Electronic Construction -- Tips, Tricks, Gens,
Traps, and Snares
(what's a "Gen" ?)
This page has general electronic construction tips (applicable to ciruits
build with solderless breadboards, wire-wrap boards, and printed wiring boards).
Do we need another page for tips specifically for PWB design ? [FIXME: seperate
"design-time" tips from "assembly-time" tips]
power supply tips
standard signal tips: series resistor, etc.
high-power and/or high-voltage tips
Signal protection. "Itâ€™s a golden rule in electronics, any input voltage applied to a device has to be within that deviceâ€™s power rails. It it is not then you risk damage being done. In most ICs there is what is know as internal protection diodes that will provide some degree of protection but they are insufficient to provide all the protection you need and sometimes they need a bit of help. In order to help limit the current for the built in protection you should put a low value series resistor. Also if the signal input will stand it put a small capacitor. This helps to absorb sudden transients, as found in static discharge, but it can degrade the signal especially if it is fast. The faster your signal the smaller capacitor it can stand before degrading the signal and the less spike protection you have."
Also, limiting diodes or chips like the BAV99 can help protect better.+
codymillerTakeThisOuT at @email@example.com
PIC-specific hardware gotchas
PIC-specific hardware gotchas
power supply tips
Protect against incorrect (reversed) insertion of batterys
Appling Vcc to the GND and Gnd to the Vcc will fry just about anything. Use
of a fuse (or one of the PTC thermal auto-resetting devices from Bourns or
Littlefuse) and a diode is a minimum. You can take an enhancement power MOSFET
and put in the negative return line from your circuit. The N-Chanel Enhancement
MOSFET is used "upside down" compared to other circuits. That is to say,
the Drain is connected to -ve, Source through the load to +ve, and gate to
+ve. The gate is driven through a 1 Meg resistor from the positive supply
terminal. A correct power connection drives the MOSFET fully ON, and everything
works. Reversing the supply connections turns the MOSFET OFF, and no current
flows and you're protected. Correct the power
connections, and you're back to normal. Select a MOSFET with a low "R(on)"
resistance, and there will be very little voltage drop.The better way to
go about this is to protect reversal based on the mechanical properties of
the battery. mostly this is done by a positive
battery terminal that is set back in the holder so only the positive battery
pole that sticks out can reach it.
Connect all power and ground pins
Many chips have several power supply and ground pins
(sometimes labeled Vdd and Vcc). Make sure each and every one of them are
connected appropriately (typically Vcc to ground, Vdd to +3V or in some cases
There are a long list of strange things that happen when you skip this step.
The PICC74A may not work properly with
it's A/D Vref set to reference the voltage on AN3 instead of chip Vdd if
pins 31 and 11 are left floating.
The Scenix SX52 may not start running if all
of its supply pins are not connected.
Q: "What does "dd" stand for in "Vdd" ?" A:
Make good power supply and ground connections.
A skinny, daisy-chained wire-wrap connection from chip to chip is probably
not going to be enough. The more robust you can make power supply and ground
leads, the better
Despike your power supply to each chip.
This means putting a .01 to .1 uFd ceramic capacitor from the +5 volt supply
to ground right at the chip power line. Spikes occur when there are sudden
current changes far from the power supply, and it's amazing how much trouble
this can cause. If there are two or more Vdd pins, you need to put a bypass
cap on each and every Vdd pin. It is critical to
de-couple power supply lines. You want the
caps to absorb well at the third harmonic of the clock. 0.1uF does well at
3 MHz, 0.01 at 30. 0.001 at 300. It's a broad response, so dont think that
there's one specific value. However, if you use 0.1uF on a 20 MHz part, you
won't get the supression that you could if you used 0.047uF. (In theory,
an ideal 0.1 uF capacitor should always give better suppression than a 0.01
uF capacitor. But real capacitors have parasitic lead inductance and other
Use the right kind of capacitors
Watch out for class 'X' and class 'Y' capacitors
which are designed for connection to mains voltages. These are fail safe
capacitors which are self healing and must be replaced with the same type.
Replacing electrolytic caps with much higher voltage rated devices can
upset circuit operation.
Electrolytics don't start behaving as proper capacitors until they reach
a fraction of their rated voltage. Also look out for special low ESR (equivalent
series resistance) capacitors found in switch mode power supplies.
Some circuits require capacitors capable of withstanding large current
Using the wrong sort of cap in this situation would lead to overheating and
other nasty consequences.
