Reference circuit program targets a growing challenge facing design engineers worldwide
Earlier in the month Analog Devices, Inc (ADI) revealed its plans to help engineers save weeks of research and design time by extending the company’s Circuits from the Lab reference circuit program.
The new program aims to address what the company claims is a growing challenge facing engineers worldwide who are being asked to design products more quickly, with fewer iterations and using an ever broadening range of technologies requiring analog, RF, power and mixed-signal expertise.
For reliable, repeatable circuit performance, the company has created reference circuits that have been carefully documented with test data, design considerations and trade-offs, and design guidelines.
eeNews Europe: What is the impetus behind the Circuits from the Lab concept?
Ledger: We do a lot of research with our customers and one of the things we have found is that 2011 is just a very difficult year to be a design engineer. There are several factors that are making this profession a lot less fun than it used to be and probably a lot more stressful at the same time. We see a lot of companies competing at a global scale which impacts the design cycles for everyone. We see a lot of engineers that used to be on teams with four or five people or eight or ten people who are now working on smaller teams because of the downsize. A couple of years ago they used to have a power guy. Now they don’t. Or they used to have an RF guy. Now they don’t. And the implication is that people have to do a lot more in less time.
People who have been professional engineers for their whole career are now being asked to move into new domains that they don’t have any experience in. There is less time to experiment and there is less time to just get a board and play around a little bit and most importantly there is less time to really learn the new concepts they are moving into.
This is particularly true when you are moving into the analog or mixed-signal world because:
a) These technologies are moving further and further to the edge of the board every year and basically becoming more complex and challenging as the performance requirements go up as a result of moving out.
b) This is an area where there is a little bit more ‘Black Magic’ in the design than maybe in other domains.
c) The last element is that there are a lot of kids coming out of school today don’t have this kind of background. A lot more people are focusing on digital so you don’t have a lot of expertise any more. A lot of the mentors are retiring now.
We see 2010 and 2011 has been a pretty rough time for a lot of design engineers particularly those working in or being asked to move into analog, mixed-signal and RF domains.
eeNews Europe: Do you think this a global problem?
Ledger: When we talk about design cycles shortening I think it is true worldwide. I think North America and Europe have been particularly impacted by shrinking design teams.
In China what we see is that there are larger teams there but we also have an entire missing generation of mentors there. What’s interesting about China is that people really having to teach themselves on the job. In a lot of European and the US companies when you come on board there are a handful of senior people there who are going to teach you the ropes and really look over your shoulder and help ensure that your design is going to work out of the chute. In China they have missed this generation of senior professionals so we see in China this voracious appetite for educational materials, examples and reference circuits.
Anecdotally we get a lot of feedback in North America and Europe that engineers coming out of school today don’t know the stuff as well as they did 20 or 30 years ago.
A lot more folks with degrees are focusing on digital because that’s where a majority of the board and software is these days. We have heard anecdotally from our customers that it is harder to hire college graduates who really know analog stuff because there are fewer of them out there. What’s more of a challenge in China at the moment is that over there are a lot of young engineers who are coming into companies that don’t really have a network of mentors to help take all of the theoretical knowledge and translate it into practical knowledge.
eeNews Europe: What do you see as the major pressures on design engineers?
Ledger: 1 – We have got the shrinking design window. Two years ago that might have been 12 months but this year it is: “Hey we have got to get a product out in eight months.” That is partially because we are playing on more of a global playing field and we have much harder competition coming from regions where we didn’t have much competition.
2 – Smaller design teams typically means people are shouldering more of the work and we have heard from customers that people who have been out in the industry for 20 years are being asked to move into new technology areas at a rate significantly higher today than what it was two years ago.
The implication is that everyone has less time. No one has time to go back to school or even just experiment. I was talking to one engineer about two months ago and he was saying: “I don’t have time to experiment any more. I used to be able to have fun and get these new products and play with them in the lab. My job now is much more like a system integrator and I really have to hope the suppliers are giving me good stuff and I have to leverage as much existing stuff as I can but at the same time I have to really guarantee that my board works on that first shot because an unexpected reason can delay by a couple of months and that is unacceptable.”
3 – I think this is particularly true in analog, mixed signal and RF which just tends to have a little bit more black magic in it than some other domains. The simulation technologies are not necessarily where they are with digital board design and there is no substitute for experience and unfortunately that’s a lot of what people don’t have these days.
