Searching \ for '[EE] Poor man's very low frequency oscilloscope.' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=poor+mans+very+low
Search entire site for: 'Poor man's very low frequency oscilloscope.'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Poor man's very low frequency oscilloscope.'
2005\02\05@032103 by ThePicMan

flavicon
face

Hello PIClisters!

I'm not extremely experienced in the PIC world, but I've already
designed and built some devices (the biggest so far was Christmas
lights for my kids :D ). Now I'd like to make a device for my own
use and usefulness. It's a "poor man's very low frequency scope",
because I need to monitor an analog low voltage (from 0.0V to 1.5V)
input that changes slowly, and record and log this waveform on my
PC via RS232, where I'll write a small software to do the rest of
the job (display the waveform, save, print, etc..).

True, I could buy or copy some existing similar project, but where
would be the fun of it?

I've chosen a 18x MPU for this project because I think the 16x are
(or are going to become) obsolete, but mostly because all my PIC
experience so far is in the 18x parts and I'd like to use a nice
PIC18F1220 for my project that is on my bench all the time looking
me with sweet eyes that seem to ask "when will you use me?". Its
eyes are just like the ones of Shrek2's cat, you know? look here:

http://www.topdownloads.net/wallpapers/view.php?id=12896

Now you know why I HAVE TO use this 1220. ;)

Now it's my question time! :)

1) I plan to use the PIC18F1220's internal 8MHz oscillator and to
  bit bang a PIC I/O pin for the UART TX part (just because it's
  more fun to do so and because I won't have to cope with the
  1220's built-in UART's baud rate divisor limitations).
  Considering that I will thus have more control on the exact
  baud rate (relative to the clock), will the INTOSC +/- 2% be
  still ok for transmitting reliably? $1 BILLION QUESTION: What
  if I calibrate the unit? Will the INTOSC remain constant in
  the following years (given the same ambient temperature and
  Vdd)?

2) Use a full MAX232 or some diode/transistor-based hack? Consider
  that communication is only one-way (from circuit to PC COM port).

3) Should I buffer the input with an op-amp?
  While I plan to use my device mostly for monitoring SLOW stuff
  (producing a change in voltage on average I think no more than
  0.1V every second), I wouldn't dislike a greater versatility
  from the device, so that in the future it would be able to give
  me a nice oscilloscope-like plot of a quickly changing high
  impedance source. Being able to measure also negative voltages
  would be nice too!
  So maybe I should really use a (FET input?) op-amp between my
  circuit's input connector and the analog input pin of the PIC.
  That would also help me add some gain to the signal. What type
  of (not too hard to find) op-amp would do you suggest me for
  this use?

4) Would it be wise to power everything off the PC's RS232? Together
  TX, RTS and DTR may give me at least 20mA even on a portable PC,
  I think. While it sure would be "wiser" to have an external power
  supply, it would certainly be better if it didn't have one. But
  if I have to make a PSU for it.. what would be the cheapest and
  lightest solution?

5) I will put the circuit board into an aluminium box. Should I
  ground it then? Also the input and RS232 connectors? I'm afraid
  I will remove some isolation this way (some devices I will
  monitor may be at a different ground potential than the PC).
  Also, should I short circuit the RS232 connector's pin 5 (SGND)
  to the chassis ground? I've read somewhere that SGND and the
  cable's shield should not be connected together.

6) Anything I should be aware of and possibly (probably!) I ain't?

Of course I will share the whole project (schematics, PCB, and info)
with the List once it's complete, running, tested and blown up! :D

Oh my God.. there's smoke on the bench.. gotta close this post now..
hurry up.. bye!! ;D

Cheers,
--
PicMan

2005\02\05@074810 by Gerhard Fiedler

picon face
ThePicMan wrote:

> It's a "poor man's very low frequency scope", because I need to monitor
> an analog low voltage (from 0.0V to 1.5V) input that changes slowly, and
> record and log this waveform on my PC via RS232

> (some devices I will monitor may be at a different ground potential than
> the PC).

I think this is the crucial point that you need to address before deciding
anything else. How much isolation do you want? And how do you want to
achieve this?

