Searching \ for '[EE] ADC Input Buffer Circuit' 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/io/atod.htm?key=adc
Search entire site for: 'ADC Input Buffer Circuit'.

Exact match. Not showing close matches.
PICList Thread
'[EE] ADC Input Buffer Circuit'
2008\08\03@180109 by Forrest W Christian

flavicon
face
I asked a similar question a while back, and that gave me some great
ideas, but I still haven't stumbled on quite the right solution.

Essentially I need to provide the functionality of a half-dozen or so
"Voltmeters" on a telemetry project I am working on.   For each
"voltmeter", I need to have a positive and negative input terminal -
just like a regular meter, and be able to read the positive or negative
voltage which appears across that pair of terminals.

Here's the problem:   The voltage sources may or may not be
interconnected with each other and/or the voltage source for the
telemetry instrument.  In addition, the voltage sources may be
interconnected but at a higher reference voltage - I.E. a high side
shunt or similar.   Think solar site where the instrument is powered
from the same batteries that you are measuring voltage from - and that
you're also measuring things like current shunts, and other devices at
the site (voltage outputs from other sensors).

In short, I need the functional equivalent of an instrumentation
amplifier which can be adjusted to measure input voltages over a wide
range (+-60V) and like I mentioned above may or may not be
interconnected electrically with the amplifier's power supply.   I'd
also like to do my best to make it "idiot-destruction" proof.

Cost is an issue...  I probably have a couple dollar budget for the
components for this section, although I could squeeze a bit more if I
found the exact right solution (I.E. a multi-input SPI Instrumentation
ADC which needs very few external components).

Accuracy isn't all that critical...  After all, I'm planning on feeding
this into a 12 bit ADC.

Ideas?

-forrest

2008\08\03@182928 by Richard Prosser

picon face
Forrest,

Could you run separate sensors & connect them by a) optoisolators  at
the digital level, or b) radio (zigbee  etc) ? Passing analogue
readings through an opto can be done but it needs care.
Alternatively you could do a voltage-frequency conversion and use
simple transformers.

A simple PIC etc for each sensor  feeding an opto is likely to be
cheapest but the Nordic devices that include a 8051 micro and a radio
transceiver may be OK also.   You'll need to provide a simple power
supply, but you'll probably need that no matter which way you go.

You don't state the update rate so I'm guessing it will be slow.
Otherwise you might need bidirectional comms to prevent collisions.
Checksum or CRC error detection is all that's likely required and
ignore corrupt data packets.

Maybe the sensor micro could use a PWM channel to regulate it's own
SMPS? - Almost a single chip + opto solution.


You may be able to group sensors to perform multiple tasks - e.g one
group measuring negative referenced signals (e.g low voltage side
battteries, temperature), one group measuring positive referenced
signals (High voltage side batteries, current shunts).

RP

2008/8/4 Forrest W Christian <spam_OUTforrestcTakeThisOuTspamimach.com>:
{Quote hidden}

> -

2008\08\03@190038 by Harold Hallikainen

face
flavicon
face
I did something like this for a battery monitoring system for an electric
car. Each 12V battery got a PIC that measured battery voltage,
temperature, and also PWM drove a shunt circuit to prevent overcharging
when the series string was charged. Each card used the UART to drive an
opto coupler that drove an open collector bus. The opto was arranged so
that when the UART was in the idling mark condition, the transistor in the
opto was off. All these opto transistors were in parallel and transmitted
packets on a somewhat random basis. A central controller gathered the
data. This "Aloha" network has an occasional collision, but these packets
are thrown out. The reduced cost of a "transmit only" node made a few lost
packets worthwhile. Each of these boards were powered by the battery it
was monitoring.

One problem with almost any sort of isolated measurement system is that
you have to power the circuitry on the isolated side. In the case above,
the isolated circuitry was powered by the battery being monitored. In many
cases, you don't have that option and need to add a DC to DC converter for
each channel being isolated. It gets expensive fast. Analog devices has
some interesting isolation amplifiers that have the isolated DC to DC
converter built in. For more info, see
www.analog.com/en/amplifiers-and-comparators/isolation-amplifiers/products/index.html
.

Harold

--
FCC Rules Updated Daily at http://www.hallikainen.com - Advertising
opportunities available!

