to one wire.

SPI-Expansion Connector J2

Motor Connector J19

Control Console Connector J23

Ethernet Connector J21

RS232 Connector J22

Power Connector J20

LCD-Display Connector J11

HEX Keypad Connector J14

Errata

J22 - Wrong Gender