Basically you can either put your circuit on the PC's ground and create a
separation in the analog input, or you can put your circuit on the input's
ground and create a separation in the serial link between the circuit and
the PC.

Creating a separation in the input may be as simple as a difference
amplifier (a few up to a few ten volts of ground difference) or as
expensive as a full-blown isolation amplifier (a few hundred up to a few
thousand volts).

Creating a separation in the serial link can be done with optos and either
a separate power supply (transformer-based) or a dc-dc converter (with
separate grounds for input and output, of course).


The answers to most of the other questions (input amplifier, serial
circuit, ground connections) depend on the decision about how you deal with
the ground isolation.

Gerhard

2005\02\05@074930 by Jinx

face picon face
> 1) I plan to use the PIC18F1220's internal 8MHz oscillator and to
>    bit bang a PIC I/O pin for the UART TX part (just because it's
>    more fun to do so and because I won't have to cope with the
>    1220's built-in UART's baud rate divisor limitations).

What limitations ?

At 8MHz there are several common baud rates with just 0.16% error

Speaking from personal experience, "not having to cope" isn't
learning but maybe it's something you might feel like tackling some
other day

>    Considering that I will thus have more control on the exact
>    baud rate (relative to the clock), will the INTOSC +/- 2% be
>    still ok for transmitting reliably?

Those two statements seem at odds - yes, you could get down
to the finest detail of timing control, but that could be anywhere
within the PIC's +/- 2%, something over which you have little
control. Why not a resonator or crystal ? Also a lower frequency
will use less power

> $1 BILLION QUESTION: What if I calibrate the unit? Will
> the INTOSC remain constant in the following years (given the
> same ambient temperature and Vdd)?

How will you calibrate it ? And how often ? AFAIK the INTOSC
doesn't deteriorate with time but you'd need to check for any
info about that at Microchip's site

> 2) Use a full MAX232 or some diode/transistor-based hack?
> Consider that communication is only one-way (from circuit to
> PC COM port)

There have been simple interfaces posted, but not all can be
used in every situation with every PC. At least with a more
sophisticated i/f you are pretty much guaranteed performance

> 3) Should I buffer the input with an op-amp?

> Being able to measure also negative voltages would be nice too!

Which you could do quite easily with an op-amp or two

> So maybe I should really use a (FET input?) op-amp
> between my circuit's input connector and the analog input
> pin of the PIC.

That depends on the drive available from the voltage source, but
pretty much any op-amp is much higher impedance than the ADC
input recommendations

>    That would also help me add some gain to the signal. What type
>    of (not too hard to find) op-amp would do you suggest me for
>    this use?

An LM324 would do, although it's not rail-to-rail. With the range
of voltages you mentioned that shouldn't be a problem, but there
are better options. Check out an op-amp spec and comparison table,
eg at Microchip or any manufacturer eg Fairchild, National, Linear,
Maxim, or retailer, eg Digikey, RS, Farnell, Mouser. I use the
TLC271 or LC272 quite a bit (mainly because I got sent 200 each
by mistake and gave up fighting the company who sent them and
believed their paperwork over me)

> 4) Would it be wise to power everything off the PC's RS232?

Wise is relative. If the draw-off is able to be supplied comfortably
by "a" PC then it should be OK. Another PC may not have the same
power available though

>    if I have to make a PSU for it.. what would be the cheapest and
>    lightest solution?

Cell-phone charger ? Cheap, light, about the right voltage range

> 5) I will put the circuit board into an aluminium box. Should I
>    ground it then?

If there's a continuous ground from source right through to PC
it probably wouldn't hurt, but it also rather depends on what you
expect the circuit to be shielded against

> Also the input and RS232 connectors? I'm afraid I will remove
> some isolation this way (some devices I will monitor may be at
> a different ground potential than the PC)

In which case you'll perhaps have to consider some optical bridge
in the circuit somewhere. Linear optos  for the input maybe or an
opto on the serial output of the PIC. You might be able to isolate
the signal somewhat using resistors but that will probably depend
on the actual application

> Of course I will share the whole project (schematics, PCB,
> and info) with the List once it's complete, running, tested and
> blown up! :D
>
> Oh my God.. there's smoke on the bench.. gotta close this
> post now.. hurry up.. bye!! ;D

Hmmm, that sounds encouraging ;-)

But hey, I nearly set the place on fire today guiltily yacking on
the phone when I knew damn well I shoulda been watching the
frying pan !

