Sunday, December 31, 2017

Why Do I Use So many Bypass Capacitors?

When you look at the design requirements of most ICs they commonly require a .1uF bypass capacitor, with one larger bypass capacitor (on the order of a few uFs) for every 3 or 5 ICs. I use a .1uF and a 2.2uF capacitor for every IC. That seems a bit extravagant both in capacitor costs and board area. So why do I do it?

Let us think about it. An IC draws a pulse of current due to a change of state. This pulse causes a change in voltage on the .1uF bypass capacitor. By having a 20 times larger capacitor right next to the .1uF the change in voltage will be 20 times smaller (actually a bit less due to the frequency response of the larger capacitor). This effectively reduces power supply noise for the IC by a factor of 20. It also decreases the peak of the current pulse needed to recharge the capacitor (the recharge will also take longer). Lowering the peak current also reduces emitted EMI.

If I was going into volume production and had a good EMI lab I could reduce the number of 2.2uF capacitors until it started making a measurable difference. Since I don't have an EMI lab in my shop - I just keep those extra capacitors and come in with a lower noise design. This is especially necessary for high frequency noise since most chips are not very good at reducing that kind of noise.

So why not throw out the .1uF capacitors and just use the 2.2uF caps? Because capacitors are to some extent frequency sensitive. The .1uF caps will respond to fast transients that will not (immediately) affect the 2.2uF capacitors.

Engineering is the art of making what you want from what you can get at a profit.

Monday, December 25, 2017

Crystal Radios

Ben Tongue of Blonder-Tongue Laboratories has written extensively on crystal radios and crystal detectors. You can find the articles here. Ben is deceased. I have collected the articles and turned them into pdfs along with some other crystal radio stuff. As soon as I up load them I will post a link to the package.

The zipped package of Crystal Radio pdfs can be downloaded here.

What got me started on this was Diode Detectors for RF Measurement. And what got me into that was the design of an SWR Meter. I will publish that design when the board layout is complete.

Engineering is the art of making what you want from what you can get at a profit.

Sunday, December 17, 2017

Radio Receiver Update

I have the very initial schematic of the radio receiver done. You can see the first page here.

Click on the image for a larger view)

And the rest of the schematic (in pdf) is here.

Comments and suggestions welcome. I will be doing a clean up (assigning part numbers and other stuff) over the next few days and will then begin layout.

Engineering is the art of making what you want from what you can get at a profit.

Friday, December 15, 2017

A Little Confection

While I'm working on a few projects, I thought it would be good to get something done. So I did. A little logic level tester. This one is a little different. It tests for voltages. Less than 0.5 volts. More than 2.0 volts. And more than 4.0 volts. And yes there is a light for each. Good for looking at mixed voltage logic. And there is also a light for greater than 0.5 volts but less than 2.0 volts. So you get a light for ambiguous (in some situations) voltages. The schematic looks like this:

Click on the image for a larger view.

The circuit will run on 3.0 volts (nominal) but the lights will be a little dim. The voltage response will be the same as it is on 5.0 volt power. The magic of voltage references and comparators.

The complete package including schematics, parts layout, and parts list (in pdf) can be found here.

Engineering is the art of making what you want from what you can get at a profit.

Tuesday, December 12, 2017

A Radio Receiver

I have a long time interest in amateur radio. This got me noodling around the 'net and I found the most interesting circuit. A SSB receiver. And the most interesting thing about the circuit is the frequency control device. The CS2000. It generates a frequency from 6 to 75MHz with just a frequency input and a digital word. Mouser has them in stock for $8.93 each in lots of one. Not a bad price for such a wide ranging frequency source.

There are of course other ways of doing things. this guy likes the AD9854. It looks good. And costs over $50 for one. A little out of my price range this week.

Naturally there were some circuit modifications to be made. Like directly driving 5 volt logic from a 3.3 volt logic source. It works - usually - but is not guaranteed. In those places (where necessary - some 5V logic is designed to respond to TTL levels. A relic from another era. But very useful in this one.) I added a TTL level compliant buffer. S0 .8V for a logic low and 2.4V for a logic high are guaranteed.

Also the detector circuit requires matched capacitors. Four for the detectors and eight for the phase shift (all pass) filters. So I have designed a capacitor matcher. With it you can read out a .1 uF capacitor to parts per ten million - or better. The accuracy is much less. But for matching lots of resolution is good. It also gives an interesting view of the world. The changes that can be detected when bodies move.

Design of the all pass filter was simple using the (free) design software found here. You need to install it on your computer. I have installed it on mine.

I'm working on schematics, parts lists, and board layouts for all of this. I will post them here. When they are done. About a week or two.

Engineering is the art of making what you want from what you can get at a profit.