2008\08\03@190104 by Forrest W Christian

flavicon
face
Richard Prosser wrote:
> Could you run separate sensors & connect them by a) optoisolators  at
> the digital level, or b) radio (zigbee  etc) ? Passing analogue
> readings through an opto can be done but it needs care.
>  Alternatively you could do a voltage-frequency conversion and use
> simple transformers.

I've actually thought about providing some basic inputs which are all
ground/Vss referenced and then building a separate device where this
won't work which includes some sort of isolation circuitry.

> You don't state the update rate so I'm guessing it will be slow.

Slow... once every few seconds...  Mainly these need to be able to
monitor the battery conditions at the site, plus things like water level
and the like.   They're more "condition sensing" than something which
needs quick updates.


> You may be able to group sensors to perform multiple tasks - e.g one
> group measuring negative referenced signals (e.g low voltage side
> battteries, temperature), one group measuring positive referenced
> signals (High voltage side batteries, current shunts).

As I continue to work though this, I'm almost convinced that I'm going
to have to shed the ability to measure things not referenced to
ground/vss.  It would make things a lot easier and turn this project
into something which requires a quad opamp and a few precision resistors.

The shunts are actually fairly easy...  There are several
high-common-mode-voltage amplifiers set up for this.  Unfortunately,
none will allow you to do more than a few volts of "signal"...  That is,
you can measure a +-5V voltage difference riding on up to +-100V of
common mode voltage.   Getting above the +-5V or so is fairly difficult.

It's when you start to say "It's a 0-24V signal, and I don't know what
it's going to be referenced to" which gets sticky.

-forrest

2008\08\03@191709 by Stephen R Phillips

picon face



--- On Sun, 8/3/08, Forrest W Christian <.....forrestcKILLspamspam@spam@imach.com> wrote:

{Quote hidden}

I suggest if you have a high side shunt resistor for measuring current you look at linear tech's line up.  Unfortunately NONE of there stuff is cheap. You stated cheap so you may need to shed the idea of the huge voltage range or measuring everything.  What I suggest is you first make something that works, then find out what you don't need and make it a lot cheaper.  How many units are intended in the final design? 1000's?

Stephen


     

2008\08\03@194021 by Richard Prosser

picon face
Forrest,

One common trick for measuring current from a high side shunt is to
convert the output voltage driver to a constant current source. Then
this can be returned to ground and a single resistor converts the
current back to  a voltage again. Provided the current driver can
withstand the maximum voltage you can level shift over a wide voltage
range.

RP

2008/8/4 Forrest W Christian <EraseMEforrestcspam_OUTspamTakeThisOuTimach.com>:
{Quote hidden}

> -

2008\08\03@200242 by Jinx

face picon face
> Think solar site where the instrument is powered from the same
> batteries that you are measuring voltage from

Can you do anything about that ?

2008\08\03@203148 by Apptech

face
flavicon
face
> Here's the problem:   The voltage sources may or may not
> be
> interconnected with each other and/or the voltage source
> for the
> telemetry instrument.  In addition, the voltage sources
> may be
> interconnected but at a higher reference voltage - I.E. a
> high side
> shunt or similar.

> In short, I need the functional equivalent of an
> instrumentation
> amplifier which can be adjusted to measure input voltages
> over a wide
> range (+-60V) and like I mentioned above may or may not be
> interconnected electrically with the amplifier's power
> supply.   I'd
> also like to do my best to make it "idiot-destruction"
> proof.

A reasonably simple instrumentation amplifier can have a
common mode range outside ground. This could use 2 or 3
sections of something as low cost as an eg LM324 or LM358
and possible meet your need.

If the input voltage sources are truly isolated from the
output they could assume very large voltage differences due
to eg electrostatic charging or stray sources. It may be
that you can "vaguely ground reference" them with eg a 1M
resistor from one leg to ground, thus allowing the
instrumentation amplifier approach to work with some safety.

For true floating inputs you can use eg analog optical
transmission using dual optocouplers custom made for this
purpose, or digitise at input and couple digital result
optically or with other isolated means. ADC conversion does
not have to use normal ADC means. It could be eg VCO,
voltage to pulse width etc. Depending on required accuracy,
something as 'horrible" as a 555 variant as a VCO or PWM
with optocoupler coupling could suffice.