2005\02\05@075915 by John J. McDonough

flavicon
face
----- Original Message -----
From: "ThePicMan" <spam_OUTthepicmanTakeThisOuTspaminfinito.it>
Subject: [EE] Poor man's very low frequency oscilloscope.


> 2) Use a full MAX232 or some diode/transistor-based hack? Consider

> 4) Would it be wise to power everything off the PC's RS232? Together

If you are always going to use the SAME PC, then use your transistor hack
and power it from the serial port.

Keep in mind that serial ports are becoming extinct, and perhaps more
frustrating, there is wide variation in PC serial ports.  I did a survey
about a year ago and serial port voltages range from +/- 15 (which used to
be the standard) to +/- 0.9 (!). On some ports, anything below about +0.7
will be counted as negative, on others, you need to get below -5.  The
MAX232 is simple and will get you a nice swing that will work with about
anything.  The transistor probably is already sitting in your parts drawer,
but what works with your current PC might not work with the next, if it even
has a serial port.

Ditto for the power.  Most PCs seem to put out about 9 volts, but much lower
voltages aren't uncommon.  If you power your circuit from the port, be
prepared for some nasty surprises from the next PC you try it on.

--McD



2005\02\05@090021 by ThePicMan

flavicon
face

Hi Gerhard,
first of all, thanks for your reply.


{Quote hidden}

How true, this is a crucial point that reached my mind shortly after hitting
the "send" button. What I decided then was to isolate the serial port, via an
optocoupler (4N25?). Basicly I would get serial high from DTR and serial low
from RTS, and then use the 4N25(?)'s optotransistor to drive RX through the
current from DTR/RTS. I don't need handshaking anyway. I will set DTR and RTS
to the right values in the my PC application.

Now I'm still not sure if I should use a transformer or if I could use the
DTR/RTS also to power an isolated DC-DC converter, like those nice ones made
by Texas Instruments (e.g. DCR010505), but they may be too noisy, perhaps.
I'm not truly attracted to the idea of a transformer.. but if it proves to
be the best solution, I'll swallow it and forget.

In any case the MAX232 doesn't seem necessary anymore..

Thanks,
PicMan

2005\02\05@090021 by ThePicMan

flavicon
face

Hello Jinx!


At 01.48 2005.02.06 +1300, you wrote:
>> 1) I plan to use the PIC18F1220's internal 8MHz oscillator and to
>>    bit bang a PIC I/O pin for the UART TX part (just because it's
>>    more fun to do so and because I won't have to cope with the
>>    1220's built-in UART's baud rate divisor limitations).
>
>What limitations ?
>
>At 8MHz there are several common baud rates with just 0.16% error

Cool.. I must have wrong memories then (16F?).


>Speaking from personal experience, "not having to cope" isn't
>learning but maybe it's something you might feel like tackling some
>other day

Or even today. Since the 18F1220 has an UART, and the error is only
0.16%, why not use it.. I'll spend my "hacking" time in something
else then.


>>    Considering that I will thus have more control on the exact
>>    baud rate (relative to the clock), will the INTOSC +/- 2% be
>>    still ok for transmitting reliably?
>
>Those two statements seem at odds - yes, you could get down
>to the finest detail of timing control, but that could be anywhere
>within the PIC's +/- 2%, something over which you have little
>control. Why not a resonator or crystal ? Also a lower frequency
>will use less power

I'd like to keep the number of components as down as possible.


{Quote hidden}

Is there a kind of op-amp, or a cheap instrumentation amp, which
has a bias control, gain control, and is rail-to-rail? In exchange
I can give away speed and noise requirements.