4 – Just hearing from people trying to hire these days it is clear the level of analog expertise coming out of college these days means that you don’t find as many people really focusing on that as they used to.
5 – In Asia there are not as many mentors particularly in China. In Europe, North America and Japan there is a generation of great analog designers who are all retiring now.
eeNews Europe: Why are design engineers being more stretched?
Ledger: Particularly in North America and Europe we are seeing there is an involvement of engineers in more engineering disciplines. There is a transition from two years ago.
We hear people saying: “Now I am on a team of three people and we are moving to RF. We are routing wireless to our product.” We also hear things like: “Our power supply guy left and now I’m doing that too”.
eeNews Europe: What do you see as a key practical benefit of the Circuits from the Lab idea?
Ledger: One of the things that I think is particularly painful is when people have to do PCB re-spins. We looked at data coming from the teams in Analog Devices that do eval boards. We looked at how many times do we re-spin a board before it has the function right and also has the performance right. Most good engineers can nail the function on the first spin.
What is particularly hard with analog and mixed-signal is that I may have got my signal chain working but I am missing 5 or 8 dB of performance and I am not sure what is contributing to this lower SNR signal chain. So as the technology becomes more difficult and complex a certain number of re-spins is almost expected in a design flow. We have also found that the more experienced engineers have one fewer re-spins than a less experienced engineer just by virtue of the fact that they are more experienced.
A re-spin can cost you and by re-spin this means building a new PCB or looking at a PCB, populating it, debugging it and realizing there are problems and going back and figuring what the problem is then changing the layout and re-manufacturing that board. That can be four to six weeks of time. What can be extremely challenging is if the board works fine but you are having to re-spin a couple of times because you know you are having some noise issues on your board or some interference issues.
At ADI we looked at this situation and we looked at the challenges with re-spins given shorter design cycles and smaller teams and we asked ourselves what can we do to help this situation?
eeNews Europe: Why is ADI extending the Circuits from the Lab now?
Ledger: The challenges out there are that we see a lot of RF and analog mixed-signal isolation technologies that a lot of people out there do not understand very well but have to apply to their designs.
ADI has been working with these technologies for a long time. Our application engineers have been doing this kind of thing for 40 years and we have a lot of experience in these domains so this ‘Circuits from the Lab’ initiative is offering an approach to try and help bridge that gap. We want too provide some of the expertise we have to customers in an offering that hopefully makes it easy for them to get their analog and mixed-signal portion of their design working right and minimize the number of re-spins and hopefully nail the necessary function of performance on the first shot rather than two or three re-spins.
The initiative was launched a little over two years ago. When we launched it the idea was that we would basically build a signal chain or subsystem. We would string a handful parts together and optimize the circuit and we would test it in the lab and then write a document which was like a kind of app note that would say here’s the circuit, here’s the problem we are trying to solve and here are the components we have selected, here’s why we selected them, here’s how this works, here’s the test data, and now here’s some layout guidelines. It was basically like a mini-reference design in a document that we offered to customers. The uptake has been great and we got a lot of feedback from customers all over the world particularly in Asia where they loved the examples and test designs.
Over the last couple of years we have being doing a lot of research and what customers have been saying to us is that these documents are a big help by seeing the test data and understanding that.
Usually, what I find when I discover circuit examples out there is that I don’t even know if they have been built before. I have no idea of what kind of performance to expect out of them. I’ve no idea how to layout my board. There’s a lot of missing information so these reference ciircuits are already a big help.
A lot of our customers went on to say that it would be really helpful to actually see the design files for this hardware. For example, they would say: "I would love to see how you split the ground plane under this converter or how you managed this impedance. Where you placed the resistors in this impedance masking circuit ".
Some customers said to us we would love to see hardware. They said: "Your test data is great but we have some conditions that we have to exercise this circuit it against so we would love to get it in our lab and bang on it a little bit".
When we talked to folks doing the embedded software for a lot of them years ago they wrote what was referred to as ‘bare metal’ software meaning that in your embedded processor they would have a home-made scheduler that would manage all of the information they were processing. Whereas today more and more people are using operating systems, like LINUX or VX Works so for them getting a device driver has been proven can save them a lot of time.
eeNews Europe: How has the Circuits from the Lab concept developed in the past couple of years?
Ledger: We have really expanded this Circuits from the Lab offering from just a document to a lot of the design collateral that you really need to take this subsystem we have built and integrate it into your design with as little effort as possible.
