The A-Z steps for building a mixed-signal SOC, Part 4: System design

(Editor's note : we are continuing our “dialogues” between these two distinguished engineers with this special, multipart series on the product development process. This is based on a real-life example and hands-on reality, not speculation or an academic perspective. If you want to see the rest of these dialogues, or other articles by these authors, you'll find a linked list here .)

This is Chapter Four of an ongoing series:

•to read Chapter One , click here

•to read Chapter Two, click here

•to read Chapter Three, click here

Dave Ritter: Well I finally got you to spring for Thai food, Dr. Schmitz. As you know, I love that Punang Curry.

Tamara Schmitz: Yeah, and the Chicken Satay was excellent!I hope it’s fueled you for some design discussions.

Dave : Sure thing, where were we?

Dr. T : You were about to wax eloquent about design. Let’s start with the system design, the top level.

Dave : That’s exactly where we started with the video equalizer we’ve been using as an example, MegaQ, and it is where most system designs begin. To do a design effectively, we need a model of both the system and the environment it will be in. The part is an equalizer, so the performance is always relative to the basic cable characteristics.

Dr. T : Well this is the 21st century, there must be good models available.

Dave : That’s true for ‘normal’ applications, but not always. Besides, if you want to do something better than it has been done before, you need to look at things in new ways.

Dr. T : What kind of new ways?

Dave : Many people accept models blindly without investigating the simplifications used to create the model. Be sure to pay attention if a model was built for high power applications, or maybe just to cover digital transmission.

Dr. T : I guess it makes sense to simplify. Sure every designer would like complete 3D models extracted for every situation, but then circuits and systems could take so long to run that it is impractical to use them.

Dave : Good job. That’s exactly right. Good engineers simplify their models for efficiency. Other good engineers need to consider those simplifications when they use information from those models.

Dr. T : Let me make sure I’ve got this right. Designers want to use models, especially for a complicated system like this one. We want a model for our system and for every input or output that will connect to it. We do research to find out what has been done before and evaluate whether or not it is adequate for our system goals. If any model is not adequate, then we need to reengineer or improve it until it matches the precision needed for our system specifications as set in our design goals.

Dave : Well done, my Padawan.

Dr. T : Are the models in Spice good enough for most designs?

Dave : There are many levels of models available for the Spice simulator. Every designer must be intimately aware of the aspects of the transistor models he or she is using. Sometimes we also work with the fab to develop changes in the process to enhance or change those characteristics. Of course, that sort of change can be very expensive and take a very long time to create and verify that it is reliable.

Dr. T : So once you have good models, you tend to use them for a while?

Dave : Definitely. But you still don’t blindly trust models. We haven’t talked about layout yet, but once a circuit is planned, or “laid out,” for fabrication, we do an extraction to calculate the effect of the topologies and even the connections we’ve chosen.

Dr. T : Ok, we’ll get to that. Let’s get back to understanding system design. How did you come up with the actual equalizer circuitry?

Dave : Our apps people had been working on discrete solutions with several customers, so I had a body of work to start with. There is an example in Figure 1 . Every one of those used simple RC-boost circuits cascaded to provide enough equalization for a few thousand feet. What I needed was an organizing principle for an easily controlled version of those circuits.


Figure 1:Example of discrete video equalizer circuit

Dr. T : I thought most equalizers were collections of poles and zeros, and you just dialed around the poles and zeros until you got a reasonable equalization curve.

Dave : That’s true. We make manual equalizers like that. And that’s part of what makes video equalization tricky. Installers must tune the equalization circuits for optimum system performance. Marketing wanted this to be automatic, though, so I needed an approach with fewer knobs.

Dr. T : Other team members on this project have told me the ISL59605 is special because the installation is so simple. What makes it simple?