          Russell McMahon

2008\08\03@210841 by Forrest W Christian

flavicon
face
Jinx wrote:
>> Think solar site where the instrument is powered from the same
>> batteries that you are measuring voltage from
>
> Can you do anything about that ?
>

No, unfortunately.

This will typically be used at remote solar-powered sites.   In fact,
almost every one of these will be powered by solar power and will be
expected to measure at least battery voltage, solar array voltage,
charging current and load current.  And, because the only source of
power is the battery array, this will be powered by the battery array.

If I was *only* monitoring the batteries, that would be easy, but the
other big piece of this is to monitor things like fuel tank levels,
generator operational status, signal strengths, etc. - all of which are
designed by different companies, and some make the assumption that
you'll be using a normal, isolated, voltmeter to do the measurements,
and as such do weird things like drive the signal below their ground, or
use a negative reference voltage, etc.

-forrest

2008\08\03@211006 by Antonio Benci

flavicon
picon face
This is a multi-part message in MIME format.
--------------080808020407070405020804
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Isolated current sensing, have a look at the following...

http://www.allegromicro.com/en/Products/Part_Numbers/0714/index.asp

Regards,

A. Benci

--------------080808020407070405020804
Content-Type: text/x-vcard; charset=utf-8;
name="Nino_Benci.vcf"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="Nino_Benci.vcf"

begin:vcard
fn:Antonio L. Benci
n:Benci;Antonio L.
org:Monash University;School of Physics
adr:;;BLD27 / RM118-119;Monash University;VIC;3800;Australia
email;internet:electronic.servicesspamspam_OUTsci.monash.edu.au
title:Professional Officer / E&IS Manager
tel;work:+613 9905 3649
tel;fax:+613 9905 3637
x-mozilla-html:FALSE
url:http://physics.monash.edu.au
version:2.1
end:vcard


--------------080808020407070405020804
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

2008\08\03@212407 by Forrest W Christian

flavicon
face
Stephen R Phillips wrote:
> I suggest if you have a high side shunt resistor for measuring current you look at linear tech's line up.  Unfortunately NONE of there stuff is cheap. You stated cheap so you may need to shed the idea of the huge voltage range or measuring everything.  What I suggest is you first make something that works, then find out what you don't need and make it a lot cheaper.  How many units are intended in the final design? 1000's?

Quantities are probably going to be in the 1000's/year range.  Perhaps
10K over the life of the device.  I'm at the stage in the design where
everything is done except the signal conditioning part, and the problem
is that if you look at the things which are likely to be hooked to this
device, they are varied and wide.

The shunt part is actually one of the easier parts to deal with - there
are lot's of options for high side monitoring, like the LT stuff you
mentioned.   Analog Devices and TI have some as well, along with I
believe Maxim.

I'm sort of leaning towards simply saying that all of the analog inputs
are ground referenced, and then providing some different ranges of
inputs, perhaps jumper selectable.   And then, for those applications
where I can't do a ground referenced input, do more exotic signal
conditioning in a separate device (with a separate power supply) - and
pass the additional cost on to those customers who need this (which are
common, but perhaps 5-10% of the total customer base).

What I had hoped was for some magic, inexpensive solution where I could
avoid issues with measuring everything in relation to ground, including
unexpected ground loops or similar.   So far, it seems like doing this
is going to be extremely expensive.

-forrest

2008\08\03@215339 by Apptech

face
flavicon
face
Hall effect

> Isolated current sensing, have a look at the following...
>
> http://www.allegromicro.com/en/Products/Part_Numbers/0714/index.asp


Or linear optocoupler
Claim 0.01% linearity
Avago HCNR201-500E
Price not marvellous, but cheaper available AND a good
enough emulation may be doable with OTS dual and quad optos.

   http://www.avagotech.com/assets/downloadDocument.do?id=1724

2008\08\03@222100 by Apptech

face
flavicon
face
> What I had hoped was for some magic, inexpensive solution
> where I could
> avoid issues with measuring everything in relation to
> ground, including
> unexpected ground loops or similar.

Several that I have already mentioned can be very cheap.
VCO or PWM to optical.
PWM can be converted back to analog at output with filter.
VCO can convert to analog with monostable.