Thursday, November 2, 2017

Hand Soldering Surface Mount Parts

The first and most important part of making soldering surface mount parts easy is to design the PCB for hand soldering. The trick is to double the LENGTH of the pads. This gives the tip of your iron room to maneuver around the other parts of the board. It also makes it easy to heat the pad which is important for good solder flow. The second trick is to use a eutectic tin-lead solder (63% tin - 37% lead - but 60-40 will do) so the soldering temperature is as low as possible. This gives you more margin when it comes to temperature rise. Too much temperature rise can damage parts. Lead free solders are not as forgiving.

My favorite iron is made up of a Weller 7400 handle with a 37UG heater and a PL-111 tip. I use a standard light dimmer mounted in a 2 outlet plus switch outlet box to control the temperature. I adjust the heat so it is just a little more than is needed to melt the solder. Surface mount parts don't need a lot of heat. So why do I have such a powerful heater? Because occasionally I want to solder wires to my boards and that can take a lot more heat. I use Weller 8001 anti-sieze to mount the tips to the heater. It makes changing tips easier.

And of course you will need solder. I like 63/37 (Sn/Pb) rosin core solder. You will need a flux with an applicator that will make temporary tacking surface mount parts easy. Chip Quik SMD291NL does the job. A rosin flux pen is also handy for coating pads on a board so they solder easier and do not become corroded. A Kester #186 pen does that job nicely.

You will also need help for times when you make mistakes. The Weller 7805 is a good solder sucker. Solder wick in several sizes (especially .025") is also very useful.

You may also need a magnifier. I use a magnifier head strap from Harbor Freight along with 3.5 diopter magnifying glasses you can pick up at any drugstore. For a bench light and general magnifier I use an E78751 magnifying light - which appears to be obsolete. This magnifying light from Harbor Freight may be a substitute. I haven't tried it.

Finally you will need a pair of tweezers for placing small parts. I have used all kinds including those for plucking eyebrows. I have been using these industrial tweezers for the last few years.

That gets you a bench set up. Not counting a conductive work space for static protection. I use black anti-static bags taped together and grounded. It is low cost and works well.

That covers your bench set up.

This video is pretty good. However I do things a little differently. I use the SMD291 flux under the chip to hold the chip in place. With this flux you can move the chip so that the pins line up with the pads. Once the pins and pads are lined up, tack the corners. And then all the rest of the pins. For added protection during soldering I like to mount the high value bypass capacitors first. This helps absorb any residual static charges while still allowing easy soldering.

Update: 24 November 2017 2156z

It occurred to me that starting novices with fine pitch surface mount parts might not be the best idea. So I designed a very simple board for testing the resistance ranges of digital meters. A board with seven precision resistors (0.1%). The first two resistors (12 ohms and 150 ohms) create a 11.11111111 ohm resistor (not counting resistor tolerances). The rest are normal decade values. which means you don't need to look for "9.000" values (which you can not get - usually) for decade dividers. The "wires" on the board add about .001 ohm to the 11.1111 ohm resistor. Which is not significant at the one part per thousand (3 1/2 digit meter) level. Especially as most of those meters can not zero out the resistance of the probe wires.

I use 805 size resistors so the board will be a little easier to solder than the 603s (inch) I normally use. Just the thing for a beginner. And you will have something you can uses when the project is done.

You can find the documentation for the project at Resistor Ladder. The cost for the 7 precision resistors is a little over a dollar total. About 15 cents each in single quantities. Quite a reduction from the days when such resistors cost around ten dollars a piece.

Update: 26 November 2017 0447z

A resistance reference is good. But for real meter testing you need voltage and current. So I designed a voltage and current source. To keep costs down there is just one voltage and one current. 1.200 volts (+/- 0.1%) and 1.200 mA (+/- 0.2%). The documentation can be found at Voltage and Current Reference 25Nov2017 -

Engineering is the art of making what you want from what you can get at a profit.

Monday, January 2, 2017

WWVB Simulator

I'm in the process of designing a WWVB Frequency Receiver and Clock that can receive either the amplitude modulated time signals or the phase modulated time signals. What I needed to go ahead with this project is a modulator that can make the phase and amplitude modulation. I racked my brain for days until I came up with a schematic of this simple concept. It is a combined phase and amplitude modulator that uses a quadrature signal generator. It sounds complicated. It is really quire simple. Have a look at the schematic. The schematic shows just the bare basics. There is no amplification or filtering or control processor. Those will be added later as I develop the design. But I just had to share my idea because it is so cute. Note that for phase reversal modulation both the I and Q channels must be switched simultaneously. If you only switch one you get a 90 degree shift. I may use that to see how the receiver behaves, but the receiver is not designed for quadrature phase modulation. Just Binary Phase Shift Keying (BPSK).

As I further develop the concept I will post more details here. Eventually I will be designing and building a board. The control microprocessor will be run off the master clock on the board so switching will be roughly coherent. With some tuned circuits to eliminate harmonics and such the exact phasing (such as WWVB does) is probably not strictly necessary. At least for testing purposes.