What are everyone's design procedures in tackling a circuit

D

Dan Charette

Guest
Hey Everyone...

I've been doing a lot of research lately trying to learn more and more
about different and subtle techniques with regards to circuit design,
analog and digital, in order to just be a bit better at this with each
subsequent design. As we all are I'm sure. But I thought I'd throw a
bone out on the table and ask some of the seasoned pros as well as the
amateur gadgeteers out there what their respective design approaches
are? I've been in this field since a wee lad smelling the ozone from
sticking that fork in the outlet at aged 3 or so and I've never worked
in a formal lab under the tutelage of a senior engineer with some road
mileage under their belt, so I've never had the opportunity to see
exactly the nuances of how some of the typical lab routines and
procedures work at say like NS, TI, AD or any number of the myriad of
different university, basement and garage based labs.
I've had my own design lab that has slowly grown into the
Frankensteinish environment that it is today, probably not too far off
from everyone elses. Anyway, when I ask about design approaches,
here's my example.
I recently had someone approach me about a circuit idea to improve
upon something he had that has failed and he needs to replace, but the
original maker has since vanished. So, I listened to his request and
asked him in detail about shortcomings he felt the unit could better
perform if redesigned. Once I had his philosophy on the unit of what
he felt would be an improvement, I then kindly but firmly asked what
he was looking to pay for the unit and how many was he willing to
commit to. I had a ballpark figure in my mind of timing that it might
take to accomplish the goal and kinda guessed at the manufacturing
side of things so I felt somewhat in the ballpark on how many of these
things it would take to get back the development time dollars. Once I
had that in mind, I started doing research on circuitry based on four
pieces of circuit designs I new would work to accomplish the task.
All I needed to do was to interface these four pieces and with some
tweeking, the circuit should fly. This research was a couple of good
solid days of reading and rough calculations on small circuits I was
looking to incorporate. During this time, I had tried a few spice
simulations in a few areas to corroborate my hand calculations. Then,
I started building individual blocks onto a bare PCB and tested each
section to make sure it would perform individually. I'm about half
way through this procedure as of today and I have roughly two weeks of
design time in the project so far. It's not a big project, but a
small gadget for use with microphones in pro audio.
Anyway, as I was looking through the newgroup postings, I see so many
very knowledgable folks with lots of experience and I just thought I'd
ask what everyone's procedures are i.e., do you draw out schematics
first? Do you sit down with soldering iron in hand and just start
slapping pieces together? Do you do complete spice simulations and
then build? Are there other details that perhaps are done that you do
that may be helpful to aspiring designers? I know there must be a
slew of approaches and I just thought I'd throw the question out on
the table and see how cut from the same cookie dough we electronics
gurus from different cultures and countries, lab backgrounds, and
overall work ethic really are.

Thanks for any input!



Dan Charette {dan_at_thesonicfrogFUZZ-dot-com}
Remove the "FUZZ" and replace the underscores and
such from my e-mail address to contact me.

"I may not always be right, but I'm never wrong."
 
Dan Charette <dan@thesonicfrogfuzz.com> wrote:
Hey Everyone...

I've been doing a lot of research lately trying to learn more and more
about different and subtle techniques with regards to circuit design,
snip
Anyway, as I was looking through the newgroup postings, I see so many
very knowledgable folks with lots of experience and I just thought I'd
ask what everyone's procedures are i.e., do you draw out schematics
first? Do you sit down with soldering iron in hand and just start
slapping pieces together? Do you do complete spice simulations and
then build? Are there other details that perhaps are done that you do
that may be helpful to aspiring designers? I know there must be a
It really depends.
If the design is not challenging, and does not involve many parts,
I may simply start plugging the parts into stripboard, making a reasonable
guess at spacing, soldering them in, and then finishing off jumpers.

For anything that I don't know the pin-numbers and datasheet off
by heart (4011, BUZ80, 7405, LM324, ...) it's likely to get
a circuit diagram drawn. (which may just be scribbled on an envelope)

If the application does not fit the datasheet well, or the device is
being used outside its intended application, or if it's a complex
circuit with lots of discretes, or involves many inputs/outputs,
I'll probably add some calcultions, and maybe even spice it.

