galvanic isolation of sensors

Hi all,

So, I want to put 10 pH/ORP probes onto a microcontroller, with each
probe being galvanically isolated from its peers so that the
measurements of different probes aren't interfering with each other.
Being more of a digital than an analogue guy, I'm not really sure how
to proceed...

Another requirement is that it's a small solution - there's not much
space to play with here, about 0.5" x 1.5" per probe PCB... And of
course, the cheaper the better, given that I want 10 of them!

One chip I found to do the isolation is the adum5241 (http://
http://www.analog.com/en/interface/d...s/product.html)
from analog devices, I could get an isolated GND, +5v, dIn and dOut
from that, which is what I need, but I also want -5v for the probe,
and I'm not sure how to go about getting it. If I try to tack on a +5v-
>-5v conversion after the adum5241, I think I'll over-run its maximum
current delivery (10mA).

So, questions:

- Do I need to isolate the -5v if the +5v and ground are isolated ?
I'd think so, but better to ask

- Is there a better way to do it ? Bound to be - this is just the
first one I've come across

Cheers
Simon.

On Thu, 12 Jan 2012 08:25:57 -0800, Simon wrote:

> Hi all,
>
> So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> probe being galvanically isolated from its peers so that the
> measurements of different probes aren't interfering with each other.
> Being more of a digital than an analogue guy, I'm not really sure how to
> proceed...
>
> Another requirement is that it's a small solution - there's not much
> space to play with here, about 0.5" x 1.5" per probe PCB... And of
> course, the cheaper the better, given that I want 10 of them!
>
> One chip I found to do the isolation is the adum5241 (http://
> http://www.analog.com/en/interface/d...5241/products/
product.html)
> from analog devices, I could get an isolated GND, +5v, dIn and dOut from
> that, which is what I need, but I also want -5v for the probe, and I'm
> not sure how to go about getting it. If I try to tack on a +5v-
>>-5v conversion after the adum5241, I think I'll over-run its maximum
> current delivery (10mA).
>
> So, questions:
>
> - Do I need to isolate the -5v if the +5v and ground are isolated ?
> I'd think so, but better to ask
>
> - Is there a better way to do it ? Bound to be - this is just the
> first one I've come across

I haven't had to select something like this in anger, so I can't tell you
the Very Best Thing.

Alternatives that I can think of are:

* Go looking for a higher power version of the same chip.
* Use two, one for +5V and one for -5V (assuming that 10mA
from each is enough)
* Use one of the data-only galvanic isolation chips,
(they're out there -- look) and do power separately.
* If you can find a teeny tiny transformer, rolling your
own power supply might be small enough, particularly
if you don't care too much about regulation.

--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com

On Jan 12, 8:25*am, Simon <goo...@gornall.net> wrote:
> Hi all,
>
> So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> probe being galvanically isolated from its peers so that the
> measurements of different probes aren't interfering with each other.
> Being more of a digital than an analogue guy, I'm not really sure how
> to proceed...
>
> Another requirement is that it's a small solution - there's not much
> space to play with here, about 0.5" x 1.5" per probe PCB... And of
> course, the cheaper the better, given that I want 10 of them!
>
> One chip I found to do the isolation is the adum5241 (http://www.analog.com/en/interface/d...products/produ...)
> from analog devices, I could get an isolated GND, +5v, dIn and dOut
> from that, which is what I need, but I also want -5v for the probe,
> and I'm not sure how to go about getting it. If I try to tack on a +5v->-5v conversion after the adum5241, I think I'll over-run its maximum
>
> current delivery (10mA).
>
> So, questions:
>
> *- Do I need to isolate the -5v if the +5v and ground are isolated ?
> I'd think so, but better to ask
>
> *- Is there a better way to do it ? Bound to be - this is just the
> first one I've come across
>
> Cheers
> * *Simon.

I'm not sure what these "ORP" probes are - if they're anything like
the pH
electrodes I've used, they are _extremely_ high impedance; and you
need
to compare their output voltage with the potential of the solution
they are
measuring. Why 10 probes? Might the solution[s] be at different
potentials,
and if so, by how much?

IF the solution[s] are at nearly the same potential, you may be able
to get
away with some good differential amplifiers and dispense with
isolation -
assuming, of course, that you aren't concerned with galvanically
isolating the
solution from the output. But that might be doable with one or two
digital
isolation devices.

