Motor Control, Budgetting and SEPIC converters

I began today by thinking about motors and capacitors.  Currently, the robot has no motor decoupling capacitors.  Thinking about both EMI from commutation and the need for PWM to remain unfiltered, it's quite simple to figure out how big the capacitors are allowed to be.  I'm aware that most DC motors are typically used with '104' ceramic capacitors (100nF).

Motor Capacitors

Knowing the typical motor resistance from the datasheet, and assuming a PWM frequency around 2.5kHz (which is a reasonable number), the capacitance needs to be much less than 6.37uF (this is the 3dB breakpoint, where half the current will go through the capacitor).  100nF is plenty less than this, so it shouldn't affect the PWM signal much at all.

I've seen two typical arrangements of capacitors on motors - the most common is a single capacitor between the terminals.  This prevents commutation noise from leaving the immediate environment of the motor.  Another less common arrangement is to add two more capacitors to connect each terminal to the motor body.  Since I've not seen this so often, and it's difficult to solder to large metal surfaces without very big irons, I'll leave this option alone for now.

Protection Diodes

H bridges are typically protected by bridge arrangements to the power rails.  I've realised that ideally these should be positioned close to the motor output terminals on the PCB, and certainly should be physically between the driver IC and output terminals.  This positioning allows inductive kicks to be dissipated before they get anywhere near any chips that might not like them.  There should probably be a capacitor next to the bridges, too, to absorb the current to the supply rails and prevent circuit noise.

The current motor control board uses 1N5401 diodes - 3A standard recovery rectifiers.  They drop about 1V typically.

I've settled on the SB340 (or 540, depending what else I might build) for the new motor controller.  They've got an extremely fast response time, and only drop 0.5V typically.  They're Schottky diodes, which I'm led to believe are the best choice for this application.

Budgeting

I thought it sensible to sketch a block diagram of the system to see if I'd forgotten anything.  I hadn't, so I made a budgeting spreadsheet.  I hate working on paper.  I checked out PCB pricing - for a double-sided Mini-ITX board, it would cost about £15.  For through-hole plating (including the additional board area required), the cost rockets to £130.  This was really just out of interest - there's no call for it in this project.

I've considered using an op-amp circuit as a level converter for the time being, if there aren't any MAX232 ICs in labs - I don't intend to buy them in just for prototyping work.

SEPIC Converters

At some stage, a DC-DC converter will be required to allow the Mini-ITX board to be powered from a battery.  The last student that did this project designed a DC-DC converter, but never built it.  It uses the Single Ended Primary Inductor Converter (SEPIC) topology, which I'd never heard of.

I did some reading to find out basically how these work.  I still wouldn't be able to recite it, but I'm sure I could get it 'close enough'.  The previous design uses an IC that's only available in an MSOP-8 package.  They're absolutely tiny - and there's no real way to prototype the design easily.

Looking on Linear Technology's website, I compared some alternatives.  Their parameter search returned a short list of ICs.  RS only stock one of these, and it's not in stock at the moment.  The previously specified IC is only available from RS in packs of 5, though Farnell will sell them individually at a slightly higher price.

I did begin to wonder - how hard would it be to use a microcontroller or discrete components to build a DC-DC converter?  I'll leave you with that thought until tomorrow!