If the components are very expensive, I'll do a lot more before
committing to a design.
 
On a sunny day (Fri, 30 Apr 2004 10:40:21 -0500) it happened Dan Charette
<dan@thesonicfrogFUZZ.com> wrote in
<t1r490d8flrf6mnppmgmrrhlos0v141ee9@4ax.com>:
Do you sit down with soldering iron in hand and just start
slapping pieces together? Do you do complete spice simulations and
then build? Are there other details that perhaps are done that you do
that may be helpful to aspiring designers? I know there must be a
slew of approaches and I just thought I'd throw the question out on
the table and see how cut from the same cookie dough we electronics
gurus from different cultures and countries, lab backgrounds, and
overall work ethic really are.

Thanks for any input!
One of the most complicated projects I ever did, I realized how to do when
in the shower.
I only use spice for filters and other things that need a lot measurement
testing otherwise.
Very often I just make some small test ciruits (for parts of the project),
I never start building something if I do not have a clear idea how it
will (have to) be.
Usually I use blocks that I know that work (from previous projects) just
like a software library almost.
Sometimes we rip apart the competions stuff to see how they did it..
And then try to improve a bit.
etc..
 
Hi Dan,

It also depends what the final application is. For my projects that is
often medical electronics. And here the rules are strict: You are required
to document well and in a tidy fashion from the very first hour of work.
There needs to be a seemless design history so anyone can track what you
did and why you designed it this way.

Sounds like a lot of red tape but after a while you also learn the values
of it. It becomes pretty much impossible to lose a part of code or a
circuit idea. It's all there and searchable.

But the most important aspect of this rigorous documentation style came
out about 10 years ago when a client's engineer had a severe accident,
landing in the hospital for a month. Luckily he documented stuff well and
early on so I could hop in without any delay and then hand it all over
again without much need to talk it through when he came back. If his board
design had stopped for that month the whole company would have missed a
major milestone.

Regards, Joerg
http://www.analogconsultants.com
 
Joerg...

With each project that comes my way, I try and develop a newer more
thorough method of documentation. In my shop, it's only me and a
software engineer. But, it's still nice to have the drawings and
sketches from stuff early on as you said, just because with so much
information that we have to get ourselves into, I find that I forget
so much of the little details about why we chose to do this or that.
A few years later when a reference may be made to a past project, it's
nice to be able to at least have an idea as to why something was done
a certain way. But, like I said, it's a work in progress.
Documentation that I did 10 years ago is nothing more than chicken
scratches whereas today, there are at least logged time/date comments
and miscellaneous notes regarding certain important aspects.
Joerg... how do you document your progress on your daily routine? Do
you have a notebook or something that you log notes in and then
perhaps a sketchbook or something for schematics and tables of
measurements and such? Or, do you keep maybe an electronic log via
some kind of computer program? For the past couple of years, I've
been using some sheets that I printed up that are like the standard
engineering paper somewhat. They have a large area with small light
graph paper divisions and then there is a heading area for project
information and dates. Along the side, I notate specific time in
terms of hours spent here or there. I then put these in a three ring
binder that pretty much encompasses the individual project.
Everything with regards to lab data, rough schematics and times are in
there. In the pursuit of finding an easier quicker way to be a little
more thorough, how do you do it?

Thanks!!!



Dan Charette {dan_at_thesonicfrogFUZZ-dot-com}
Remove the "FUZZ" and replace the underscores and
such from my e-mail address to contact me.

"I may not always be right, but I'm never wrong."
 
Most of the stuff we do is complex; a typical product will have a uP,
an fpga (or five), a lot of analog stuff, power supplies, and some
sort of bus or serial or Ethernet interface or something. A breadboard
would be a nightmare and waste months. We just design the final
product on paper, formally release all the documentation, lay out
production-quality boards, and let our manufacturing people build a
few first articles for us to test. The FPGA designs and firmware are
of course released after everything is working. If all goes well, we
can sell the first release.

