Searching \ for '"isolated" analog input' 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=analog
Search entire site for: '"isolated" analog input'.

Truncated match.
PICList Thread
'"isolated" analog input'
2007\03\26@212119 by Forrest W. Christian

flavicon
face
I'm currently working on a remote telemetry project.   One of the
requirements is the ability to measure various DC (and possibly AC)
voltages.   These may be above or below ground, ground referenced, or
not.   I do have some flexibility with specifying what inputs are
permitted (voltage), but don't have enough flexibility to say that "all
signals must be ground referenced and must not be negative in reference
to ground".  (Actually I might be able to do this, but I'm not going to
be happy about it long term).

Oh yeah, I'm not thrilled about the prospect of needing a negative
(below ground) power supply to do this since this is a cost sensitive
application.  But if I must, I do realize that I could do this with some
sort of level shifting opamp - but that isn't really the way I'd like to
do this.

External adc's are not out of the question if needed...

What I'd love is a dirt cheap completely isolated adc, but I realize
that this isn't going to happen.

Ideas?

-forrest

2007\03\26@214518 by Richard Prosser

picon face
Forrest,

Voltage & Current Requirements?? Isolation Voltage?

Is complete Isolation required, or could an AC coupled charge pump
type circuit be sufficient? (i.e. DC isolation only)

RP

On 27/03/07, Forrest W. Christian <spam_OUTforrestcTakeThisOuTspamimach.com> wrote:
{Quote hidden}

> -

2007\03\26@215115 by Cristóvão Dalla Costa

picon face
On 3/26/07, Forrest W. Christian <.....forrestcKILLspamspam@spam@imach.com> wrote:
>
> I'm currently working on a remote telemetry project.   One of the
> requirements is the ability to measure various DC (and possibly AC)
> voltages.   These may be above or below ground, ground referenced, or
> not.   I do have some flexibility with specifying what inputs are
> permitted (voltage), but don't have enough flexibility to say that "all
> signals must be ground referenced and must not be negative in reference
> to ground".  (Actually I might be able to do this, but I'm not going to
> be happy about it long term).



You should probably use instrumentation amplifiers with weakly grounded
inputs. Basically you'll use a resistor network to bring down the voltage to
levels readable by the amplifiers then, for each input, both the input
ground and the input signal will be connected to your circuit's ground via a
1 MB resistor or similar. The input ground goes to the opamp's inverting
input and the signal to the non-inverting. The opamp will subtract the
common mode and leave you the signal.

Ideally the opamps ground should convert to 50% of the ADC's full scale so
that it can swing both over and below ground. Make sure the full scale
readable swing is at least 4 X the highest input voltage you want to
measure. You should probably use a 16 bit ADC if you want good resolution.

Use diodes to clamp inputs, high resistors in series to limit current, etc.
HTH.

2007\03\26@215656 by Martin Klingensmith

picon face
Forrest,
The typical isolation schemes apply such as pulse transformers and
opto-couplers. The HCPL-3120 is an opto-coupler that has one emitter and two
photo transistors. One is used on the sending side as feedback to compensate
for non-linearity (which WILL be present in any "linear" device) It will
have significant latency no matter how you do it. Other methods involve
digital coupling using voltage to frequency converters and optocouplers or
pulse transformers.
--
Martin K

On 3/26/07, Forrest W. Christian <forrestcspamKILLspamimach.com> wrote:
{Quote hidden}

> -

2007\03\27@000236 by Rich

picon face
For analog isolation a very effective method is V/F (voltage to frequency)
and F/V (frequency to voltage) translation.  There are some very good
components with less 1% nonlinearity.  Sometime you can use an optocoupler
that directly interfaces with a D/A converter hat you can scale for the
range and span you desire.  These, I believe are the most effective and
linear.  I have achieved a Pearson Product Moment of 0.99999 with these
techniques.
   If you would like assistance implementing either, contact me off list
and I will be happy to assist you.



{Original Message removed}

2007\03\27@002124 by Vasile Surducan

face picon face
It's weird you've thinking to isolate the measured signals instead of
isolating the communication... which means the same.
If 10 bit is enough, then one PIC12F675 for every measured voltage,
floated AT the voltage level. Communication from PIC to the rest of
the world optoisolated with max 2 optocouplers. Very cheap and not so
dirty.

On 3/26/07, Forrest W. Christian <.....forrestcKILLspamspam.....imach.com> wrote:
{Quote hidden}

> -

2007\03\27@002535 by Vasile Surducan

face picon face
I forgot. If you need more resolution (14 to 21 bits) and high input impedance,
use a serial ADC, put it in the instrumentation amplifier loop (yes,
again let the ADC to float at the input signal voltage or on other
levels) and optoisolate the serial data and commands to/from the rest
of the world. It's the most cost effective methode.

On 3/26/07, Vasile Surducan <EraseMEpiclist9spam_OUTspamTakeThisOuTgmail.com> wrote:
{Quote hidden}

2007\03\27@022439 by Forrest W Christian

flavicon
face
Richard Prosser wrote:

>Forrest,
>
>Voltage & Current Requirements?? Isolation Voltage?
>
>Is complete Isolation required, or could an AC coupled charge pump
>type circuit be sufficient? (i.e. DC isolation only)
>
The main processor on these products are from the PIC18fxxJ60.   Mainly
we're measuring things like battery charge currents (which can be + or
-, and may be referenced either from +BAT (up to 60V) or Ground
(typically the battery supply and the telemetry supply are tied together
in this application, but not always - and the voltage out of the shunt
may be positive or negative), Per-battery voltages, etc.   Plus, other
instruments which may vary from site to site.