Digikey shows  LTV84x quad optos down to under $US0.40 in 2k
quantity.
10 cents per opto.

       search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=LTV-847S-TA-ND
       http://media.digikey.com/pdf/Data%20Sheets/Lite-On%20PDFs/LTV-817_827_847(M,S,S_TA1).pdf

Use as PWM, VCO couplers OR in analog mode.

While there are no claims as to matching you MAY find these
are well enough matched per package to allow two sections to
be used in the analog feedback model used by fancy matched
parts. This WILL work at some accuracy with any optos. How
accurate depends on the parts.


In digital modes a cost of under $US0.50/channel seems
possible. Maybe less.



       Russell

2008\08\03@224504 by Antonio Benci

flavicon
picon face
This is a multi-part message in MIME format.
--------------010303010703020305050108
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

The Avago device is not a bad option BUT the Allegro 714 is ready to go, no
extra external semi's required. Sensor and signal processing is provided on
chip. All that is required is a resistor and a capacitor.

A. Benci

Apptech wrote:
{Quote hidden}

--------------010303010703020305050108
Content-Type: text/x-vcard; charset=utf-8;
name="Nino_Benci.vcf"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="Nino_Benci.vcf"

begin:vcard
fn:Antonio L. Benci
n:Benci;Antonio L.
org:Monash University;School of Physics
adr:;;BLD27 / RM118-119;Monash University;VIC;3800;Australia
email;internet:@spam@electronic.servicesKILLspamspamsci.monash.edu.au
title:Professional Officer / E&IS Manager
tel;work:+613 9905 3649
tel;fax:+613 9905 3637
x-mozilla-html:FALSE
url:http://physics.monash.edu.au
version:2.1
end:vcard


--------------010303010703020305050108
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

2008\08\03@224715 by Stephen R Phillips

picon face



--- On Sun, 8/3/08, Forrest W Christian <KILLspamforrestcKILLspamspamimach.com> wrote:

{Quote hidden}

4 - 20ma current source through a 250 ohm resistor isn't good?
You can supply a 12V power source. Since it's not fast this is a standard interface. If your voltage reference is 2.5V then 120ohm 1% or 0.1% resister is good. I made a system that measured 6 J type thermocouple inputs (using an AD594 which gave open sensor output as well) and 6 4-20ma inputs using a 12 bit ADC and a bunch of muxs doing that.  It worked just fine and was very simple.  4-20ma inputs are useful and very standard although it may limit your range a bit more you don't seem worried about that.


     

2008\08\03@224800 by Forrest W Christian

flavicon
face
Don't these all require a separately-derived power supply on the
measurement side?

That would drive the price up...   Which is kinda why I'm thinking at
this point that perhaps adding this as an "option" might be better.

Unless you have a suggestion for how to provide an appropriate power
source on the measurement side....which would be inexpensive to implement.

-forrest

Apptech wrote:
{Quote hidden}

2008\08\03@230604 by Jinx

face picon face
>> Think solar site where the instrument is powered from the same
>> batteries that you are measuring voltage from
> >
> > Can you do anything about that ?
> >
>
> No, unfortunately.
>
> because the only source of power is the battery array, this will be
> powered by the battery array

What I was thinking was for the PIC, powered by the solar array, to
have a charged cap for its Vcc and disconnect itself from the solar
array power

For example a normally-open relay (or FETs), passing Vcc and Vss,
held closed by the PIC when not measuring, and released to measure,
isolating the PIC completely from everything. Temporary connections
to lines to be measured could be via FETs controlled by the PIC. With
the PIC electrically isolated it could take differential readings across
shunts etc

Or something along those lines

An alternative would be V-to-F conversion using a 555 or LM2917,
opto-coupled to the PIC

2008\08\03@232115 by Bob Blick

face
flavicon
face
Forrest W Christian wrote:
> Don't these all require a separately-derived power supply on the
> measurement side?

Not if the voltage is about a volt or more(an IR LED drop). Opto #1 LED
sees the measured side. Basically the output of the two optos fight it
out for the opamp's input - the opamp drives opto #2 LED.

There are examples on the web, many from opto makers that supply matched
pairs. Russell was suggesting doing your own matching, I think.