We only breadboard small circuits, generally to characterise a part or
a circuit whose performance we can't predict. This is almost always a
small PC board (tiny surface-mount parts, picosecond design rules)
with typically 5 to maybe 20 parts max. For example, we are now
characterizing a low-jitter LC oscillator and a simple vernier
time-delay circuit, both etched on a single board about the size of a
playing card.

Seldom use Spice. NEVER use those plastic proto things.

John
 
Hi Dan,

Since I am a consultant and usually work with several clients the typical
lab book is not an option for me. There needs to be a strict separation
between all projects. So here is in essence how it goes:

The client and I discuss the projects and come up with specifications.
Once solidified these get entered into a document, immediately. A new file
system is created with blank files for schematics, lists, calculations,
simulations, text, whatever is expected to be needed. Consistency in file
names is key here. The text file will now have entries (the specs).

Now it's off to the web, to find suitable components. Those again get
written down, that way I don't check out a part again that I had already
ruled out before. Much of the stuff that doesn't go into a final document
may be handwritten but it is still filed. Once most parts are selected and
prices are established the actual design begins. First the top down
architecture, then the design of the most difficult modules. I like to
keep the easy stuff for the end, when time may be short. The other
advantage is that when running into a hardcore snag I can call the client
early on and discuss a spec modification.

When the schematic is entered the text document is generated in parallel,
with placeholders for part designators since these are going to change. At
the end, when all tests are done and the schematic is finalized these
placeholders are replaced by the final designators. Then everything is
scrutinized for errors and off it goes.

This may be a bit oversimplified of an explanation but the point is that I
always document while working, not after having finished a design.

Regards, Joerg
http://www.analogconsultants.com
 
I can certainly second the last paragraph. Over the years I have seen
tremendous grief caused by prototyping on "stick and wire" boards. One
"project" I inherited was just wild, nothing worked at all. Even my AM
radio went quiet when power was applied to this "circuitry". Then I
soldered it all on an experimenters board with a solid ground plane. Short
paths, no wires flopping around and so on. The miracle was instant: With
nary a change to the schematic everything worked great.

Regards, Joerg
http://www.analogconsultants.com
 
Ian Stirling <root@mauve.demon.co.uk> wrote in message news:<qFukc.36327$h44.5396877@stones.force9.net>...
Dan Charette <dan@thesonicfrogfuzz.com> wrote:

Anyway, as I was looking through the newgroup postings, I see so many
very knowledgable folks with lots of experience and I just thought I'd
ask what everyone's procedures are i.e., do you draw out schematics
first? Do you sit down with soldering iron in hand and just start
slapping pieces together? Do you do complete spice simulations and

Paper, paper, paper, for me. Design, tweak, fix, redesign bits, until
its all fine and dandy, because its way easier to fix bugs on paper
than after building. I dont think I ever just pick up the iron and
build. This also helps a lot to get an optimised efficient design. It
makes it easy to concentrate on each cct section one by one, picking
up any issues I didnt spot while in overview mode, and trying out
cheaper or simpler substitutions.

Spice, urggghh, I'll get used to it one day.


Regards, NT
 
On Fri, 30 Apr 2004 23:50:31 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

I can certainly second the last paragraph. Over the years I have seen
tremendous grief caused by prototyping on "stick and wire" boards. One
"project" I inherited was just wild, nothing worked at all. Even my AM
radio went quiet when power was applied to this "circuitry". Then I
soldered it all on an experimenters board with a solid ground plane. Short
paths, no wires flopping around and so on. The miracle was instant: With
nary a change to the schematic everything worked great.

Regards, Joerg
http://www.analogconsultants.com
I hate those plastic proto things, and can't visualize the circuit
when the wires keep overlapping and then disappearing inside the
block.

Very simple things can be haywired directly on a piece of copperclad
in a few minutes, even very fast stuff if you're careful. Hardline
coax shields can be soldered right to the board, and the gate of a fet
or an SRD or something can be soldered directly to the little bit of
inner conductor hanging out.