On Thu, 12 Jan 2012 08:25:57 -0800 (PST), Simon <>
wrote:

>Hi all,
>
>So, I want to put 10 pH/ORP probes onto a microcontroller, with each
>probe being galvanically isolated from its peers so that the
>measurements of different probes aren't interfering with each other.
>Being more of a digital than an analogue guy, I'm not really sure how
>to proceed...
>
>Another requirement is that it's a small solution - there's not much
>space to play with here, about 0.5" x 1.5" per probe PCB... And of
>course, the cheaper the better, given that I want 10 of them!
>
>One chip I found to do the isolation is the adum5241 (http://
>http://www.analog.com/en/interface/d...s/product.html)
>from analog devices, I could get an isolated GND, +5v, dIn and dOut
>from that, which is what I need, but I also want -5v for the probe,
>and I'm not sure how to go about getting it. If I try to tack on a +5v-
>>-5v conversion after the adum5241, I think I'll over-run its maximum
>current delivery (10mA).

You can use an ICL7660 charge pump to get -5V.

The whole galvanic isolation thing can be a bit tricky, especially
when it comes to hundreds-of-megohm source impedance millivolt-level
signals so I'm not going to say that this will work for you. Suggest
you pay attention to the asymmetrical source impedance of the ORP/Ph
electrodes and design accordingly..

On Jan 12, 9:45*am, cassiope <f...@u.washington.edu> wrote:
> On Jan 12, 8:25*am, Simon <goo...@gornall.net> wrote:
>
>
>
>
>
> > Hi all,
>
> > So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> > probe being galvanically isolated from its peers so that the
> > measurements of different probes aren't interfering with each other.
> > Being more of a digital than an analogue guy, I'm not really sure how
> > to proceed...
>
> > Another requirement is that it's a small solution - there's not much
> > space to play with here, about 0.5" x 1.5" per probe PCB... And of
> > course, the cheaper the better, given that I want 10 of them!
>
> > One chip I found to do the isolation is the adum5241 (http://www.analog..com/en/interface/...um5241/product......)
> > from analog devices, I could get an isolated GND, +5v, dIn and dOut
> > from that, which is what I need, but I also want -5v for the probe,
> > and I'm not sure how to go about getting it. If I try to tack on a +5v->-5v conversion after the adum5241, I think I'll over-run its maximum
>
> > current delivery (10mA).
>
> > So, questions:
>
> > *- Do I need to isolate the -5v if the +5v and ground are isolated ?
> > I'd think so, but better to ask
>
> > *- Is there a better way to do it ? Bound to be - this is just the
> > first one I've come across
>
> > Cheers
> > * *Simon.
>
> I'm not sure what these "ORP" probes are - if they're anything like
> the pH
> electrodes I've used, they are _extremely_ high impedance; and you
> need
> to compare their output voltage with the potential of the solution
> they are
> measuring. *Why 10 probes? *Might the solution[s] be at different
> potentials,
> and if so, by how much?

ORP is a similar probe, yes - it's an Oxidation/Reduction potential
probe.

The pH solutions are likely to be similar (within +/- 1 pH at most)
and it's for measuring / controlling the pH of an aquarium when you're
introducing chemicals at various places in the water column.

>
> IF the solution[s] are at nearly the same potential, you may be able
> to get
> away with some good differential amplifiers and dispense with
> isolation -
> assuming, of course, that you aren't concerned with galvanically
> isolating the
> solution from the output. *But that might be doable with one or two
> digital
> isolation devices.

I doubt there'd be much correlation in the readings between an ORP
probe and a pH probe, though, and I wouldn't want the addition of one
to affect existing readings. The same system will be reading
temperature (since pH is temperature-dependent) although that's of
lesser concern due to the nature of the probe.


On Jan 12, 11:46*am, Spehro Pefhany
<speffS...@interlogDOTyou.knowwhat> wrote:

>
> You can use an ICL7660 charge pump to get -5V.
>

That's one I hadn't come across, and for the my needs, it would appear
to produce a good -5v (the required load should be very low due to the
massive impedance of the probe). Thanks!


On Jan 12, 9:42*am, Tim Wescott <t...@seemywebsite.com> wrote:

>
> * Go looking for a higher power version of the same chip.
>

And, of course, there is indeed one such - in fact it handles 0.5A.
This still wouldn't be sufficient for the original 5v -> -5v converter
chip (which can need 1->2A (!!) on startup), but the ICL chip doesn't
seem to suffer from that.

