Reference designs are circuit diagrams that appear on a datasheet, usually near the beginning. But where do they come from? This seemingly simple question has led to a number of interesting conversations. The answers vary widely among product families and target audiences. If you’ve ever wondered how they’re chosen and why, we have some answers for you.
If you wander into an engineer’s cubicle and ask where reference designs come from, the engineer will look at you strangely and probably say that they’re just basic circuits. Pressing further, we find that there is no industry standard or governing methodology for choosing these circuits. Engineers can only answer for the products they have worked with, so let’s try to bring those thoughts together.
Remember that the datasheet is a sales and marketing tool. It is designed to show you how great an IC is and how simple it is to use. The direct tradeoff of highlighting simplicity is providing all the details necessary to ensure successful operation. (If something is so easy to use, why would it need elaborate amounts of explanation?) This same tradeoff appears in the reference design.
On The Bench
Let’s start with a simple example, a voltage reference. This handy little chip will provide the bias voltage that your system needs. How accurate or noisy this voltage may be depends on supply bypassing. The datasheet may choose to include the supply bypass capacitors in the reference design. However, the proper types and values of these capacitors will depend on the frequencies in the system where it is operating.
If it’s a broadband system, multiple capacitor values in parallel would be best to overcome the resonant behavior of each component. (Resonance occurs between the capacitance and the parasitic series inductance of the layout and package.)
All precision ICs will need similar protection from bypass capacitors dependent on their system. Remember that it’s good practice to place bypass capacitors to protect every device from any spikes or unwanted noise on the power supplies. So, use bypass capacitors whether they appear in the reference design or not.
Layout is the largest common variable in system design. A simple circuit diagram, and therefore a reference design, doesn’t usually account for any parasitics generated by layout. This gives applications engineers headaches. Even when a sample layout is provided, customers must adapt the IC to fit in their system with their available footprint. Part of the challenge (and the fun) of system design is making the appropriate choices and tradeoffs for a successful system to come together.
Can a reference design represent all the situations where an IC will be used? The answer is yes, sometimes. It depends on how application-specific or general-purpose the IC is designed to be.
Continue on next page
It’s tougher to create the reference design for a general-purpose device than an application-specific one. By the simple fact that it is general purpose, there are many different uses for the part. Companies want to show to things with their reference design: how easy the part is to use and how to use it in your system. In this case, these desires are in conflict.
For example, op amps can be used for gain, filtering, buffering, and more. Which application should the engineer highlight in the datasheet? The engineer could choose to draft multiple reference designs. Typically, this is not optimal since the reference design also becomes the eval board circuit. It isn’t cost-effective to have multiple eval boards for every possible application of every different product.
Instead, most reference designs try to accommodate the top one or two applications. This means using switches to hook in different feedback circuits or connectors so each application can be prototyped. Of course, switched circuitry for multiple uses means that the reference design seems more complicated.
Figure 1 shows how simple it is to hook up a general-purpose op amp (as long as you consider adding five external passive components to be easy). It’s easiest to accept the presence of RF and RG, since they’re needed to complete the feedback that stabilizes the amplifier and sets the gain.
The next components to consider are the two bypass capacitors connected between the supply pins and ground. Many system designers mistakenly remove these devices to save bill of materials (BOM) cost or board space. This is rarely a good idea. Remember, bypass capacitors protect the op amp from spikes and noise on the power-supply lines. They are more effective when they are placed closer to the device.
The final component, RL, is the only one that might be optional. RL represents the load resistance. When the load is connected, this resistor isn’t needed. But if the load isn’t connected, then the resistor simulates actual operation so you can observe the op amp under similar conditions.
It is also important to note that this reference design serves multiple devices. The EL2170 is a single op amp. The EL2270 is a dual version of the same part, while the EL2470 is the quad version. This same reference circuit applies to all three versions of the part. Note, however, that the bypass capacitors are only needed per IC. That means that the single device has five passive components, the dual will have eight components, and the quad will have 14 components.
Figure 2 illustrates a second reference design. Here we find an op amp hooked up in a buffer configuration. (The output pin is connected to the inverting input.) Instead of a triangle, which is the symbol of an op amp, the engineer chose to represent the IC with a box. This might be more user-friendly for a system designer who “just needs an op amp.”
This is a higher-frequency part, two different capacitors are suggested on each supply pin. The 4.7 µF takes care of lower-frequency issues, while the 0.1-µF capacitor addresses higher-frequency signals on the power lines. As for loading, this time the load is modeled with a resistor and a capacitor in parallel. There’s no surprise here since any higher-frequency system will be more sensitive to the capacitance on any signal path node (like the input and output, in this case).
Put To The Test
Once customers believe a device might work in their design, it should be tested. Again, the reference circuit comes in handy. Eval boards commonly include switches to allow for the testing of basic parameters. This allows customers to verify the parameters critical to their system design.
Sometimes a circuit labeled “Test Circuit” is included in the datasheet. This clearly demonstrates the configuration used during the data-gathering phase of compiling the datasheet. Otherwise, pay close attention to the small print included above and within the design specification table to check for specific testing configurations.
One final question, possibly the most important: How do you know the reference design works? The ideal answer should be because the engineer who wrote the datasheet built and tested it. But people make mistakes, datasheets go through multiple revisions, and simplifications might limit the applications for which the reference design is reasonable. Nothing replaces your own tests in your own system. Get an eval board and hook it up. It’s quick and easy, and it will give you the peace of mind that your system will come together successfully.