Generating a stable clock is far from easy…
Small, fast and stable
Generating a stable clock is far from easy: while current microcontrollers usually have some kind of embedded clock source, accuracy requires external components. Integrated oscillators provide a very interesting alternative, one of which we will look at here.
The ASEM-Series by ABRACON is a very reliable standard oscillator chip. In principle, the part does not need any kind of external circuitry. Simply look at the pinout, which presents itself as following…
…The datasheet diagram does not include decoupling capacitors.
ABRACON promises an accuracy of 5 PPM for the first year – quite a good value, which should satisfy most applications where extreme timing accuracy is not required.
ABRACON offers a whole family of parts, which can also be programmed to custom frequencies. Information on what you actually get can be obtained by looking at the numbers of the individual parts, which are set up according to the following scheme:
It takes but one look the part numbers to find out what you are dealing with…
The old way…
For comparison’s sake, let us look at traditional crystals such as 9C-8.000MEEJ-T. Its datasheet promises an accuracy of three PPM – this, of course, is quite a bit smaller.
The price is cheaper, too – this can be seen in the OEMsecrets comparison widget embedded. Keep in mind that there is no such thing as a free lunch. Using a crystal requires you to provide additional components – in the end, integrated crystals can be cost-efficient in small, medium also larger designs.
Fool around town…
Oscillators such as the above-mentioned ASEM-Series generate very steep waveforms. This is a problem if your design has to pass EMC certification or needs to work in an EMI-sensitive environment. Playing it safe is easy; place one or two small decoupling capacitors very close to the part. Furthermore, it is also a smart idea to transfer the clock using some kind of transmission line – you are not in the mood to play around with a network analyzer, at least try to stick to a design similar to the one shown in the sketch.
…The ground trace below the signal should minimize the loop area
Capacitor prices… all hands on deck, brace for impact!
MLCC capacitor troubles are real: since Murata decided to exit the market, prices go up. A similar trend can also be seen in flash memory.
Memory prices have long shown themselves to be subject to the “pigs cycle”. This is shown by looking two types of flash memory originally developed by Cypress, both of which are hit by supply problems…
An interesting aspect is that manufacturers focus on efficiency: older or “larger” products are eliminated in favor of smaller ones which can be produced with less resource outlay.
For example, take a look at the S25FL132K0XNFI013; a venerable SPI flash memory, which recently got hit with the hammer of discontinuation. At least, an application note available here provides some information on alternatives.
The MLCC capacitor situation, which we discussed recently – has not improved much. It takes but one look at the OEMSecrets toplist to see that MLCCs still are among the most-searched parts, with more than 25% of the toplist being populated by capacitors of one form or another.
While you might be tempted to claim that your systems are safe – you, after all, use MicroSD memory cards and electrolytic capacitors – be advised that this is untrue. MLCC capacitors and flash memory are used all over the place: one customer of yours truly and recently was hit 20% price increase on the process computer used to power his GPS tracker.
What to do?
Randomly stockpiling parts usually is an inefficient coping strategy. Sevelopers should instead keep prices of commonly used parts on the radar: if they suddenly start to rise, targeted stockpiling can be a good idea.
One way to find out about price trends is provided by TTI – the Berkshire-Hathaway owned company provide a list of of useful bits of info at https://goo.gl/7zh9Xs.
Better stock up if the exclamation mark shows up
As OEMsecrets mission is customer satisfaction, we always invest in finding new ways to help you. Our position in the market allows us to provide you stock level and price information of our own.
Sadly, as of this writing, we don’t know when the feature will be available – however, stay tuned to the OEMsecrets blog as we will gladly inform you the moment the feature rolls out!
RS485 sounds like just another serial protocol…
RS485 sounds like just another serial protocol which no one really uses anymore. Sadly, this cannot be further from the truth – read on to find out about its benefits and about an IC which makes implementing it really easy.
Differential signaling is helpful wherever noise can interfere with your signal: Jacob’s classic on Industrial Control Electronics can read as a collection of horrors of sorts. RS485 finds a home not only in the cabin of many aircraft, but – for example – also in some advanced model railroad systems.
Sadly, designing and RS485 transceiver by hand – while possible – definitely is not comfortable experience, especially for process computer jockeys who have a healthy respect of all things related to analog circuit design.
Fortunately, Analog Devices comes to the rescue with their ADM483E. Their IC contains both a sender and a receiver component, as can be seen by looking at one of the examples given in the datasheet:
…A decoupling capacitor is all which is needed for your new RS485 bus.
The part generates all voltages on the fly – as long as the supplied 5V is relatively stable, the rest is handled automatically.
Need for speed!
Another interesting aspect is its high-speed – while many “ready” transceiver modules lack speed, the parts can work reliably at up to 250kbps. Of course, the part also supports multidrop, allowing you to add up to 32 nodes to your RS485 network.
Should you find yourself working on an environment where certification is required, Analog also has you covered. Not only is the company – it recently brought Linear Technology – of excellent repute; the part itself is intended to work with TIA and EIA standards. Pricing, of course, is quite humane – look at our OEMsecrets pricing box to find out more.
For the capacitor, in principle, any partcan be used. As it must smooth out switching transitions, the use of a high quality decoupling part is recommended. These parts are not particularly expensive, and a single failed certification test can pay for literally thousands of them…
When adjusting voltage levels for process computers…
Resilient attenuator on the cheap!
When adjusting voltage levels for process computers, accuracy is not always important: in many cases, good isolation from the incoming high voltage spikes is of more significance.
A primitive design for an attenuator usually follows the lines shown in figure 1 – a small diode and a “stock” resistor is used, which works while the incoming spikes are not particularly large.
