dgrieg, to produce good circuits using text we used to use DOS EDIT. the closest one can get now os MS notpad, which works fairly well. But I produced a lot of cable diagrams using edit and "wordstar" non-document. Back in the olden days. Those were the days my friend.
WilliamK, the value of Bipolar over MOSFET's or vice versa is a very convoluted relationship between voltage, current and switching speed as you well know. Bipolars have a VCEsat figure that can be markedly lower than the RDSon of a FET when they're hi voltage devices. Then there's the to SOA foldback behaviour of bipolars compared to the straight line SOA with MOSFET's which is really handy if there's a lot of load in the inductance so having the bipolar/MOSFET split per your description may make real good sense for some applications. I think a heatsink (or rather lack of need for one) could very easily make up for the added cost of drive circuitry.
That said, SOA would be the one thing that I would be keeping a real close eye on.
Interesting side note, the authour introduced us to the black box concept by talking about Opamps, I would have thought that they too should never be black-boxed until after the details are nutted out.
You can do anything with software. However, it does not run with out power. And when you violate the laws of physics there will be smoke and perhaps fire. We had a software engineer who did not bother to read the rise time rateing advised. As a ressult the circuit bounced when good deivices were used. The orginal fix to the fast software was slow devices with little gain. The fix was do change the software.
The H-bridge as shown will be about 90% efficient because half of the MOSFETs are hard-switched. This is a lot of electronics loss in a multi-kilowatt-level motor. Soft switching circuits can be added to the basic bridge to bring the efficiency to over 95% for AC or DC motor drives, maybe to 98%. Soft switching has the additional benefit of being less stressful on the MOSFETs and the motor windings, so it increases reliability.
A third H bridge configuration - 4 N channel stacked DS-DS-DS-DS, with or without series Schottky's and parallel HV super-duper-fast diodes. Why? - the DC rail voltage exceeds the FET DS voltage limits.
I have seen an alternative arrangement for H-bridges that used to mosfets as the upper transistors and two NPN devices at the lower side. The benefit that was claimed was that the upper sidedevices had the lower voltage drop while conducting while the NPN devices worked faster for the PWM control function. And the shunting diodes could be external to the transistors so that their conduction loss during switching would not affect the transistor dissipation. Aside from making the drive circuit a bit more complex it certainly sounded like a worthwhile option.
Is there a downside that I have missed? Aside from not fitting into a single IC package?
Actually there are two H-bridge configurations, this is one of them. The other is made of all N-channel devices, the idea is that Rdson is always higher for P-channel devices, so by using all N-channel with a heavy load you will tend to lower losses. This is the more common configuration when an IC with a high level of integration is used to drive the power devices. Since such a configuration requires a drive voltage that is 5 to 10 volts higher than the supply voltage, the IC must include a charge pump, and it is by providing same that the IC "earns its keep" (merits the possibly higher cost in the application circuit). Such designs don't generally require additional analysis however, since presumably the design of the switchover time provided by the IC "guarantees" that cross-conduction in the bridge cannot occur.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.