datasheets.com EBN.com EDN.com EETimes.com Embedded.com PlanetAnalog.com TechOnline.com  
Events
UBM Tech
UBM Tech

Design Article

Analog: back to the future, Part 3

Steve Taranovich

12/2/2012 8:56 PM EST

The birth and evolution of the switching voltage regulator—The SG1524
The birth and evolution of the switching voltage regulator—The SG1524

 “All I did was being an enabler to help others apply switching technology to power supplies easier”, humbly spoken by the man who is called the “Father” of the switching regulator controller. Bob Mammano’s biggest battle was to convince the marketing team at Silicon General (Founded by Mammano) that it would be a good idea for him to spend time on this project.

The SG1524 came out of this effort and was unique in the industry among 1975 power solutions because it put all of the blocks a switching power supply needed together in one chip. As a matter of fact, it can be arguably classified as the first mixed-signal IC design.


SG1524 block diagram
In 1975 there were a series of technical challenges in fitting all the pulse-width modulation (PWM) control circuitry into a single IC. At that time IC’s were either analog or digital and both had to be integrated into this design. To compound the problem, analog circuits were designed on a bipolar process to get the voltage and current needs of a power management solution. How would one build a digital circuit on the same process?

Mammano went back to an even older technology for digital logic circuitry, a basic bipolar technology. This would also now work for the analog portion of the circuit.  This design was unique in that it was also the first design to implement protection as well as control on an IC and became the mainstay of all future power controllers. Another innovative idea was to limit the output current on a pulse-by-pulse basis, a key requirement for switching regulators.

Mammano had the unique experience of founding the Unitrode IC division in 1981 and his team developed the first PWM controllers with current-mode control, the UC1843 and UC1846. Later at Texas Instruments he was on the first digital power committee, when digital power was just a “name”.

Modern digital power architectures have PWM’s integrated into the microcontroller with time granularities of 150 ps, unheard of and unachievable in the past.  The Texas Instruments Piccolo processor is a good example of this capability.

Texas Instruments Piccolo microcontroller family has high resolution ePWM capabilities

Higher integration technology also now enables integration of MOSFET switches right into the PWM IC.

Mammano says, “Power supply solutions are determined by the needs of the system it’s going to power. The challenge is to optimize the supply and make it more cost-effective, smaller and more efficient. The problem has always been that a different system needs a different power supply which is optimized to that particular system needs.”

In the past any technique that could give an analog supply more versatility increased the price and size, but with the advent of digital power, software now is the enabler for power supply versatility, at equal or lower ultimate system cost.

The DSP or microcontroller now gives better control management, sequencing, monitoring and margining for adaptive capability that is most evident in being capable of maximizing efficiency in near real-time as loads quickly change.




mtripoli

12/4/2012 2:13 PM EST

Seeing the photo of the "Evolution of..." brings back memories. When I was a "kid" I worked at Analogic (briefly) on hybrid D/A converters (if I remember correctly) "adjusting" resistors with a decade sub box. I did that for about a week before, I, um, "left"...

Sign in to Reply



Brian Lowe

12/4/2012 10:37 PM EST

I think it's worth a mention of the pioneering work of Harris Semiconductor in dielectrically isolated bipolar in the 1970s. This was a high performance complimentary bipolar plus thin film resistor process with Cold War era desirable radiation resistance. They produced the highest performing op amps and sample-holds available, also the most expensive. They also had one of the early silicon gate CMOS processes in the late 1970s that could work at a (then) breath-taking clock frequency of 30MHz. Our custom products group also used it to develop some of the very first, and much slower, monolithic pacemaker chips.

Also some important history lies in the evolution of IC testing, such as the early Fairchild and Teradyne systems in which they had to design their own computers and write their own operating systems. In the 1977 Sentry II you had to multiplex two pins together to get a 20MHz digital signal to a device. I once saw a co-worker debug a logic error found by the CPU diagnostics. He ultimately traced the problem to, I believe, a quad XOR gate. He pulled out the PCB, replaced the IC package and the diagnostics then passed.

At ITT Semiconductor in 1977 I heard a story about a large run of wafers that had been tested and the bad dice inked. While the wafers were in storage awaiting scribe and packaging, ants decided they liked whatever was in the ink and ate it all off the wafers, which had to be retested; a new brand of ink was also procured.

Sign in to Reply



steve.taranovich

12/8/2012 9:27 PM EST

Excellent choice Brian---I remember when I was with Burr-Brown and then TI---our OPA627 and other amazing performing op amps were on the Harris dielectrically isolated bipolar process---great process and made some awesome op amps!

The only reason I did not mention Harris is that I could not locate someone from the 60's/70's who was still around to tell the story like the companies I put in these articles.

Sign in to Reply



Please sign in to post comment

Navigate to related information

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)