I don't 100% need complete isolation.  I.E. completely separate
optoisolated inputs for each analog input is overkill.  I just need
something which would work under typical situations.   For that matter,
if I *had* to say everything is ground/Vss referenced I could probably
live with that (but would prefer not to).

Inputs are going to be around +-60V at the high end.  I will need to
work on scaling so that I provide a range of inputs as needed.

-forrest

2007\03\27@023025 by Forrest W Christian

flavicon
face
Vasile Surducan wrote:

>It's weird you've thinking to isolate the measured signals instead of
>isolating the communication... which means the same.
>If 10 bit is enough, then one PIC12F675 for every measured voltage,
>floated AT the voltage level. Communication from PIC to the rest of
>the world optoisolated with max 2 optocouplers. Very cheap and not so
>dirty.
>
The question I would have in this case is how do I end up with a
separate VCC/VSS for each of these devices.  It would seem to me that
deriving a separate power supply for each of these pics would get very
expensive fairly quickly.

-forrest

2007\03\27@025416 by Forrest W Christian

flavicon
face
I realized I should probably better define what I'm looking for, so I
can get more useful answers.  I guess this is one area where I can't
just say "how do I isolate an analog input".   Not to say that the
answers so far haven't be useful, as they have pointed me in some
directions I haven't thought about - for instance both the V/F and the
Instrument amplifier are good ideas which I will delve into further.

That said, what I really need to be able to do is have a half dozen or
so "voltmeters" which are attached to the PIC18F*J60.  Cost is
definately an object in this case.    Although I will need to deal with
different scales so that a 2V signal can be read with enough resolution,
on the high end I need to measure up to about 60VDC.  The input can
swing either way.  These will be used to measure things like battery
voltage level (typically the *same* array that the unit is powered from,
but not always), charge/discharge current measured as shunt voltage (may
be on the + or - side of the bat), and other inputs such as fuel level,
valve positions, etc.  

In many cases I could say "everything is measured in relation to
ground/Vss", and this wouldn't pose a problem.  In others, this would be
an issue.   In an ideal world, the inputs would be completely isolated
from each other and from ground/Vss.  That is, work just like multiple
battery powered voltmeters.   I realize that this may be unworkable or
significantly expensive.   As a result, the needs may be able to be
redefined such that everything must be ground referenced.  However,
being able to measure voltages where the input is below Vss is important
since one of the main applications for this device is to monitor battery
utilization.

Hopefully this rambling helps describe what I'm trying to figure out.  
I'm at the "we need to be able to measure x, and what are our options to
do so" stage so I really don't want to through any ideas out at this
point, even if they don't fully meet my ideal situation.

-forrest






2007\03\27@045650 by Forrest W Christian

flavicon
face
I hate replying to myself :)... but...

Forrest W Christian wrote:

>As a result, the needs may be able to be redefined such that everything must be ground referenced.  However, being able to measure voltages where the input is below Vss is important  since one of the main applications for this device is to monitor battery utilization.
>  
>
I do realize that this specific case (I.E. force everything to be
"ground" referenced, and allow +- voltage input) is easily accomplished
with an opamp level shifting the voltage upward, and scaling accordingly
(So +-5v maps into say 0-3.3V, where 0V becomes something like 1.65V).  
Of course more novel and less expensive solutions to this case are welcome.

(And of course, ideally I would love to not have to ground reference
anything).

-forrest

2007\03\27@080557 by rlistas

picon face
Another simple solution:

Use the hall effect current sensors from Allegro and
convert voltage to current, and read in the PIC A/D.

The sensor are completely isolated from the signal.


Best regards,

Rubens


At 05:58 27/3/2007, you wrote:
{Quote hidden}

>-

2007\03\27@124156 by Vasile Surducan

face picon face
On 3/26/07, Forrest W Christian <@spam@forrestcKILLspamspamimach.com> wrote:
> Vasile Surducan wrote:
>
> >It's weird you've thinking to isolate the measured signals instead of
> >isolating the communication... which means the same.
> >If 10 bit is enough, then one PIC12F675 for every measured voltage,
> >floated AT the voltage level. Communication from PIC to the rest of
> >the world optoisolated with max 2 optocouplers. Very cheap and not so
> >dirty.
> >
> The question I would have in this case is how do I end up with a
> separate VCC/VSS for each of these devices.  It would seem to me that
> deriving a separate power supply for each of these pics would get very
> expensive fairly quickly.

Yes, if you buy the DC-DC converters instead of designing those.
Think creative. One converter with n voltage outputs will supply all modules.

2007\03\27@125450 by Vasile Surducan

face picon face
On 3/26/07, Forrest W Christian <KILLspamforrestcKILLspamspamimach.com> wrote:
> I realized I should probably better define what I'm looking for, so I
> can get more useful answers.  I guess this is one area where I can't
> just say "how do I isolate an analog input".   Not to say that the
> answers so far haven't be useful, as they have pointed me in some
> directions I haven't thought about - for instance both the V/F and the
> Instrument amplifier are good ideas which I will delve into further.

Search the arhive for optoisolated amplifier. I've posted there the schematic.
You can't design a performance tool without spending some money.
Please understand that. I will forget to use a 10 bit ADC for such
universal purpose like yours where resolution could be an issue.

Vasile

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...