But a lot of surface-mount parts can only be handled by making a real
PC board. We use AP circuits a lot for a few pieces in a few days.
Lately, I've been making all the guys assign a real board number to
every such proto, and saving the schematic and PCB files on a
backed-up server drive - almost a formal production release - so we'll
know what the hell these little boards are for six months from now.

John
 
Hi John,

That's what I do as well. Every prototype gets a number and when several
iterations or mods are done the Rev level will be bumped X1, X2 and so on.

There is (was?) something worse than plastic prototype boards: Wire wrap. Oh,
did I hate that. And I never, ever, allowed my circuits to be wire wrapped.
Noise everywhere, even clocks talked over so badly that their crosstalk reached
the TTL thresholds at times. Then once in a while, ploink, a little wire
slipped off or broke. The broken ones could only be found by tugging on the
wires which, of course, made more wires pop off. I have seen grown up engineers
on the verge of bursting into tears. Needless to say, most wires were the same
color because the techs didn't want to change spools. Wire guns were so
expensive and sometimes rather fickle that they could not have more than one
per person.

SMT: Sometimes I use Wainright strips. When it needs to be "RF proof" what I
sometimes do is cut some strips of whatever materials can be safely glued down.
The sticks in Dove ice cream bars are great (vanilla with almond/chocolate
coating works best for me...). Then the ground pins are bent down and soldered
to the ground plane. Decoupling caps can be placed almost right on the pins.

With these newer MSOP packages this gets tough. Half a millimeter of pin
spacing really strains the eyes. Alternative up and down bending works but not
really on larger packages.

Regards, Joerg
http://www.analogconsultants.com
 
Joerg wrote:

Hi Dan,

Since I am a consultant and usually work with several clients the typical
lab book is not an option for me. There needs to be a strict separation
between all projects. So here is in essence how it goes:

The client and I discuss the projects and come up with specifications.
Once solidified these get entered into a document, immediately. A new file
system is created with blank files for schematics, lists, calculations,
simulations, text, whatever is expected to be needed. Consistency in file
names is key here. The text file will now have entries (the specs).

Now it's off to the web, to find suitable components. Those again get
written down, that way I don't check out a part again that I had already
ruled out before. Much of the stuff that doesn't go into a final document
may be handwritten but it is still filed. Once most parts are selected and
prices are established the actual design begins. First the top down
architecture, then the design of the most difficult modules. I like to
keep the easy stuff for the end, when time may be short. The other
advantage is that when running into a hardcore snag I can call the client
early on and discuss a spec modification.

When the schematic is entered the text document is generated in parallel,
with placeholders for part designators since these are going to change. At
the end, when all tests are done and the schematic is finalized these
placeholders are replaced by the final designators. Then everything is
scrutinized for errors and off it goes.

This may be a bit oversimplified of an explanation but the point is that I
always document while working, not after having finished a design.

Regards, Joerg
http://www.analogconsultants.com
In general I've found that saying "we'll document it when we're done" is
form-fit-function compatible with "build it, ship it, wonder what you did".

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
 
On Fri, 30 Apr 2004 10:40:21 -0500, Dan Charette
<dan@thesonicfrogFUZZ.com> wrote:

Hey Everyone...

I've been doing a lot of research lately trying to learn more and more
about different and subtle techniques with regards to circuit design,
analog and digital, in order to just be a bit better at this with each
subsequent design. As we all are I'm sure. But I thought I'd throw a
bone out on the table and ask some of the seasoned pros as well as the
amateur gadgeteers out there what their respective design approaches
are? ....
Interesting topic. I have not had someone lay out a procedure for me
so have developed my own, but have always wondered if there was a
better way.

At work we have software to track engineering projects, but it doesn't
do everything I need and is somewhat slow, so I supplement that with
my own documentation. The first thing I do is start an MS Word
document for the project and put down the project description.
Then I add specifications and any other details that are available.
If it's a complex project, I break it down into bite-sized circuit
sections and create topic headings for those in the document. I use
Word's index feature to build a comprehensive index at the top of the
document. The index is handy if the notes are printed, but I mostly
use the document on the computer and use the search function to find
the section of current interest.