Cheerful regards,

Bob

2008\08\03@234958 by Forrest W Christian

flavicon
face
Bob Blick wrote:
> Not if the voltage is about a volt or more(an IR LED drop). Opto #1
> LED sees the measured side. Basically the output of the two optos
> fight it out for the opamp's input - the opamp drives opto #2 LED.

Ok, my brain hurts... but I think I've got my brain twisted around what
you two are trying to explain to me.

All of the linear opto schematics I had seen had a single LED driving
two internal photo transistors.   On the input side of these schematics,
there was typically an opamp with a separate power supply which
basically completed the feedback loop...   That is, the opamp's feedback
loop went through the LED and one of the transistors so that the LED was
turned on enough that the phototransistor was turned on enough that the
voltage out of the "input side transistor" matched the input voltage.
This of course needs a power supply for the opamp on the input side.
On the output side transistor there was typically a unity gain opamp set
up just as a buffer.  The theory being that both transistors will be
turned on the same amount so once you "balanced" the input side, the
output side would also be producing the same voltage.

What I'm hearing you say (or at least I think I am hearing you say) is
to take a more-traditional dual optocoupler (perhaps matched) and feed
the signal of interest into one of the LEDs.  Then effectively wire both
phototransistor outs into another opamp which then drives the second LED
in the optocoupler.  The theory being that both sides are going to react
the same at the same current level through the LED, so that when the
opamp is stable, the voltage out of the opamp (and into the second LED)
on the output side is going to pretty much match the input signal level.

Is this more or less correct?

2008\08\04@001300 by Apptech

face
flavicon
face
>> Don't these all require a separately-derived power supply
>> on the
>> measurement side?

> Not if the voltage is about a volt or more(an IR LED
> drop). Opto #1 LED
> sees the measured side. Basically the output of the two
> optos fight it
> out for the opamp's input - the opamp drives opto #2 LED.

> There are examples on the web, many from opto makers that
> supply matched
> pairs. Russell was suggesting doing your own matching, I
> think.

I suggested both purpose-built parts and the possibility of
"roll your own" with multi unit COTS optos - the latter will
be rather cheaper than purpose built parts.

Even with a LED only input you'd probably want some sort of
buffer.

An isolated supply can be made reasonably easily. eg

- A two winding coil with suitably high frequency AC feed
and rectification.

- Or even two capacitors passing AC.

- Extra points :-). A photovoltaic panel and optical power
input - LED or bulb.
Lessee. White LED 50 l/W, 20 mA.
About 66 mW in. 50 l/W ~= 15% light out = say 10 mW light.
Couple half this to panel and panel = 5% (amorphous) = 0.5
mW. Maybe 1 to 2 mW with good coupling, good panel etc.
Enough to do the job with care. Some calculators have small
multi-cell panels that may do. Vout ~= cells/2 max.

As the main requirement for isolation is to float supplies
that are not formally connected the isolation system
(probably) need not have an overly capable isolation spec.
Of the above options the two winding coil would probably be
the easiest, possibly the cheapest and probably the highest
power transfer per $. PV panel method is the most elegant
[tm] but probably dearest.



       Russell


2008\08\04@001401 by Bob Blick

face
flavicon
face
Yes, exactly! It's very old-school, but if you can work within all the
limitations it isn't bad at all.

Cheerful regards,

Bob

Forrest W Christian wrote:
{Quote hidden}

2008\08\04@043655 by Alan B. Pearce

face picon face
> Analog devices has some interesting isolation amplifiers that
> have the isolated DC to DC converter built in.

TI also have the Burr-Brown originated isolated amplifiers. See
http://focus.ti.com/paramsearch/docs/parametricsearch.tsp?family=analog&familyId=512&uiTemplateId=NODE_STRY_PGE_T&paramCriteria=no
for details (watch long URL).

2008\08\04@044902 by Alan B. Pearce

face picon face
>I suggest if you have a high side shunt resistor for measuring current
>you look at linear tech's line up.  Unfortunately NONE of there stuff is
>cheap.

Analog Devices, Zetex and Supertec also make high side current monitors.

www.analog.com/en/amplifiers-and-comparators/current-sense-amplifiers/products/index.html
http://www.zetex.com/3.0/3-3-2a.asp?scatid=16
http://www.supertex.com/pdf/misc/2K8_SFC.pdf (Look for HV78xx series on page
12 - these will work at mains voltages).