…This simple design works, but is not particularly resilient.
As EMV demands increase, cascaded zener diode designs get attention – the first references to them can be found inside Seifart’s Classic from the GDR. In principle, these designs follow the ideas outlined in figure 2.
The second diode limits the voltage across the first one…
While this design works, it is not particularly efficient from size and cost points of view. Furthermore, the voltage of Z1 is usually selected to be significantly larger than that on Z2, thereby leading to the well known antipattern of “wagging the reference”.
The brutality dividend
A very interesting alternative comes from Micro Commercial Components, and has seen quite a bit of use in process computer designs by yours truly. The SMBJ5xx8 family contains a set of Zener diodes which can dissipate toasty five watts of power – let us look at the SMBJ5342B-TP, which is priced as following:
With that settled, a circuit similar to the one shown in figure three can be designed. Here, the main loss takes place in the filter resistor, with the series resistor limiting the current flowing into any clamping diodes inside of the process computer.
…A high-power zener makes life a lot easier
In practice, such designs can survive insanely high voltages if the resistor is beefy enough – a system designed by yours truly handles literally dozens of volts while transferring down to 3V3. Of course, this problem can be solved in a variety of ways-for example, multiple smaller resistors can be connected into a series circuit. But this, sadly, is a topic for a rainy day…
The datasheet of this part family contains an explicit warning about fakes. This practice, which happens relatively rarely, indicates that this part is especially “hot” – make sure to purchase the diode from approved distributors, which you can easily find at OEMsecrets.
Controlling a large current with a small one without a physical connection…
Controlling a large current with a small one without a physical connection, traditionally leads to a knee-jerk reaction of relays. Optocouplers traditionally were limited to low currents.
For quite a bit of time, solid-state relays have been on the market. What once was a relatively large and expensive component is now available in comfortable six pin surface mount ICs – one example, which would have cost a lot a few years ago, is shown in the box below:
First of all, most solid-state relays, that includes the LCB710, are based on the principle of the optocoupler. This means that a small control current lights an LED, which then emits photons which trigger a set of photo transistors.
In the case of our part, the internal structure – as shown in the product data sheet – looks like this:
The two field effect transistors allow for AC and DC switching…
One strength of traditional relays was that they did not give much of a fuss about what signal they were handling: once the switch was closed it – essentially – carried whatever you chased along it. Solid-state relays were problematic in this regard, as many of them could conduct current only in one direction. Fortunately, the component at hand does not have this problem.
Furthermore, the turn-on resistance is quite acceptable, with an average range of like 0.5 ohms. Forgive yours truly for reprinting the datasheet here – the following things are very very cool:
…Printing standard deviations in a datasheet is a surefire way to endear your product to this reviewer
One key benefit of solid state relays is their quick turn on time: in principle, you can expect turn-off in less than 2 ms, while activation is done about 0.75 ms in the worst case – in both cases, expect the majority of the parts to do significantly better, which can be seen as an invitation to select components by hand.
Finally, also keep in mind that performance is quite satisfactory: current of up to 1A can be controlled with 2mA. The galvanic insulation barrier is quite beefy, being able withstand to 3750 V RMS.
To select, or not to select?
An unverified rumour circulating in Eastern Europe claims that one of the reasons why Solartron went tits-up was their expensive selection process. Selecting parts provides a way to drive additional revenue to your company: selling prepared replacement parts is fun.
However, be careful; you need to hold additional stock and have quite a bit of labour costs. If you can design around it, to so by all means…especially as parts also change as they age…
Old hands are not necessarily warm towards the idea of surface-mount in general; surface-mount inductors are considered the devil’s handiwork. On the other hand, integration and assembly cost pressures motivate even the most conservative designer to take a second look at SMD.
Playing around in your workshop is likely to make you paranoid; in addition to “inductor levitation” issues which laid flat many Palm m5xx organizers, it takes but a cursory look at the frontend of the venerable HP 34401A to see impossible-to-find Gowanda inductors going AWOL and causing the dreaded Error 612, 613, 615, 617, 618, 619, 621 combination.
Read a book…
Hollomon’s classic “Surface Mount Technology for PC Boards” contains a surprising, albeit logical statement:
“in principle, SMD components are more vibration robust than their THT compatriots due to their lower weight.”
This is easily explained – keep in mind that the force F is a function of mass multiplied with acceleration. Reducing weight due to missing leads, thus, leads to better attachment of the component to the board during vibration tests.
Before diving into specific interesting products, let us mention that this rule does NOT apply to connectors. With them, force is applied by the user and is not a function of weight – the larger area of the solder joint can be beneficial.
Inductors come in a large variety of types. One product, which we found to get a lot of attention recently, was a rather simple drop-in replacement for a classic coil. Simply deploy it to get rid of two or four through-hole joints, thereby reducing assembly costs.
This, of course, is not all you can do. Switching legacy switching ICs for high-frequency ones such as the ACT4065 we introduced recently allows you to reduce inductor size, thereby killing both weight and cost.
Just like in classic inductors, a large variety of construction styles exists. Wurth’s WE-PD family, for example, has a shielded core which reduces inductor emissions – this is helpful if the power supply is physically close to sensitive circuitry like a GPS receiver, but can not be shielded efficiently.
Bourns provides a very interesting application note which goes into more detail on the various inductor types – it is a well-recommended reading material for all those who are not 100% firm in inductor design and behavior.
Electrical engineering and electronics is both art and science. However, “tribe mentality” can be a valid guide – given that tons and tons of SMD coils are sold every month, things can not be all that bad. Of course, we look forward to your feedback.