Keep the distance between the Vdd and Vss pins on the same chip as short
as possible. There should be one connection from a chip to the power
supply. If two leads on the chip, the best method is to make a T with equal
left/right sides if possible. Don't kill yourself for it, though. If you
need to make a main feeder which splits into two short (unequal but not greatly
disparate) lengths, so be it.
The importance of this depends on frequency. Remember, a DC signal is an
AC signal with an infinite number of frequencies all out of phase with one
another. Also, you _always_ have some ripple on a trace. As the trace lengths
change, the relative position of the chip on the ripple changes. Think of
a long boat on the ocean. The keel is always "ground" but if one end of the
boat is higher than the other (i.e., there is a potential on one ground pin
vs. the other), then you're asking for problems. As the lead lengths change,
the differences become more pronounced. Worst case would be when one lead
was 1/2 wavelength from the other.
standard signal tips
Don't "despike" your signal
lines, add a resistor instead.
David Vanhorn says:
[The worst thing is ] putting a cap to ground (some arbitrary point in the
ground system) from a signal line, to "smooth it out"... Grrr. Now the driver
needs much more current on each transition, and the problem is worse.
Resistance in series solves this problem, but it's not so convenient to implement
usually. (Tip: high speed clock lines should always have a resistor at the
source. Even if later you choose the value to be zero, though likely 100
ohms will be better)
(see "Spice provides signal-integrity clues" article by Ken Boorom, EDN,
for more details. )
Cirrus (2006) recommends:
"... In addition to standard mixed-signal design techniques, system performance
can be maximized by following several guidelines during design. ...
- Place a buffer on the serial data output very near the A/D converter.
Minimizing the stray capacitance of the printed circuit board trace and the
loading presented by other devices on the serial data line will minimize
the transient current.
- Place a resistor, near the converter, beween the A/D serial data output
and the buffer. This resistor will reduce the instantaneous switching currents
into the capacitive loads on the nets, resulting in a slower edge rate. The
value of the resistor should be as high as possible without causing timing
problems elsewhere in the system."
-- from the 2006 datasheet for the CS5368 24-bit A/D converter
Keep your digital and analog circuitry physically separate whenever
Digital switching, especially at microprocessor buss or video card speeds,
can throw all sorts of noise and trash into analog or audio circuitry via
the power supply. Adaquate
Isolation is a must.
Add Grounded Shields or Ferrite
Beads to long cables carrying high frequency signals
it won't solve all problems, but it's better than nothing.
A few more tips at Massmind: Avoiding Noise.
high-power and/or high-voltage tips
Cut the PC board to seperate positive and negative power terminals in
high current products
Opening and closeing a power switch can cause a pico second arc which slowly
deposits a metal film across the board. There may also be an impurity in
the laminate that occasionally gets you. Then when you close the switch after
a number of depositions of metal, there is enough of a path for current,
and its downhill from there once there is a carbonized path. ...this is why
slots are often cut in boards around these sorts of areas, to stop potential
carbonization tracks with dirt and moisture build up.
Don't use silicon sealant to mount wire-wrap sockets or to seal/insulate
This stuff is handy and common but it is not an insulator. It will leak small
currents, which may not matter in logic circuits but can wreak havoc in
high-impedance analog circuits. Lawrence Lile [lilel at toastmaster.com]
Also, don't use it because some silicone sealants contain acetic acid (smell
'em if you don't believe me) which will react with copper or iron, causing
corrosion. Corroded copper and rust are famously poor conductors.
Use IC sockets on prototypes.
After you have things debugged, you might consider soldering chips directly
to boards to save cost and eliminate connections that can oxidize or come
loose, but during the design/testing phase you need to be able to swap chips
without repeated desoldering.
Cover the window on any chip with EPROM even if you don't care about the
EPROM part of the chip
PIC processors for example are sometimes affected by non-uv light entering
the EPROM window and shineing on the silicon. This can actually inject signals
into the non EPROM parts of the chip!
Always provide mechanical strain relief for wires soldered to a PCB
Wagner Lipenharski says:
Another possibility is drilling two extra holes at the board (little bit
bigger than the wire diameter), just to zig-zag the wire (snake path) through
them after soldering. If you don't use any mechanical lock to the wire, sooner
or later it *will* break up at the solder point.
If you DO end up just soldering the three wires to the board, the back
presumably, put a small dab of silicon sealer over the wire adjacent to [but
not in contact with] the solder point. This will hold the wires if you're
worried about movment loosening them but will still give access to the joint.
Silicon holds fast but is very easily cut off with a scalpel.
unsorted other tips
Make sure you can get the parts before basing a design on it.
