4QD-TEC Electronics Circuits Reference Archive
Fully complimentary switching
In part 7 we explained how a full bridge worked and went on to explain how two suitable half bridge could be connected 'back-to-back' to make a full bridge. In that exercise, the two controllers were operating independently and, in normal operation, only one half bridge was chopping. Only when the permanently on half bridge went into current limiting would both half bridges be chopping.
But what happens if we deliberately switch all four bridge legs simultaneously? This page examines this as a possibility. This is a possibility which was suggested to me - the original 'idea man' is unfortunately anonymous.
In the drawing of the bridge below, the scheme is that legs A and D will be on while B and C are off and vice versa. So, if the duty cycle is 50%, then the voltages on each end of the motor will be equal, at exactly half of the supply voltage and no motor current will flow.
Now consider the situation when leg A (and D) is on for more of the cycle that is leg B (and C). Clearly the voltage on the left hand end of the motor will be more positive than that on the right and net current will flow as shown by the red arrow. Since we chose a chopping frequency high enough that the motor current is, essentially, constant throughout the cycle, when leg A is off, flywheel current will flow in the path shown by the green arrow.
When the motor back emf rises above the controllers average voltage output, the motor current will of course reverse as it starts generating and the currents shown above will reverse.
We have seen that, at 50% duty cycle - there is no net output from the controller. When A+D are on more than 50%, motor will be driven forwards. When B+C are on more than 50% - the motor will be driven in reverse. So in this mode, there really is no difference whatsoever in the chopping action between forward and reverse or between acceleration and braking.
- Drive waveforms
Only two pwm waveforms are required to drive the bridge - and one is an invert of the other. This of course assumes the timing (to prevent transconduction, to avoid A and C being turned on simultaneously) of the bridge switching is done in the drive and level shifting circuitry.
- Current limit
At all times, i.e. in all four quadrants of operation, the motor current is flowing either in leg C or in leg D - and one of these must be positive. This should enable simpler current detection circuitry to be used.
- Main Capacitor
The main capacitors are being worked harder. In other full bridge circuits, flywheel current recirculates only locally, through motor and MOSFETs. Here, it recirculates through the battery lines, thence trough the main decoupling capacitors, which will therefore need to be more numerous.
We have seen in part 6 that commutation is a high dissipation point in the MOSFETs. In the system described above, both bridge halves are switching at all times (other than at full speed of course), so both halves of the bridge will have to dissipate the commutation losses. This means that overall efficiency will be less than other switching systems - i.e. it will get a bit hotter.
However - to mitigate this a little, in the usual system, one hiside bridge leg is turned fully on and is conducting current continuously. Its junction temperature will be high because of I2R heating and this will increase the on resistance of the silicon. If the conducting MOSFET is off for part of the cycle, it will be able to cool when it is not conducting so its average junction temperature may not be as high. This will offset, to some extent, the commutation heating.
To actually calculate the difference would involve a heat integration over the cycle and one fairly involved graphing of dissipation. The calculation involves junction thermal mass, thermal resistance, ambient cooling and, of course, switching times.
Overall Pros and Cons
It should be clear then the benefits and advantages are very much down to the implementation: down to the drive circuitry and how well this handles all the normal design compromises.
These articles have, we hope, given you a good idea of the principles involved. However they barely scratch the surface of what can be done using software. Electronics has evolved much since the articles were first written!
With a microcontroller controlling a full bridge, circuitry can be much simplified and lots more monitoring can be done so that any one bridge leg can be switched on and off independently. This can be used to test for external faults and to react accordingly in time, hopefully, to avoid damage to the controller. Faults can be logged for later diagnosis and clever tricks can be done so that the machine is better protected: motor current can be profiled so overheating is avoided and control can be retained so the machine breaks safely even if the battery falls off.
We would be prepared to discuss such things with a reader who has sufficient interest (and ability) that a commercial design could be realised.
Other relevant pages
© 2001-2011 4QD-TEC
First published: 7th September, 2001.