Combining the two suggestions, seems like that would work

Thanks for all the help

Simon

"Simon" <> wrote in message
news:70d01634-9eb6-4ced-90ac-...
> Hi all,
>
> So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> probe being galvanically isolated from its peers so that the
> measurements of different probes aren't interfering with each other.
> Being more of a digital than an analogue guy, I'm not really sure how
> to proceed...
>
> Another requirement is that it's a small solution - there's not much
> space to play with here, about 0.5" x 1.5" per probe PCB... And of
> course, the cheaper the better, given that I want 10 of them!
>
> One chip I found to do the isolation is the adum5241 (http://
> http://www.analog.com/en/interface/d...s/product.html)
> from analog devices, I could get an isolated GND, +5v, dIn and dOut
> from that, which is what I need, but I also want -5v for the probe,
> and I'm not sure how to go about getting it. If I try to tack on a +5v-
>>-5v conversion after the adum5241, I think I'll over-run its maximum
> current delivery (10mA).
>
> So, questions:
>
> - Do I need to isolate the -5v if the +5v and ground are isolated ?
> I'd think so, but better to ask
>
> - Is there a better way to do it ? Bound to be - this is just the
> first one I've come across
>
> Cheers
> Simon.

I can see all sorts of potential 'gotchas' in this scenario. With Hi-Z
inputs, things like transients from switching power supply devices like the
7660 could cause problems unless layout and packaging is just right. As
other posts have suggested, things are a lot simpler if you can avoid
isolation of the inputs, is the CMV really so high that you need it? If you
don't, then you can use a single power supply that can be well packaged to
avoid any noise or spikes if it incorporates switching. In fact, I've always
found that for audio (similar need for power rail cleanliness), a simple
rectifier/filter/linear reg arrangement is hard to beat.

On Jan 13, 9:17*pm, "Bruce Varley" <b...@NoSpam.com> wrote:
> "Simon" <goo...@gornall.net> wrote in message
>
> news:70d01634-9eb6-4ced-90ac-...
>
>
>
>
>
>
>
>
>
> > Hi all,
>
> > So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> > probe being galvanically isolated from its peers so that the
> > measurements of different probes aren't interfering with each other.
> > Being more of a digital than an analogue guy, I'm not really sure how
> > to proceed...
>
> > Another requirement is that it's a small solution - there's not much
> > space to play with here, about 0.5" x 1.5" per probe PCB... And of
> > course, the cheaper the better, given that I want 10 of them!
>
> > One chip I found to do the isolation is the adum5241 (http://
> >http://www.analog.com/en/interface/d...products/produ...)
> > from analog devices, I could get an isolated GND, +5v, dIn and dOut
> > from that, which is what I need, but I also want -5v for the probe,
> > and I'm not sure how to go about getting it. If I try to tack on a +5v-
> >>-5v conversion after the adum5241, I think I'll over-run its maximum
> > current delivery (10mA).
>
> > So, questions:
>
> > - Do I need to isolate the -5v if the +5v and ground are isolated ?
> > I'd think so, but better to ask
>
> > - Is there a better way to do it ? Bound to be - this is just the
> > first one I've come across
>
> > Cheers
> > * Simon.
>
> I can see all sorts of potential 'gotchas' in this scenario. With Hi-Z
> inputs, things like transients from switching power supply devices like the
> 7660 could cause problems unless layout and packaging is just right. As
> other posts have suggested, things are a lot simpler if you can avoid
> isolation of the inputs, is the CMV really so high that you need it? If you
> don't, then you can use a single power supply that can be well packaged to
> avoid any noise or spikes if it incorporates switching. In fact, I've always
> found that for audio (similar need for power rail cleanliness), a simple
> rectifier/filter/linear reg arrangement is hard to beat.

What I've decided to do (because even the solution above works out to
be expensive, given the number of probes) is to prototype a couple of
boards and see if I actually need the isolation. If there's no
measurable difference I'll proceed without. If there is, I'll have to
spend some time figuring out / learning more about how to fix it...

Cheers
Simon.