>> 4) Would it be wise to power everything off the PC's RS232?
>
>Wise is relative. If the draw-off is able to be supplied comfortably
>by "a" PC then it should be OK. Another PC may not have the same
>power available though

Yes.. and considering the quality of the (latest, before they
completely disappear) COM ports founds in PC's.. there's not much
to be optimist in this field.


>> 5) I will put the circuit board into an aluminium box. Should I
>>    ground it then?
>
>If there's a continuous ground from source right through to PC
>it probably wouldn't hurt, but it also rather depends on what you
>expect the circuit to be shielded against

I reckon I should seek the ground of the interfering eletromagnetic
field of interest.


>> Also the input and RS232 connectors? I'm afraid I will remove
>> some isolation this way (some devices I will monitor may be at
>> a different ground potential than the PC)
>
>In which case you'll perhaps have to consider some optical bridge
>in the circuit somewhere. Linear optos  for the input maybe or an
>opto on the serial output of the PIC.

But are linear optos precise enough? For me it's important to
contain the measurement error within about 1%.


{Quote hidden}

:D


>But hey, I nearly set the place on fire today guiltily yacking
>on the phone when I knew damn well I shoulda been watching the
>frying pan !

S*it happens. :]

Have a nice weekend!

2005\02\05@090022 by ThePicMan

flavicon
face

Hello John, thanks for your reply!


{Quote hidden}

Yup. And what about using an optocoupler, driving the PC's RX via it and
getting from the PC's DTR/RTS the high and low (basicly the optotransistor
would put RX at RTS or DTR potential, depending on the light reaching it)?

If it's a good idea, could you/anybody suggest me a schematic for implementing
the above? I guess I'll need apart the optocoupler (4N25?) also a resistor,
something like:

             .------- DTR (placed at "low", i.e. + ~12V)
             |
             |
             \
             / R = ? ohm
             \
             /
             |
             |
             *------- RX
             |
             |
  4N25  == |/
        == |\
             |
             |
             \------- RTS (placed at "high", i.e. - ~12V)


How do I calculate the optimal value of R?


>Ditto for the power.  Most PCs seem to put out about 9 volts, but much lower
>voltages aren't uncommon.  If you power your circuit from the port, be
>prepared for some nasty surprises from the next PC you try it on.

I'll use an external power supply then.. :P

Greets,
ThePicMan

2005\02\05@115423 by davedilatush

picon face
ThePicMan wrote...

>...Now I'd like to make a device for my own
>use and usefulness. It's a "poor man's very low frequency scope",
>because I need to monitor an analog low voltage (from 0.0V to 1.5V)
>input that changes slowly, and record and log this waveform on my
>PC via RS232, where I'll write a small software to do the rest of
>the job (display the waveform, save, print, etc..).

I did something like that a few years ago, using a PIC18F242 and
an LTC2400 24-bit A/D converter.  The schematic diagram and MPASM
source files are at:

http://mywebpages.comcast.net/dilatush/picdaq.zip

Maybe there'll be something useful in there.

Dave D.

2005\02\05@134902 by ThePicMan

flavicon
face
At 16.54 2005.02.05 +0000, you wrote:
>ThePicMan wrote...
>
>>...Now I'd like to make a device for my own
>>use and usefulness. It's a "poor man's very low frequency scope",
>>because I need to monitor an analog low voltage (from 0.0V to 1.5V)
>>input that changes slowly, and record and log this waveform on my
>>PC via RS232, where I'll write a small software to do the rest of
>>the job (display the waveform, save, print, etc..).
>
>I did something like that a few years ago, using a PIC18F242 and
>an LTC2400 24-bit A/D converter.  The schematic diagram and MPASM
>source files are at:
>
>http://mywebpages.comcast.net/dilatush/picdaq.zip
>
>Maybe there'll be something useful in there.

Thanks Dave! I'll take a look at it.