2008\08\04@045557 by Alan B. Pearce

face picon face
>What I had hoped was for some magic, inexpensive solution where I could
>avoid issues with measuring everything in relation to ground, including
>unexpected ground loops or similar.   So far, it seems like doing this
>is going to be extremely expensive.

You might like to look at the Avago Liner Optoisolators, like the HCPL-7800.
This is designed for giving isolated analog feedback of motor currents, so
the input voltage required is around 0.5V. The input side amplifier (and the
Output as well IIRC run off 5V.

http://www.avagotech.com/search/results.jsp?src=&siteCriteria=hcpl-7800&searchButton.x=8&searchButton.y=10

2008\08\04@050101 by Alan B. Pearce

face picon face
>All of the linear opto schematics I had seen had a single LED driving
>two internal photo transistors.   On the input side of these schematics,
>there was typically an opamp with a separate power supply which
>basically completed the feedback loop...

The link to the Avago device that I posted does all this for you inside the
chip. The device Russell posted is the bare bones device, and one would need
to do all this yourself.

2008\08\04@082319 by Enki

picon face
On 3 Aug 2008 at 20:49, Forrest W Christian wrote:

>
> Unless you have a suggestion for how to provide an appropriate power
> source on the measurement side....which would be inexpensive to
> implement.
>
> -forrest
>

       A simple 34063 SMPS with a dozen separated outputs?

       Mark Jordan


2008\08\05@000141 by Rich

picon face
I have not followed this thread but the title caught my eye.  One very
important thing to remember when driving the input to  an A/D is that the
output impedance of the driver should be kept as low as possible in order to
avoid aliasing and conversion error.

{Original Message removed}

2008\08\05@080139 by olin piclist

face picon face
Rich wrote:
> I have not followed this thread but the title caught my eye.  One very
> important thing to remember when driving the input to  an A/D is that
> the output impedance of the driver should be kept as low as possible
> in order to avoid aliasing and conversion error.

Conversion error, yes.  But A/D source voltage impedence has nothing to do
with aliasing.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2008\08\05@093803 by Rich

picon face
I disagree!  Think it through.  What happens if the source impedance is to
high and cannot source the required current or even sinks current from the
converter?  Analog Devices has some good articles on this.  True, they are
older articles, but they can still apply depending on the choice of devices
used.


{Original Message removed}

2008\08\05@111210 by Michael Rigby-Jones

picon face


> -----Original Message-----
> From: TakeThisOuTpiclist-bouncesEraseMEspamspam_OUTmit.edu [RemoveMEpiclist-bouncesspamTakeThisOuTmit.edu] On
Behalf
> Of Rich
> Sent: 05 August 2008 14:37
> > Rich wrote:
> >> I have not followed this thread but the title caught my eye.  One
very
> >> important thing to remember when driving the input to  an A/D is
that
> >> the output impedance of the driver should be kept as low as
possible
> >> in order to avoid aliasing and conversion error.



> From: "Olin Lathrop" <olin_piclistEraseMEspam.....embedinc.com>
> >
> > Conversion error, yes.  But A/D source voltage impedence has nothing
to
> > do with aliasing.


> From: EraseMEpiclist-bouncesspammit.edu [RemoveMEpiclist-bouncesEraseMEspamEraseMEmit.edu] On
Behalf
> Of Rich
>
> I disagree!  Think it through.  What happens if the source impedance
is to
> high and cannot source the required current or even sinks current from
the
> converter?  Analog Devices has some good articles on this.  True, they
are
> older articles, but they can still apply depending on the choice of
> devices used.

If the source impedance is too high, then it simply reduces the accuracy
of the converted measurement.  It does not cause aliasing which is an
artefact of sampling below the Nyquist frequency.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2008\08\05@115259 by Forrest Christian

flavicon
face
Aliasing has to do with sampling a high-frequency signal at a too-low
rate..    Generally you need to sample at two times the highest
frequency component of the input signal.   For instance, if you sample a
10khz signal at 5khz, you will not have collected enough data to
determine the frequency of the signal..  In fact, a 10khz signal sampled
at exactly 5khz will appear to be DC - since you are sampling it at the
same point in the cycle and only getting the one sample.   Instead, you
need to sample at at least 20khz so you get a representative sample of
the input signal.  