Simon wrote:

> On Jan 13, 9:17 pm, "Bruce Varley" <b...@NoSpam.com> wrote:
>> "Simon" <goo...@gornall.net> wrote in message
>>
>> news:70d01634-9eb6-4ced-90ac-...
>>
>> > Hi all,
>>
>> > So, I want to put 10 pH/ORP probes onto a microcontroller, with each
>> > probe being galvanically isolated from its peers so that the
>> > measurements of different probes aren't interfering with each other.
>> > Being more of a digital than an analogue guy, I'm not really sure how
>> > to proceed...
>>
>> > Another requirement is that it's a small solution - there's not much
>> > space to play with here, about 0.5" x 1.5" per probe PCB... And of
>> > course, the cheaper the better, given that I want 10 of them!
>>
>> > One chip I found to do the isolation is the adum5241 (http://
>> >www.analog.com/en/interface/digital-
isolators/adum5241/products/produ...)
>> > from analog devices, I could get an isolated GND, +5v, dIn and dOut
>> > from that, which is what I need, but I also want -5v for the probe,
>> > and I'm not sure how to go about getting it. If I try to tack on a +5v-
>> >>-5v conversion after the adum5241, I think I'll over-run its maximum
>> > current delivery (10mA).
>>
>> > So, questions:
>>
>> > - Do I need to isolate the -5v if the +5v and ground are isolated ?
>> > I'd think so, but better to ask
>>
>> > - Is there a better way to do it ? Bound to be - this is just the
>> > first one I've come across
>>
>> > Cheers
>> > Simon.
>>
There have been a few useful suggestions for Isolated DC-DC amplifiers
already so I won't add to that list. However, as someone who has always had
to deal with having to provide full galvanic isolation for inputs from
plant, I can offer some advice on the power supply side. Trying to get small
isolated power on the plant side can be a real pain in the neck if you are
doing it on an individual channel basis. A better solution is to provide a
multi-output power pack where all the DC outputs are isolated from each
other. This usually requires a custom built power pack but that works out
cleaner and more stable than the individual chip based solutions.

--
************************************************** ******************
Paul E. Bennett...............<email://>
Forth based HIDECS Consultancy
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-510979
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
************************************************** ******************

On Jan 14, 1:52 am, Simon <goo...@gornall.net> wrote:
> On Jan 13, 9:17 pm, "Bruce Varley" <b...@NoSpam.com> wrote:
>
>
>
>
>
>
>
>
>
> > "Simon" <goo...@gornall.net> wrote in message
>
> >news:70d01634-9eb6-4ced-90ac-...
>
> > > Hi all,
>
> > > So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> > > probe being galvanically isolated from its peers so that the
> > > measurements of different probes aren't interfering with each other.
> > > Being more of a digital than an analogue guy, I'm not really sure how
> > > to proceed...
>
> > > Another requirement is that it's a small solution - there's not much
> > > space to play with here, about 0.5" x 1.5" per probe PCB... And of
> > > course, the cheaper the better, given that I want 10 of them!
>
> > > One chip I found to do the isolation is the adum5241 (http://
> > >http://www.analog.com/en/interface/d...products/produ...)
> > > from analog devices, I could get an isolated GND, +5v, dIn and dOut
> > > from that, which is what I need, but I also want -5v for the probe,
> > > and I'm not sure how to go about getting it. If I try to tack on a +5v-
> > >>-5v conversion after the adum5241, I think I'll over-run its maximum
> > > current delivery (10mA).
>
> > > So, questions:
>
> > > - Do I need to isolate the -5v if the +5v and ground are isolated ?
> > > I'd think so, but better to ask
>
> > > - Is there a better way to do it ? Bound to be - this is just the
> > > first one I've come across
>
> > > Cheers
> > > Simon.
>
> > I can see all sorts of potential 'gotchas' in this scenario. With Hi-Z
> > inputs, things like transients from switching power supply devices like the
> > 7660 could cause problems unless layout and packaging is just right. As
> > other posts have suggested, things are a lot simpler if you can avoid
> > isolation of the inputs, is the CMV really so high that you need it? If you
> > don't, then you can use a single power supply that can be well packaged to
> > avoid any noise or spikes if it incorporates switching. In fact, I've always
> > found that for audio (similar need for power rail cleanliness), a simple
> > rectifier/filter/linear reg arrangement is hard to beat.
>
> What I've decided to do (because even the solution above works out to
> be expensive, given the number of probes) is to prototype a couple of
> boards and see if I actually need the isolation. If there's no
> measurable difference I'll proceed without. If there is, I'll have to
> spend some time figuring out / learning more about how to fix it...
>
> Cheers
> Simon.

