Modules for embedded systems
Modules are a good way to save time and money on an embedded system. Here's a list of some modules; add your favorite modules to the list.
Usually, when I embark on a design, I like doing everything from scratch. Doing this gives one the most control over the direction the project is going. Also, it gives the core designer the maximum flexibility in solving price / performance and other design issues.
However, it behooves one to consider all options when selecting from the vast array of technologies to solve a specific problem. I remember very early in my consulting career a conversation with one of my engineering friends at one of my client's companies. The conversation went approximately like this:
Engineer: "Hey Ken, this DC-DC converter looks pretty cool. Plus, it is very small with a less than 20mm square form factor. It can deliver three voltages and can easily supply the power that is required by the project."
Ken Wada: "Why would you do this? Designing a buck mode converter with multiple outputs with the desired voltages is pretty easy. All you have to do is to select the proper core, rig up the SG3524 converter chip, order up some prototype magnetics from the local winding house, and you're pretty much good to go! Plus, designing one from scratch will give you the best price vs. performance too!"
Engineer: "Ummm, I don't know how to select the core. You may know all about saturation and permeability, but I just need a simple converter. Plus, this module is only going to cost me $0.50 in quantity"
Ken Wada: "$0.50 in quantity!? Dang! You cannot even purchase the switch mode controller chip for this price in quantity! Well, I suppose it is OK, since the product will never exceed 10K quantity per month no?"
So, I learned a quick lesson that nothing in the design world should be outright dismissed. There are ways to get a bunch of stuff out there to work well. I can tell you this much, I have now designed in a ton of modules in a lot of projects for my clients. And to this date, all of them have been very happy with the results and could care less that I have used modules in my design. Many of my clients mostly care about whether or not the design works. Also, they do care about price vs. performance.
There was a time when I thought the system on a chip (SOC) technology would make the module technology a distant thing of the past. In today's environment, the multi-chip module is still strong and shows no signs of abating. Now, I'm not going to down play the current SoC technology. Indeed, the current array of SoC solutions is wonderful--a topic for a future article here at Embedded.com.
As in all designs, there are many considerations and ways to solve a particular problem. As with any element, when I consider using a module to solve a particular design issue, I have a few things that are always on my checklist. In a nutshell, here's my checklist on module selection.
- First and foremost, exactly what am I trying to solve? Can it be done using with fewer parts? Can it be done using an architecture which costs less?
- Many times, when using a module, the designer is tacitly implying that he is 'outsourcing' part of the design. At this point it becomes necessary to gauge the trust in the source. This is important due to the non-2nd source nature of the module. Even though there are competing modules, there are usually no standards when it comes to the mechanical configuration. There are, of course, exceptions to this rule. Some very good examples of a standard mechanical configuration for a module are the secure digital (SD) flash module or standard formfactor pluggable (SFP) laser modules.
- Price versus expected volume is very important. Many times, we opt for 'outsourcing' part of the design because we know that we can get to a decent result much faster, this achieving a better time to market, at maybe the expense of adding to the unit cost. However, in today's environment, I have seen many modules that come in at a very reasonable price point even at +10K quantity!
- It is especially important to get as much information about the module as soon as possible. The ability to acquire this information, before a selection is made, can tell one a ton about product and technical support well down the road. All too many times, we tend to ignore this one--especially if it looks like we can simply purchase a solution that will fit our immediate needs.
What follows here are some examples of modules and module technologies I have used in the past. Also included are some examples of some module technologies I thought were interesting and may find a way into my design toolbox for future projects, or just stuff that I found that just looked cool. By no means is this an exhaustive list. What follows here is a very, and I mean very, small sampling of what I have used, and what is currently available.
This list is not an endorsement by me. What I'm showing are simply some examples of what I've used, or what some of my friends are using currently or have used in the past.
DC to DC converters
It was the humble DC-to-DC converter example above that changed my attitude toward doing everything from scratch. Later, I thought the SoC was going to make modules obsolete. However, even today, it's difficult to create a silicon-based inductor. Therefore, the DC-to-DC converter is still going strong and is here to stay. Here are a few examples of DC-to-DC converter modules and companies I have used in the past.