For a complex project, I may create a rough schematic in block form.

Next I go to the web to research components and ideas, beginning with
the more challenging parts of the circuit. I find the Digi-Key web
site to be the most useful for parameter searches for components from
a pool of manufacturers. It is also the fastest route to a data
sheet. I make notes about the more promising components in the
project document, including sources, pricing, names, telephone
numbers, email addresses (the buyer will need that later). I'll try
to get some sample components in DIP packages and build the circuit,
or small sections of the circuit if it is complex, on a solder
protoboard. Occasionally, I'll use a plug-in type protoboard, but
only for very simple circuits--it seems to be faster than soldering
for simple circuits, but slower if you try to build something complex.
For the project I am working on now, I have 8 seperate proto solder
boards that connect together with board interface connectors and
ribbon cables. This gives me the freedom to alter various parts of
the circuit by rebuilding a small board rather than risk damage to my
proto board by ripping up and replacing components. I sometimes use
Spice as well, but don't feel I can rely on that only.

The results get noted in the project document and I start adding dates
with the information so I can later tell what is old and what is new.
I'll start drawing schematics for successful trials. I may add some
of these to the project document via cropped screenshots. I may
include scope images.

When I have enough information, I will assemble it into a single
schematic and create the PCB.

Recently I have discovered the digital camera. I got one with a macro
lense so I can take closeups of PCBs. I use a ring lamp with the
glass removed, put it close to the PCB and take the shot through the
center hole. It's like a microscope minus the eye strain. With this
I can document manufacturing problems, assembly instructions,
inspection instructions, test instructions, and troubleshooting
instructions. These images go into the project document and can later
be pulled out to create other documents such as the assembly and
troubleshooting instructions.

TD
 
Hi Dan,

Interesting that Teddy mentioned the MS-Word index feature. Those things
are truly helpful and what most people don't know is that all this often
comes pre-packaged with their computer. In other words you've already paid
for it with the machine.

I must admit that I still do indices by hand but I did use MS-Word's HTML
indexing and linking when working on a project with half a dozen other
people. Now when someone added a document the others could all see it
immediately. Everybody just had to be diligent about uploading right away.
The old files were automatically archived, just in case.

Databases: These I use extensively to track repeating stuff. Not just work
hours but also parts prices, problem frequencies in an EMI measurement
run, or simply addresses, you name it. What often surprises me is that
while other engineers are using spreadsheets regularly, almost none of
them are familiar with databases and certainly not with creating reports.
Same here: If you buy a PC pre-loaded with MS-Works, you've got a database
program. If it is MS-Office, you usually have a database as well. They are
all pretty compatible, even with d-Base and other formats that are popular
in larger businesses.

Regards, Joerg
http://www.analogconsultants.com
 
Hi Joerg...

Thanks for all the valuable info. It's interesting to hear the
different methodologies by which people organize themselves to perform
very similar tasks. It's kinda like a chili cookoff or something.
Everyone has their own recipe and in the end, there are a bunch of
pots of chili sitting around to ladel up and make the tummy smile.

You know, with regards to your database comment, I do have a parallel
question with this in mind. As our projects here become larger and
more ornate I'm starting to find myself bogged down with generating
approximate costs to determine what a piece of gear might cost. What
would really be so useful would be to somehow correlate my parts list
from my schematics into some kind of database that contains current
pricing and the quantity pricing breaks on those parts. Then, with a
quick little number entry, I can determine with relative accuracy a
total parts cost if I choose to build 1, 10, 25, etc. units so that I
may be able to better plan a strategy for what it'll take for an
investment.
Does that make sense? Is that something that you worry about with
your design duties or is that a parts ordering/accouting person that
handles that. It's definately a job in and of itself on say a larger
project. I can see very easily how choosing to build 10 units of
widget A or 25 of them could really screw up the long term planning on
the product and it's final retail price. We all know it's about
quantity, but finding a nice way to be able to really accurately
predict the outcome would be extremely handy.