Our intent is to help ensure that we are documenting the circuit, we’ve gotten this performance out of it and we are trying to make it as easy as possible for the customer to take that same design and replicate what we have found and get there hopefully with no re-spins.
We have the circuit documentation plus the test data we have been doing all along. We’ve added the design files, the schematics, the bill of materials, and layout files (PCB Gerber layout format). We are also doing LINUX device drivers when there is a digital component on the board and in a growing number of cases we are commercializing the hardware. We are making the hardware available. In some cases the hardware we have built this on comes off a demonstrator platform that we don’t sell broadly but we can at least provide the design files and photographs of the hardware that we used but in a growing number of cases we are actually selling these circuits as a low-cost eval board that you can get your hands on and put on your bench and run your own stimuli through it.
eeNews Europe: How many circuits are now offered under the Circuits from the Lab program?
Ledger: Today we have about 160 circuits up on our web site. About 20 to 30 have this expanded offering and all of the new circuits we have in the pipeline going forward will use the new format. We will be offering all the design files and device drivers where applicable. Most of the new ones we are doing will have hardware you can buy.
eeNews Europe: Will you be offering circuit design material on your older designs?
Ledger: We started this new standard in the autumn of 2010 and went back through our most popular circuits then retro-fitted the old ones and in some cases made the design files available. So while we are moving forwards we are sort of going backwards and pull those circuits forward as well. I am not sure if we are going to go through the entire collection and bring them up to the standard.
One of the ways we initially positioned this offering was that these were tested designs.
There are lots of great cook books out there. There are lots of great sites that aggregate circuit designs and give people lots of good starting points and good inspiration for new designs. But I think one of the big challenges is that you never really know who built that design or how well it works. You are never really sure what kind of performance you can expect out of that circuit. We are trying to address both of those issues as people look for designs that they can base their designs on.
We have spent time testing these circuits and exercising them to take a little bit of the unknown out of the engineering work. What we do is test these in a standard lab environment at room temperature. In some cases where a circuit is targeting a specific market segment where additional testing is required or of interest we will do that as well.
Because these are not full blown reference designs there is a lot of other things that will impact the performance around these circuits which is one of the reasons we don’t do full characterization on them.
With a lot of our newer circuits we are also proving information about the test set-up we use. We will outline how we went about getting the test data.
eeNews Europe: Why do you decribe your concepts as ‘reference circuits’?
Ledger: A question we get asked a lot of times is ‘"Where does this fit in the whole scheme of things? Are they app notes? Are they eval boards?"
The reason we chose the term ‘reference circuit’ is because they sort of fit between the component and a reference design. When you have a component you typically have documentation in the form of a data sheet. You typically have an eval board. You typically have design files for that eval board. So it looks kind of similar.
The difference is that usually an eval board is designed to showcase the performance of one part so the design decisions that went into that board are really aligned around showing that one part in that signal chain is running at its peak performance. The implication is that sometimes eval board circuits don’t make great application circuits. They tend to be a bit theoretical.
In reference circuits we really look at the whole design and we optimize all the components to meet a certain performance threshold but also to design a circuit that is reasonable and that meets the cost targets of a certain industry or meets other system level requirements. So while these circuits look kind of similar from an anatomical perspective they are actually pretty different. Largely because the design intent is fundamentally different from the beginning. We think of these as a subsystem design that is almost like a monolith. You can almost take the subsystem and drop it into your design. We have hopefully given you as much information as you need to get it really into your design and get it humming the way it should.
These fall a little bit short of a full reference design which typically has software and is closer to an end product. The difference there is that they are very hard to support as a semiconductor vendor’s reference design can be particularly involve software and require a lot of support.
The reference circuits are kind of a sweet spot between support load and complexity. We provide a fairly complex offering but it is right before the support starts to grow exponentially and because it is not a full blown design and it is more modular it fits into to a lot more situations.
We might have thousands of potential customers for a product and a reference circuit may resonate with hundreds of those customers but a reference design may only fit a handful of those customers because they get closer and closer to a full design.
eeNews Europe: How do the ‘reference circuits’ differ from simple Web-based design tools?
Ledger: A question we get asked from time to time is: "What about those Web-based tools out there?". For example, ADI has a tool called ADIsimPower which is a tool that you can go into and you can put in your rails and the amount of current you need then you can pull a lever and it will generate a reference design for you. We will give you a handful of options and you can adjust some dials if you want to optimize for price or performance and it will do simulation for you and take all your components and give you the schematics. It is basically a design that you can take out of that Web-based tool and drop in to your product.