You may find the ideal integrated circuit for your application in a data
book, but it may not be in production, may be unavailable from
distributors, or may be too expensive. Especially
if you are creating a design you hope to produce for a while, it's wise to
choose devices that are widely available and that (you hope) won't be
NEVER plug untested circuitry into a computer backplane slot!!!
Never never never, if you love your PC. All it takes is a simple little wiring
error and your motherboard and disk controllers will be toast. While IBM-PC/AT
buss interfacing isn't rocket science, it's all too easy to do expensive
damage to your machine. Consider other interfacing methods (parallel/serial
ports, commercial control/data acquisition boards) first. There are buffer
cards that transparently permit prototypes cards to be used while protecting
the computer buss, but they are not cheap.
Pick chip types with the correct switching speed for your design
Digital logic chips having very fast edge speeds (dV/dt), such as 74S, 74AS,
74F, and 74FACT can cause RF crosstalk and interference (especially in poorly
designed or laid out cicuits) that will frustrate the most brilliant engineer.
See Signal Levels@. Avoid them unless you need a super
high frequecy, low propagation design.
Don't trust data sheets labeled "Preliminary Information".
This means that the data sheet was written before the chip was actually in
production, and the device may change significantly by the time it is actually
released. The device performance may be different; still more important,
the pinout may be completely rearranged in the final design. Preliminary
data can help you choose a device but don't set your design in stone based
Don't rely on simulators
Simulations are doomed to succeed. Reversed biased
Transisters are a good example.
Watch out for temperature variation in component specs
Shottky's aren't better, just different
Chris Eddy says:
Just a quick look at a Diodes Inc. catalog shows that the shottky's are typically
1mA reverse current, and the regular diodes are typically 5uA reverse current.
I found out the hard way when I built a circuit that needed protection from
negative going signals. I put the Shottky across the diode in an RC filter
between two op amps, so the resistor served two purposes. I didn't look real
hard when I tested it, and the customer complained about temperature dependency.
We quickly discovered that when the cover was removed and a heat gun was
used, an easily observed error at ambient became a nearly zero signal level
out at high temperatures. The figures shown above for leakage are at 25C!
Needless to say, it was a rude awakening to the drawbacks of Shottky.
Make sure you know how your connectors actually connect
Some DIN connectors internally connect
pin 2 to the shield
Don't assume that a reverse biased bipolar
transistor will not conduct through a reverse
Adding a series diode in series with the transistor's collector is a good
idea when reverse biased C-E junction will occur if well defined operation
I believe that this document contradicts its self. In one part the use of
silicon sealer is highly discouraged because of the acid content. Below a
quote from "Jinx" says to use the sealer! Placing an acidic substance on
the back of a PCB, with the traces, is definately a bad idea.
Newton replies: Jinx specifically said not to let the sealant come
in contact with the solder point. I would expand that to include any metal
at all, which, I'm sure, is what he ment. +
/techref/pcbs.htm With a complex design,
using good PCB design software can save a lot of time.
Another point to note
about the use of silicon sealant I read once on an RS datasheet - don't use
it anywhere near switches that are switching high current - any arc that
is produced will encourage the formation of Silicon Carbide on the contacts,
which is a good insulator (at lowish voltages
Greg Pulley says:
Don't be all that concerned about the neat breadboards people post - the high-speed performance is terrible, made worse by all the mutual coupling and parasitics beyond what the crappy breadboard saddles you with.
rule one: don't worry about what it looks like.
rule two: make all bus bits the same length between parts
rule three: run clocks and critical control signals as point-to-point direct as possible. if they are above 5-8MHz, twist them with a dedicated ground wire (look up how to make twisted pair with a cordless drill).
rule four: run power and ground wires first. add plenty of 100nF decoupling caps directly across the parts. It's also good to use tall sockets and solder the decoupling caps underneath straight across the power pins. Add 10uf bulk electrolytics every 5 chips. add 100uf at the input to the circuit where the PSU comes in. ferrite beads on your power supply leads are good idea too to reduce EMI.
rule six: breadboards suck if you want whatever you build to work more than 2-3 days without becoming flakey. Use protoboards and solder, or if you have the tools, wire-wrap is a good choice too. For WW, dedicated power grids made with 14-16AWG wire are best.
| ||©2023 These pages are served without commercial sponsorship. (No popup ads, etc...).Bandwidth abuse increases hosting cost forcing sponsorship or shutdown. This server aggressively defends against automated copying for any reason including offline viewing, duplication, etc... Please respect this requirement and DO NOT RIP THIS SITE. Questions?|
<A HREF="http://www.massmind.org/techref/electips.htm"> Electronic Construction - Tips, Tricks, Gens, Traps, and Snares</A>
Welcome to www.massmind.org!