Thanks...

Dan


Dan Charette {dan_at_thesonicfrogFUZZ-dot-com}
Remove the "FUZZ" and replace the underscores and
such from my e-mail address to contact me.

"I may not always be right, but I'm never wrong."
 
Hi Dan,

Often I have to design for rock bottom pricing and that requires keeping
close tabs on the bill of materials. Some CAD programs allow to enter
pricing info in a "hidden" field. That gives you an instant cost rundown
every time you run the BOM.

Of course you can also use a database but then you have to create a link
between schematic and database (ideal) or enter it by hand (not so ideal).
Mostly I use a very archaic route and it works well on moderate complexity
designs. I keep a paper sheet, vellum for easy erasing, and chalk up the
parts as I place them, erasing those that I delete again. The trick is to
use the same parts over and over again, which is something that needs to
be done anyway to keep SMT part feeder rigging costs down. So when I use
only one part of a BAV99 double instead of a single diode the reason is
reduced parts variety. Each semiconductor needs their own line on the
sheet but resistors and many ceramic caps can share one line if their
pricing is identical.

I know this is archaic but it works. Electricity used to go out a lot here
and then the batteries only allow a few more hours. With a "paper
database" I can go on forever.

The one thing I wish all manufacturers would learn is to make budgetary
prices available without any gimmickry. When it says "call" or plain
nothing I usually move on, no time for that. Some manufacturers seem to
have no idea how much business they are losing because they fail to state
budgetary pricing on the web site.

Regards, Joerg
http://www.analogconsultants.com
 
On Sat, 01 May 2004 19:44:36 GMT, Joerg
<notthisjoergsch@removethispacbell.net> wrote:

Hi Dan,

Often I have to design for rock bottom pricing and that requires keeping
close tabs on the bill of materials. Some CAD programs allow to enter
pricing info in a "hidden" field. That gives you an instant cost rundown
every time you run the BOM.

Of course you can also use a database but then you have to create a link
between schematic and database (ideal) or enter it by hand (not so ideal).
Mostly I use a very archaic route and it works well on moderate complexity
designs. I keep a paper sheet, vellum for easy erasing, and chalk up the
parts as I place them, erasing those that I delete again. The trick is to
use the same parts over and over again, which is something that needs to
be done anyway to keep SMT part feeder rigging costs down. So when I use
only one part of a BAV99 double instead of a single diode the reason is
reduced parts variety. Each semiconductor needs their own line on the
sheet but resistors and many ceramic caps can share one line if their
pricing is identical.

I know this is archaic but it works. Electricity used to go out a lot here
and then the batteries only allow a few more hours. With a "paper
database" I can go on forever.
I do most of my designing on D-size blue-gridded vellum. Design
Automation is an electric pencil sharpener and an electric eraser.

John
 
Tim Wescott <tim@wescottnospamdesign.com> wrote in message news:<1096afneco2bn66@corp.supernews.com>...
Joerg wrote:

In general I've found that saying "we'll document it when we're done" is
form-fit-function compatible with "build it, ship it, wonder what you did".

I've always handwritten design notes because it allows speed, which is
very important, a permanent record, permits mixing all types of
text/diagrams/etc, and adding notes later on. I've always though there
must be something computer-based to do this, but have not found
anything really satisfactory. I use a text editor a fair bit, but it
wont do everything. Is there a package you use that allows all this to
be done, and most importantly, done fast? I'm talking here about the
as I go along stuff, not final paperwork.


Regards, NT
 
Hi John,

It's even more retro in my case. Mechanical sharpener, sometimes grandpa's old
knife. Good old hand held eraser.

I have never seen an electric eraser...

Regards, Joerg
http://www.analogconsultants.com
 
Hi NT,

Not really I guess. The closet may be these PDAs which have a stylus to enter sketches. But for me they are
to small and text entry into these is a real pain.

What I really look for is an affordable electronic white board that can also be used in a teleconference.

Regards, Joerg
http://www.analogconsultants.com
 

Welcome to EDABoard.com

Sponsor

Back
Top