I'm pretty confident that you don't need the isolation between
probes. Are they all going into the same fish tank, or into different
tanks in a multiple aquarium setup?

I would ask what is the purpose of combining them into one
instrument? If you can make the instrument small and inexpensive, it
might be better to make 10 separate instruments. That has some
advantages such as cheaper prototyping and ease of repair by just
replacing a single inexpensive unit with a spare rather than having to
debug and repair a more costly multiple input device.

Rick

On Jan 12, 2:36*pm, Simon <goo...@gornall.net> wrote:
> On Jan 12, 9:45*am, cassiope <f...@u.washington.edu> wrote:
>
>
>
> > On Jan 12, 8:25*am, Simon <goo...@gornall.net> wrote:
>
> > > Hi all,
>
> > > So, I want to put 10 pH/ORP probes onto a microcontroller, with each
> > > probe being galvanically isolated from its peers so that the
> > > measurements of different probes aren't interfering with each other.
> > > Being more of a digital than an analogue guy, I'm not really sure how
> > > to proceed...
>
> > > Another requirement is that it's a small solution - there's not much
> > > space to play with here, about 0.5" x 1.5" per probe PCB... And of
> > > course, the cheaper the better, given that I want 10 of them!
>
> > > One chip I found to do the isolation is the adum5241 (http://www.analog.com/en/interface/d...um5241/product......)
> > > from analog devices, I could get an isolated GND, +5v, dIn and dOut
> > > from that, which is what I need, but I also want -5v for the probe,
> > > and I'm not sure how to go about getting it. If I try to tack on a +5v->-5v conversion after the adum5241, I think I'll over-run its maximum
>
> > > current delivery (10mA).
>
> > > So, questions:
>
> > > *- Do I need to isolate the -5v if the +5v and ground are isolated ?
> > > I'd think so, but better to ask
>
> > > *- Is there a better way to do it ? Bound to be - this is just the
> > > first one I've come across
>
> > > Cheers
> > > * *Simon.
>
> > I'm not sure what these "ORP" probes are - if they're anything like
> > the pH
> > electrodes I've used, they are _extremely_ high impedance; and you
> > need
> > to compare their output voltage with the potential of the solution
> > they are
> > measuring. *Why 10 probes? *Might the solution[s] be at different
> > potentials,
> > and if so, by how much?
>
> ORP is a similar probe, yes - it's an Oxidation/Reduction potential
> probe.
>
> The pH solutions are likely to be similar (within +/- 1 pH at most)
> and it's for measuring / controlling the pH of an aquarium when you're
> introducing chemicals at various places in the water column.
>
>
>
> > IF the solution[s] are at nearly the same potential, you may be able
> > to get
> > away with some good differential amplifiers and dispense with
> > isolation -
> > assuming, of course, that you aren't concerned with galvanically
> > isolating the
> > solution from the output. *But that might be doable with one or two
> > digital
> > isolation devices.
>
> I doubt there'd be much correlation in the readings between an ORP
> probe and a pH probe, though, and I wouldn't want the addition of one
> to affect existing readings. The same system will be reading
> temperature (since pH is temperature-dependent) although that's of
> lesser concern due to the nature of the probe.
>
> On Jan 12, 11:46*am, Spehro Pefhany
>
> <speffS...@interlogDOTyou.knowwhat> wrote:
>
> > You can use an ICL7660 charge pump to get -5V.
>
> That's one I hadn't come across, and for the my needs, it would appear
> to produce a good -5v (the required load should be very low due to the
> massive impedance of the probe). Thanks!
>
> On Jan 12, 9:42*am, Tim Wescott <t...@seemywebsite.com> wrote:
>
>
>
> > * Go looking for a higher power version of the same chip.
>
> And, of course, there is indeed one such - in fact it handles 0.5A.
> This still wouldn't be sufficient for the original 5v -> -5v converter
> chip (which can need 1->2A (!!) on startup), but the ICL chip doesn't
> seem to suffer from that.
>
> Combining the two suggestions, seems like that would work
>
> Thanks for all the help
>
> Simon

I'm with Rick on not needing isolation between channels. The
electrodes shouldn't
alter the solution being tested; and unless you have an incredible
voltage gradient
through the channel (you haven't suggested anything like that), they
should all
be well within the common-mode range of a decent set of amplifiers.
What are you
using for a ground reference? One or more than one electrodes? What
kind of
electrode?