Dave : Most equalization systems use a block diagram like the one shown in Figure 2. There is pre-equalization circuitry or boost that is placed before the signal enters the cable and post-equalization circuitry that is installed at the receiver. That may sound simple, but think about most security systems. The cameras might be placed all around a building or a property. They might be extremely hard to reach, so installing pre-equalization circuitry can be costly and complicated.


Figure 2:A standard security-video set-up

with equalization components at the

transmit and receive side of the cable.

Dr. T : The MegaQ is only installed at the receiver, though, right?It looks like Figure 3. Most security systems have a central computer or center where all of those signals come together. It is far more convenient to have all of the equalizer circuitry at this point.


Figure 3:The security-video set-up with MegaQ equalization.

Dave : Bingo. So it is only on the receiver side and is fully automatic. It will even flip the input connections if the installer happens to accidentally switch the wires. And don’t forget that it will compensate for 5,000 feet of cable, far more than any chip offered by a competitor.

Dr. T : I’m always impressed by features that not only think ahead to make a customer’s life easier but allow new and exciting applications like the extension to almost a mile of cable. But let’s get back inside the design. I think the coolest part of MegaQ is the fact that it is fully automatic. How did you accomplish that?

Dave : Start with the discrete solutions that you are already familiar with. We agreed that they would need to be adjusted for each installation. Think of these adjustments like “knobs.”Each knob in the equalizer required a control loop to set it and a sense circuit to measure it. So it was important to keep it simple from a control standpoint.

Dr. T : I suppose the simplest would have just one knob: length. You would just turn it up or down to adjust for cable length. I suspect that the cable, like most cables, is going to act like a low-pass filter. Therefore, the sensor should sense the high frequency content and adjust the equalization until the high frequencies match the low frequency content. That way the circuit has corrected for the losses in the cable.

Dave : That’s pretty close. What we came up with was a length based equalizer. Each stage represented the boost required to equalize a given length. Let’s assume it was 1000 feet.

Dr. T : Why 1000?Why not 2000?

Dave : Another interesting question!It depends on the equalization circuit you are designing with. Look at the frequency response of your equalizer stage. The bandwidth of the signal sets the needed bandwidth for your stage. The gain available will help you calculate how much length can be compensated. We were working with 5MHz video, so it is reasonable to assume a compensation length per stage of approximately 1000 feet.

Dr. T : So to do more than 1000 feet you just added more stages.

Dave : Yep.

Dr. T : And that’s it?

Dave : Not quite. Several other issues need control, like DC offsets or variation in driver signals, loads or passive elements.

Dr. T : Do you test each block separately?

Dave : Yes, that’s why it is so important to have a functional model. It keeps the simulation time down to a reasonable level.

Dr. T : I’ll bet it was also helpful to go back and probe your prototype.

Dave : We would have never gotten it working if we hadn’t built the prototype and sorted out many issues before we started silicon design. I want to emphasize how invaluable it is to build a working prototype.

Dr. T : So you took all of that system work, and created a discrete, PCB level version of the architecture. Next time we need to talk about getting this design into real silicon.

About the authors

Dave Ritter was at Intersil when he co-wrote this series.

Tamara Schmitz grew up in the Midwest, finding her way west with an acceptance letter to Stanford University. After collecting three EE degrees (BS, MS, and PhD), she taught analog circuits and test-development engineering as an assistant professor at San Jose State University. With eight years of part-time experience in applications engineering, she joined industry full-time at Intersil Corporation as a principal applications engineer

Editor's note : If you liked this article and are interested in “analog” issues such as signal input/output (sensors and transducer, real-world I/O); interfacing (level shifting, drivers/receivers); the signal chain, signal processing (op amps, filters, ADCs and DACs); and signal integrity, then go to the Planet Analog home page here for the latest in design, technology, trends, products, and news. Also, sign up for our weekly Planet Analog Newsletterhere. You won’t be disappointed, and we won’t waste your time, that's a promise!

0 comments on “The A-Z steps for building a mixed-signal SOC, Part 4: System design

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.