Cheers,
ThePicMan

2005\02\05@165608 by Jinx

face picon face
> Or even today. Since the 18F1220 has an UART, and the error
> is only 0.16%, why not use it.. I'll spend my "hacking" time in some-
> thing else then

Given the choice, I would take advantage of the USART that
Microchip thoughtfully included ;-)  Now, you might get lucky
and be able to tune INTOSC to, or closer to, 8.192MHz, a hex
value which is just a smidgeon over 1% higher than 8MHz. It's a
much better timimg base, reducing errors to 0% for many common
baud rates. And remember that mA / MHz figure for lower power

> Is there a kind of op-amp, or a cheap instrumentation amp, which
> has a bias control, gain control, and is rail-to-rail? In exchange
> I can give away speed and noise requirements.

You said that measurement error below 1% is OK and that you'd
like to measure -ve voltages. A non-RtoR amp would be quite useable.
Bias the output to say 2.5V, which = 0V op-amp input. Your PIC
ADC could measure -2.5 to 1.5V directly. Adding +ve gain could
prove problematic with just 5V supply (and o/p clipping below the rail),
it would be less so with a higher op-amp V+ and resistive divider on the
o/p

I doubt if noise is going to seriously affect the reading. It's generally
in the nV region, far less than 1 bit. Similarly speed. I think you said
that the voltage is way less than 1Hz. Even slow CMOS amps have
several 10's kHz bandwidth

> But are linear optos precise enough? For me it's important to
> contain the measurement error within about 1%.

There are purpose-built linear optos. They generally have two LEDs.
If you don't trust the opto or want to cobble your own, characterise it
and use a table

2005\02\06@080228 by Gerhard Fiedler

picon face
ThePicMan wrote:

>>> (some devices I will monitor may be at a different ground potential
>>> than the PC).
>>
>> I think this is the crucial point that you need to address before
>> deciding anything else. How much isolation do you want? And how do you
>> want to achieve this?

> What I decided then was to isolate the serial port, via an optocoupler
> (4N25?).

> Now I'm still not sure if I should use a transformer or if I could use
> the DTR/RTS also to power an isolated DC-DC converter, like those nice
> ones made by Texas Instruments (e.g. DCR010505), but they may be too
> noisy, perhaps.

I don't think that you will see noise problems in the precision range you
need. (You wrote somewhere 1%.) As others wrote, you may have trouble with
different PCs. Depends on your requirements of inter-exchangeability
between PCs.

So now that you have decided that you want your circuit on the ground level
of the input, you obviously don't want to connect your case to the PC
ground. Besides the serial connections and possibly the DC-DC converter
input, nothing in your circuit should be connected to anything coming from
the PC.

I think a common ground point near the input would be good. You connect the
input, the DC-DC converter ground and the circuit ground to it. Possibly
not the case -- I'm not sure you want a case that can be touched to be on
the same potential as a circuit under test. Depends on the potentials...

Gerhard

2005\02\06@125002 by Mark Rages

face picon face
On Sat, 05 Feb 2005 09:22:01 +0100, ThePicMan <.....thepicmanKILLspamspam@spam@infinito.it> wrote:
>
> Hello PIClisters!
>
> I'm not extremely experienced in the PIC world, but I've already
> designed and built some devices (the biggest so far was Christmas
> lights for my kids :D ). Now I'd like to make a device for my own
> use and usefulness. It's a "poor man's very low frequency scope",
> because I need to monitor an analog low voltage (from 0.0V to 1.5V)
> input that changes slowly, and record and log this waveform on my
> PC via RS232, where I'll write a small software to do the rest of
> the job (display the waveform, save, print, etc..).
>

PicMan,

Here's an alternate solution.

Use a voltage-to-frequency converter (such as LM331), passing the
output through an optocoupler or transformer if isolation is
necessary, then to your computer's sound input, where you have a
program running to count zero crossings of the audio signal.

Advantages:
1) simple
2) 0.01% linearity. (According to the datasheet.  Probably requires a
nice mylar capacitor to achieve)
3) easy isolation.

Disadvantages:
1) you must have an open soundcard input.
2) your software must do some rudimentary signal processing (counting
zero crossings)
3) range is limited by your soundcard's frequency response.  If it has
a 20Hz-10kHz response, you have about 54 dB of dynamic range.   Notice
the range doesn't have to be flat and linear, the soundcard just has
to register zero crossings at the frequency.

