Thoughts on the circuits you should publish

-December 19, 2017

When it comes to published articles, there are two fundamental categories that define what can and what should not be published: professional and consumer. The consumer categories once included magazines such as Popular Electronics and Electronics World, magazines you could find in a supermarket targeted at a predominantly hobbyist base. The professional magazines are best represented by the two top tier magazines, EDN and Electronic Design (ED).

In general, virtually any circuit was safe to publish in EDN and ED because of a professional audience that usually understood any apparent hazards. And most circuits that had hazards had them pretty well spelled out as any professional writer usually would anyway. But hobbyist magazines should have (I say should have because they often overlooked this requirement) adhered to a certain standard of avoiding even publishing dangerous circuits. During their latter years as our society became even more litigious, it is amazing what circuits continued to be published.

In ignoring dangerous circuits, there is also the issue of whether a circuit was really a good circuit or a piece of junk. For that matter some professional and consumer magazines suffered from a certain amount of poor quality circuits. One would think that the profession of publishing would somehow act as a gatekeeper ensuring that only quality circuits would be published. The problem, as in so many industries, is that many of the really good engineers are busy at the companies where circuits get designed.

Magazines, unfortunately, simply couldn’t get top tier engineers (a phenomenon that has only gotten worse as time has gone on). Normally that shouldn’t be a problem since many articles are submissions by the brilliant engineers working in industry. The problem is that many of the magazine engineers are gatekeepers rendering opinions on the technical merits of articles that are often beyond their technical ability. Although the modern internet world has further degraded to where articles are simply published with no review whatsoever.

Treading on dangerous ground

Let's look at one of those authors' most memorable electronic engineering disasters. It astonished me to see it published in the most popular consumer electronics publication in 1996. It’s almost unbelievable that this was ever published in our modern highly litigious society. And, regardless of the legal issues, publishing this circuit was genuinely irresponsible. But all of this gets ahead of some of the lesser but still important issues.

This circuit is for a Class-D switching audio power amplifier. This technology was not only thoroughly described (in some of the very same popular consumer electronics publications) as far back as 1975, but Sony manufactured the first commercially available switching audio power amp in that time frame. At the time, Sony had developed devices called V-FETs, named that for their vertical J-FET type structure. Being FETs they easily realized the high speeds necessary for a 250kHz switching rate commensurate with high quality audio (that dictates a sampling rate an order of magnitude greater than the desired upper frequency limit). The fundamental topology of a generalized Class-D amplifier resembles that of a sigma-delta modulator:

Figure 1
This represents the basic topology of a properly done Class-D switching audio power amplifier.

Note that this generalized topology does all the right things. The modulator is contained within a closed feedback loop ensuring faithful reproduction of the input signal. The output filter is outside the feedback loop which greatly simplifies stability issues and actually permits far greater bandwidth. This basic topology diagram omits many details. For instance, driving the gates on the power devices (including Sony’s original V-FETs) presents a number of challenges requiring circuits like cascaded followers.

One difficulty unique to Class-D switching designs is that they depend on the recirculation of unused energy in their output stage to achieve their efficiency. This creates a problem when driving a DC level into a load from a single output stage. To understand, we’ll utilize the basic circuit in Figure 2. The circuit shown is based on the assumption that we are trying to generate a negative output voltage. It also includes some items not exactly present in real world circuits, D3 and D4. These diodes are included to dramatize the fact that most power supplies are good current sources, and poor current sinks.

Figure 2
This circuit is used to show why a single-ended Class D must only be used for AC signals with no DC component.

The upper circuit in Figure 2 depicts the lower MOSFET Q2 on, and supplying the necessary current to the load to develop the negative output. Any intermediate output voltage will dictate a duty cycle less than 100% (or greater than 0%) so eventually Q2 will turn off and Q1 turn on, as shown in the lower schematic of the figure. Under these conditions the current continues to flow in the same direction, a consequence of the inductance in the output filter. The only path for this current is via the flyback diode D1 from Q2’s source, to Q2’s drain, and into the positive supply. The direction of this current is such that it will cause the positive supply to rise a little each cycle, until it is large enough to damage something.

This circuit cannot ever be exposed to DC inputs or allowed to develop an offset that can appear as a static DC output. Under such conditions, recycled energy from the output filter will have the effect of raising the power supply voltage on the rail opposite from the one supplying the load (e.g. a positive DC level at the load will have the effect of pushing up the negative power supply rail). Sony dealt with this by AC coupling the input and including a “supply rising detector” that shut the amplifier off. A more elegant way of dispensing with this problem is to set up switching amplifiers as full bridges where the energy gets an opportunity to be recycled.

Loading comments...

Write a Comment

To comment please Log In