About the only function that the input circuitry can have on aliasing is
whether or not it blocks or creates a high-frequency signal.   In the
DSP world it is pretty common to add a low-pass filter which blocks at
least everything over 50% of the sample rate to avoid aliasing.

I can see how a too-high impedance can create a symptom that *looks*
like aliasing, but is not.   Aliasing has to do with the signal being
accurately delivered to the ADC, and then the interaction between the
ADC's sample rate and the input signal causes "aliased" frequencies
below the too-high frequency.   About the only thing that a too-high
impedance on an ADC will cause is for the reading to be off (sampling
error) - which may cause a signal with artifacts related to the sampling
error, but won't be aliasing in the true meaning of the word.

-forrest

Rich wrote:
> I disagree!  Think it through.  What happens if the source impedance is to
> high and cannot source the required current or even sinks current from the
> converter?  Analog Devices has some good articles on this.  True, they are
> older articles, but they can still apply depending on the choice of devices
> used.
>
>
> {Original Message removed}

2008\08\05@120856 by olin piclist

face picon face
Rich wrote:
>>> I have not followed this thread but the title caught my eye.  One
>>> very important thing to remember when driving the input to  an A/D
>>> is that the output impedance of the driver should be kept as low as
>>> possible in order to avoid aliasing and conversion error.
>>
>> Conversion error, yes.  But A/D source voltage impedence has nothing
>> to do with aliasing.
>
> I disagree!  Think it through.

I have.  Do you understand what aliasing means?

> What happens if the source impedance
> is to high and cannot source the required current or even sinks
> current from the converter?

Then you get a offset on the conversion.  The A/D pin current times the
source impedence is the offset voltage.  However this has nothing to do with
aliasing.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2008\08\05@195623 by Rich

picon face
Yes, of course of course I do.


----- Original Message -----
From: "Olin Lathrop" <RemoveMEolin_piclistspam_OUTspamKILLspamembedinc.com>
To: "Microcontroller discussion list - Public." <RemoveMEpiclistTakeThisOuTspamspammit.edu>
Sent: Tuesday, August 05, 2008 12:10 PM
Subject: Re: [EE] ADC Input Buffer Circuit


{Quote hidden}

> --

2008\08\05@205455 by Jinx

face picon face
Rich, I don't recall Microchip using the term "aliasing" and can't see it
in any ADC documentation. I always understood it to be applicable to
repetitive sampling of AC signals

Perhaps you meant "quantisation" (jitter) ? Although that too has nothing
to do with source impedance. Not directly anyway

2008\08\05@222402 by Rich

picon face
Aliasing is related to the nyquist sampling.  It may not be related to the
driving impedance.  My thoughts were that the error could appear as
redundant codes, that is the same code for more than one voltage or current
input, due to the sourcing of current from the converter to the driver.  I
have actually seen such errors from improper driving circuits.  It has
happened that when some op amp is used to drive the A/D converter errors
occur because the op amp was not properly matched to the converter.  I
apologize if I used the term aliasing incorrectly.  But the fundamental idea
of driving the A/D with a low impedance is a good rule to keep in mind.  Of
course, all components are not the same so it is necessary to review all of
the specifications for each.


{Original Message removed}

2008\08\06@012610 by Rich

picon face
Yes, Thank you for the clarification.  I did misrepresent my point.  I was
thinking that the appearance of aliasing could occur because the sampling
rate, of course, is not affected by the impedance factors.  It is good that
you have clarified it so that no one is actually misled.  I have
experimented with A/D converters in order to design a test fixture for
evaluation.  I was speaking from that experience.

{Original Message removed}

2008\08\06@071430 by olin piclist

face picon face
Rich wrote:
> Aliasing is related to the nyquist sampling.  It may not be related
> to the driving impedance.

Try "is not" instead of "may not be".

********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.

2008\08\06@080822 by Apptech

face
flavicon
face
>> Aliasing is related to the nyquist sampling.  It may not
>> be related
>> to the driving impedance.

> Try "is not" instead of "may not be".

To be PIC-y, in the limit, increasing driving impedance will
probably improve the aliasing performance, albeit very
slightly in most cases.


       Russell




More... (looser matching)
- Last day of these posts
- In 2008 , 2009 only
- Today
- New search...