Regards.
AnalogMan
markrages@gmail


--
"We act as though comfort and luxury were the chief requirements of
life, when all that we need to make us happy is something to be
enthusiastic about."

- Einstein

2005\02\07@060357 by Alan B. Pearce

face picon face
>1) I plan to use the PIC18F1220's internal 8MHz oscillator and to
>   bit bang a PIC I/O pin for the UART TX part (just because it's
>   more fun to do so and because I won't have to cope with the
>   1220's built-in UART's baud rate divisor limitations).


I don't know what limitations you mean, but unless you specifically need to
use the RX pin for something else, I would use the hardware uart. You do not
say what language you are programming in, but if using the Microchip C18
compiler, there is a software uart module already there, as well as a
hardware uart module. I would also look at having two way communication, and
build in some form of "speed selection" that changes the analogue sampling
rate, by sending commands from the PC. You may also want to have multiple
channels that you can switch off/on, again from the PC. The internal
oscillator should be perfectly adequate for baud rate generation given that
the other end should be crystal controlled in the PC, so you can use pretty
well all your frequency tolerance at the PIC end.

>2) Use a full MAX232 or some diode/transistor-based hack? Consider
>   that communication is only one-way (from circuit to PC COM port).

definitely use one of the MAX series chips, or one of the many clones. You
will thank yourself in the long run, and they are readily available at low
price.

>3) Should I buffer the input with an op-amp?



Yes, get yourself a quad rail-to-rail op amp.
>   So maybe I should really use a (FET input?) op-amp between my
>   circuit's input connector and the analog input pin of the PIC.
>   That would also help me add some gain to the signal. What type
>   of (not too hard to find) op-amp would do you suggest me for
>   this use?

A rail-to-rail op amp, of which there are many to choose from will give you
all this. There are also the possibility of changing the gain as well,
another reason to have two way communication. Using a quad op-amp will allow
you to have up to 4 channels which I am sure you will find a use for, even
if you cannot think of one at this stage.

>4) Would it be wise to power everything off the PC's RS232? Together
>   TX, RTS and DTR may give me at least 20mA even on a portable PC,


No, get yourself a wall-wart, and properly filter and regulate it. You will
thank yourself in the long run - think in terms of being able to disconnect
the PC, and go do something else while your PIC keeps logging the data into
internal EEPROM, as a later feature you can fit. Then come back some time
later and download - another reason for two way comms.
>5) I will put the circuit board into an aluminium box. Should I
>   ground it then? Also the input and RS232 connectors? I'm afraid


You will need to have some sort on connection between the box and the ground
side of the PIC. It may not need a direct connection, you may be able to
have a 10k-100k resistor which should be a low enough impedance most of the
time, but you may want to make provision for a direct connection using a
switch to short out the resistor.

>6) Anything I should be aware of and possibly (probably!) I ain't?

See reasons above for two way comms, and possible enhancements like multiple
channels and un-attended data capture. All easily added with absolutely
minimal extra hardware, provided you think about it at the design stage.

>Of course I will share the whole project (schematics, PCB, and info)
>with the List once it's complete, running, tested and blown up! :D

I am sure James would appreciate another project he can list on the
piclist.com pages.

>Oh my God.. there's smoke on the bench.. gotta close this post now..
>hurry up.. bye!! ;D

Yeah, and here comes my boss ...



2005\02\07@110738 by James Newtons Massmind

face picon face
This triggered a REALLY outside the box idea...

...use analog electronics to reduce the amount of information that needs to
be sent to the computer. Analog data compression if you will.

E.g. use a frequency to voltage converter to find what frequency the signal
is, and then filter out that AC component. Send only the frequency and the
remaining DC signal to the PC. Now, in the PC, use a program to reconstruct
what the original signal was based on the information.

How you do that is a good question. I know exactly how it is done in
software, and I could guess at a digital circuit that would do it, but in
the analog world I'm worse than clueless: I come up with really strange
ideas! <GRIN>

Perhaps something like an audio equalizer? Is it not true that a filter both
removes part of the signal but can also be used to measure how much of the
signal is being removed and by knowing the electrical characteristics of the
filter that signal can be reconstructed and re-injected into the remaining
signal at a later date?

