Searching \ for '[EE] Current sensing - op amps' in subject line. ()
Help us get a faster server
FAQ page: www.piclist.com/techref/power.htm?key=amps
Search entire site for: 'Current sensing - op amps'.

Exact match. Not showing close matches.
'[EE] Current sensing - op amps'
2005\11\21@035913 by

Hi List,

I have a project where I'm trying to sense current going through a 0.25 Ohm
resistor, and so far I wasn't able to get good results at the low end of the
range. The requirements are:

- Input current range: 0-2000 mA
- Output voltage range: 0-5V

In principle, the lowest current that I care about in this application is 7
mA (1.75mV input voltage), but it would be nice to differentiate between 0
mA (no current) and 7 mA (idle current).

I'm using the MC34071A op-amp, which has a low input voltage offset (0.5mV
typical), but the output starts out at about 180 mV and doesn't change until
the input voltage reaches ~5 mV. After that point it works perfectly,
multiplying the voltage by a factor of 10.

I have to admit that op amps are not my forte, therefore I would greatly
appreciate any help.

Best regards,

Vitaliy

> I'm using the MC34071A op-amp, which has a low input voltage
> offset (0.5mV
> typical), but the output starts out at about 180 mV and
> doesn't change until
> the input voltage reaches ~5 mV. After that point it works perfectly,
> multiplying the voltage by a factor of 10.

You probably need an opamp that includes ground in both its input range
and its output range.

If the current and/or voltage range is a problem you could try this: use
two opamps, one for the lower part of the range and one for the higher
part, each connected to an A/D input. When the measurement is in the
higher range the opamp for the lower range will saturate, but who cares?
(but don't exceed the allowed input voltage for the A/D pins!).

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Wouter wrote:
> You probably need an opamp that includes ground in both its input range
> and its output range.

Could you please elaborate?

> If the current and/or voltage range is a problem you could try this: use
> two opamps, one for the lower part of the range and one for the higher
> part, each connected to an A/D input. When the measurement is in the
> higher range the opamp for the lower range will saturate, but who cares?
> (but don't exceed the allowed input voltage for the A/D pins!).

Can't do. I basically need a "black box" that takes 0-0.5V and outputs 0-5V.
The PIC is on a different board, and the idea is to make each module
application-independent. For all intents and purposes, the PIC should not
care what is connected to its A/D inputs. Otherwise, I could have connected
them directly to the current-sensing resistors, and performed normalization
in software (resolution is not too important).

Something like the MAX471 would have been ideal, if I could control the
relationship between the current and the output voltage.

Best regards,

Vitaliy

>> You probably need an opamp that includes ground in both its
>> input range
>> and its output range.
>
> Could you please elaborate?

An opamp has limitations. You must use it within its limitations. As
your inputs are likely to be near the ground of your opamp (or do you
use a negative supply?) you will need an opamp that includes ground in
its input range. Likewise you want you opamp to produce very low
voltages, near ground. So the opamp you select must be able to do this.
If you need 5 V output too you will need a rail-to-rail opamp.

> Can't do. I basically need a "black box" that takes 0-0.5V
> and outputs 0-5V.

Take my two opamp approach and add a PIC after it to do the A/D and D/A
conversion :)

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

> Something like the MAX471 would have been ideal, if I
> could control the relationship between the current and the
> output voltage

I use the 472 in this circuit to convert 4-20mA sensors for PIC
ADC, works well across the whole 0-5V range. One sensor is
a 5m depth gauge (1mV / mm), and MAX951 o/p is linear from
0mV to 5000mV

http://www.maxim-ic.com/appnotes.cfm/appnote_number/823

Maybe there's a similar app note for the 471 ?

Hello Wouter,

Thank you for your help.

> An opamp has limitations. You must use it within its limitations. As
> your inputs are likely to be near the ground of your opamp (or do you
> use a negative supply?) you will need an opamp that includes ground in
> its input range. Likewise you want you opamp to produce very low
> voltages, near ground. So the opamp you select must be able to do this.

Can you suggest one? As I mentioned, I'm a newbie when it comes to op-amps.

> If you need 5 V output too you will need a rail-to-rail opamp.

Actually, I am stealing 6.3V from the servo power supply, so that's not an
issue.

>> Can't do. I basically need a "black box" that takes 0-0.5V
>> and outputs 0-5V.
>
> Take my two opamp approach and add a PIC after it to do the A/D and D/A
> conversion :)

I'd like to avoid that if I can. :)

Best regards,

Vitaliy

>Can't do. I basically need a "black box"
>that takes 0-0.5V and outputs 0-5V.

I would think that starting with a 4-20mA interface and working on it to
handle your 0-2000mA would make a starting point.

Alan B. Pearce wrote:
> >Can't do. I basically need a "black box"
>>that takes 0-0.5V and outputs 0-5V.
>
> I would think that starting with a 4-20mA interface and working on it to
> handle your 0-2000mA would make a starting point.