These types of tools are becoming very popular and in many ways they are very complementary to reference circuits because they fundamentally solve two different kinds of problems.
The Web-based tools typically don’t generate a very optimal solution but it is good enough and it is usually for a problem that generally works but it is not going to be something that will differentiate your product from your rivals.
Power supplies are a great example. A lot of companies don’t have the luxury of having a power guy so using tools to generate a power supply that is going to work for them is just great.
Reference circuits tend to address more complex problems where you really have an engineer that is trying to balance a number of trade-offs or a lot of parameters to get an optimal design. The simulation and auto-generation tools that are out there today are not really up to standard to do these kind of designs.
On our site today reference circuits co-exist very peacefully with our filter wizard that will generate filter designs and our ADIsimPower tool that will auto-generate power supply designs.
One of the big differences is that with the reference circuit you are going to learn a lot about the process and how it works and you are going to get a little bit more insight into RF design. These Web-based tools don’t really teach you anything. They will give you a design that works but you are not necessarily going to know how it works or why it works the way it does. That’s a fundamental difference. There is a lot more education that happens with these reference circuits.
eeNews Europe: What application areas are you addressing with Circuits from the Lab?
Ledger: We are very focused with this offering. There are few domains that we are really hyper-focused on. We are really focusing on data conversion, RF and isolation as three main technology areas that we see customers are wrestling with.
We have a lot of DAC output circuits for instrumentation applications. We also have a lot of RF and ADC driver circuits for communications.
Isolation is a domain that we found that a lot of people need now but very few people understand. Those areas are rapidly growing. In the ADC and DAC areas those are continuing to grow slowly but we have a pretty guide foundation collection.
Examples of circuits that have proven pretty popular on the site include a kind of PLL modulator combination that becomes a full direct RF conversion transmitter. It’s a very broadband transmitter and with this circuit we provide an eval board, all the design files and bill of materials. The circuit solves a tricky problem and provides a broadly applicable solution for broadband transmitters which has found its way into a lots of different kinds of application.
We also have an example of one of our high-speed ADC circuits and this is a clever technique of using a VGA to expand the dynamic range of a converter. This is another example where we have a broad set of design files available and hardware.
There is also an example using a precision ADC and the circuit diagram looks a little bit simpler. In some cases we have two parts that work so well together to solve a problem that there isn’t necessarily quite as much Black Magic in the circuit but they are just really well-matched components that solve a common problem so we are providing that as a block as well.
So you will see a spectrum of solutions. You will see solutions that address the really hard challenges and you will see some solutions that may look a little simpler that provide more of an elegant solution to a common problem. We see an equal uptake and adoption rate for each type so both ends of the application spectrum are getting a lot of attention.
We also have an example of one of our USB isolator circuits. USB has become an ubiquitous standard, however, in industrial and healthcare applications the fact that it doesn’t have any isolation precludes the standard from its adoption there. This a circuit which will add isolation to the hub cable peripheral configuration and provide several kilovolts of isolation protection. For customers wanting to add a USB port to a medical device this circuit has been very popular.
Then we have an example of one of our application specific circuits which is a current loop for a process control application that is a 4 mA-to20 mA very low power implementation.
All these solutions are available for free you don’t even have to register on our site. We have a browser that allows you to browse by application, or circuit type, or what it has been optimized for, or the kind of design collateral that comes with it, so you can quickly drill down into the kind of circuits that you are interested in.
eeNews Europe: What do you see as the major design challenges out there facing the industry at the moment?
Ledger: Isolation is an unknown proposition for many engineers. RF is also an area that a lot more people are working in now and I think it is an area that a lot of people don’t know a lot about. And finally, just in general, using data conversion.
Just building the right driver output circuits for data converters particularly at the higher end if we are dealing with a 12-bit DAC for most engineers that is within their comfort zone. They are able to get that right but when you are talking about a 16-bit or 18-bit ADC then the techniques you need to use to get the performance out of those circuits tends to push the comfort zone of your average engineer. Some of these more modern data conversion problems are another domain that we find people need a lot of help with.
ADI aims to help engineers solve design challenges in analog, RF and mixed-signal systems
More about Circuits from the Lab at www.analog.com/circuits/about
Watch a one-minute video
Reference Circuits from the Circuits from the Lab program include:
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