There was a system used years ago called "audio companding" which
"compressed" the high frequency signal for tape recorders. It was combined
with another circuit for play back that re-expanded the signal on playback.

Some sort of calibration would probably be needed to set up the software as
an accurate negative model of the analog component...

A quick Google for ["analog data compression" circuit] came up with:
http://www.freepatentsonline.com/image-4441183-1.html which looks really
interesting, although it is more digital than analog.

---
James.



> Use a voltage-to-frequency converter (such as LM331), passing
> the output through an optocoupler or transformer if isolation
> is necessary, then to your computer's sound input, where you
> have a program running to count zero crossings of the audio signal.


2005\02\07@112813 by Alan B. Pearce

face picon face
Following on from previous messages, you might want to look at the following
items (discovered as a result of trawling around the Linear technology
website after looking at Dave's project)

For an input op amp, use an LT1990 (will withstand +/-250V on inputs, a bit
hard to blow up accidentally, has switchable gain of 1 or 10) or an LT1991
(has switchable gains through internal resistors which can be mixed and
matched to select gain, configurator spreadsheet also downloadable). Either
one could provide useful gain switching under software control with a bit of
thought.

2005\02\07@120553 by olin_piclist

face picon face
James Newtons Massmind wrote:
> E.g. use a frequency to voltage converter to find what frequency the
> signal is, and then filter out that AC component. Send only the
> frequency and the remaining DC signal to the PC. Now, in the PC, use a
> program to reconstruct what the original signal was based on the
> information.

This works in theory if the input signal is a single repeating wave riding
on a slowly varying bias, and the wave shape and phase are not important.  I
haven't been paying attention to this thread that much, so I don't know if
these restrictions are true.

However, in practise it's very difficult to "then filter out that AC
component" given only a voltage representing its amplitude.  This sort of
thing is much better suited to digital techniques.  You could have a DSP
identify the repeating wave, and separate the original signal into the
repeating AC component and a DC component.  I did something similar once in
a system that had to eliminate repeating noise that was only roughly known
ahead of time.  Once you have the amplitude of the repeating signal and the
DC level you send them on to the PC as you suggested.

In other words, your general concept might work, but your algorithm is much
better suited to a DSP than to analog electronics.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

2005\02\07@123744 by Mark Rages

face picon face
On Mon, 7 Feb 2005 08:07:30 -0800, James Newtons Massmind
<jamesnewtonspamKILLspammassmind.org> wrote:

> E.g. use a frequency to voltage converter to find what frequency the signal
> is, and then filter out that AC component. Send only the frequency and the
> remaining DC signal to the PC. Now, in the PC, use a program to reconstruct
> what the original signal was based on the information.

That's all you need for a signal that is DC+a single frequency
component, although you cannot construct the original signal without
also having phase and amplitude information.

The real problem with this approach is that most signals will have
more than one frequency component.

> How you do that is a good question. I know exactly how it is done in
> software, and I could guess at a digital circuit that would do it, but in
> the analog world I'm worse than clueless: I come up with really strange
> ideas! <GRIN>

You need to understand Fourier analysis to understand this kind of
problem.  Not *how* to do the Fourier transform, just the concepts
involved.

{Quote hidden}

Actually that looks like something you might see in a PIC datasheet: a
way for one pin to measure digital or analog voltages, switchablew by
software.

If this kind of thing iterests you, you might want to look at
"pre-digital" computer technology, such as the Philbrick Applications
Manual for Computer Amplifiers.

Regards,
Mark
markrages@gmail

--
"We act as though comfort and luxury were the chief requirements of
life, when all that we need to make us happy is something to be
enthusiastic about."

- Einstein

2005\02\07@142335 by James Newtons Massmind

face picon face
I was thinking that you would send BOTH the frequency component AND what was
left of the original signal after removing that component.

The point is to gain intelligence about the signal and therefore transmit
less information while still conveying the entire original.

The big problem is the bandwidth into and out of a PC. If you can use analog
components to compress the amount of data needed, then a smaller, slower A2D
can still be effective.

---
James.



> {Original Message removed}

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...