That would mean that I have to use a MAX471 plus an op-amp, for each of the
seven current sensing inputs. Too expensive, and an overkill for this
project.

The simple op-amp circuit works for currents above 20 mA (>5mV). Wouter
mentioned that there are op-amps which include ground in their input range,
I think that's all I would need to solve the problem. Part numbers, anyone?
:)

Best regards,

Vitaliy

>> Something like the MAX471 would have been ideal, if I
>> could control the relationship between the current and the
>> output voltage
>
> I use the 472 in this circuit to convert 4-20mA sensors for PIC
> ADC, works well across the whole 0-5V range. One sensor is
> a 5m depth gauge (1mV / mm), and MAX951 o/p is linear from
> 0mV to 5000mV
>
> http://www.maxim-ic.com/appnotes.cfm/appnote_number/823
>
> Maybe there's a similar app note for the 471 ?

Jinx,

Unfortunately, even if there is, I cannot afford to pay \$4 per input. If
worst comes to worst, I can live with not being able to measure currents
below 20mA, and simply document the "feature."

Best regards,

Vitaliy

> Can you suggest one? As I mentioned, I'm a newbie when it
> comes to op-amps.

No, but there are many piclisters that are more analog-oriented than I
am, so you are bound to get more suggestions than you'll like.

> Actually, I am stealing 6.3V from the servo power supply, so
> that's not an issue.

Note that putting more than Vdd on a PIC input is not considered a good
idea - espacially if you have aother analog inputs.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Wouter wrote:
>> Can you suggest one? As I mentioned, I'm a newbie when it
>> comes to op-amps.
>
> No, but there are many piclisters that are more analog-oriented than I
> am, so you are bound to get more suggestions than you'll like.

I'm still waiting for one. ;)
I was under the impression that an op amp without an input offset voltage
does not exist, I guess I was wrong?

>> Actually, I am stealing 6.3V from the servo power supply, so
>> that's not an issue.
>
> Note that putting more than Vdd on a PIC input is not considered a good
> idea - espacially if you have aother analog inputs.

In the tests we did, the maximum current never exceeded 1.5A (=3.75V). Two
amps (5V) is the absolute worst case scenario.

Best regards,

Vitaliy

> Unfortunately, even if there is, I cannot afford to pay \$4 per
> input. If  worst comes to worst, I can live with not being able
> to measure currents below 20mA, and simply document the
> "feature."

I asked for 2 samples of each from Maxim and got them.
Unknown to me, my customer had asked for 10 of each, and
to my great surprise actually got them. So my sensor convertors
cost nothing

>> No, but there are many piclisters that are more
>> analog-oriented than I am, so you are bound to
>> get more suggestions than you'll like.
>
>I'm still waiting for one. ;)

Look at any of the rail to rail input and output op-amps. Often these will
come in quad amp packages. Microchip have some in their range. Many of these
are CMOS devices. You do need to be a little careful though, as many that
claim to be rail-to-rail outputs only get close to the rails before the
output saturates.

There are others, often using JFET inputs which will go outside the supply
rail (VDD or GND, depending on P or N channel input devices). I cannot think
of any part numbers at the moment.

>I was under the impression that an op amp without an
>input offset voltage does not exist, I guess I was wrong?

No, you are not wrong, but there are ones around which do have extremely
small offset voltages and currents.

> I'm still waiting for one. ;)

be patient, wait til the USA wakes up.

> I was under the impression that an op amp without an input
> offset voltage does not exist, I guess I was wrong?

Every opamp has a (maximum) input voltage offset, and maxium drift with
time, temperature, etc. But some opamps have a much higer one than
others.

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Alan B. Pearce wrote:
>>> No, but there are many piclisters that are more analog-oriented than
>>> I am, so you are bound to get more suggestions than you'll like.
>>
>> I'm still waiting for one. ;)
>
> Look at any of the rail to rail input and output op-amps. Often these
> will come in quad amp packages. Microchip have some in their range.
> Many of these are CMOS devices. You do need to be a little careful
> though, as many that claim to be rail-to-rail outputs only get close
> to the rails before the output saturates.

How about Analog Device's AD8551 (or 8554 for quad). Has Vos 1uV and
can swing to within 10mV of ground. Note 10mV is the specified max,
typical is 1mV (but we all know better than to use the typical
figures, don't we?)

As was mentioned, no OpAmp is perfect e.g. they all have an offset
voltage and none can truly swing rail to rail, but the AD855x comes
pretty darn close. I'm sure there are other mfg. that have
similar types.

If you truly want to go to 0V, you will need to power the OpAmp
with both positive and negative supplies.
Vitaliy wrote:
> I have a project where I'm trying to sense current going through a 0.25
> Ohm resistor, and so far I wasn't able to get good results at the low
> end of the range. The requirements are:
>
> - Input current range: 0-2000 mA
> - Output voltage range: 0-5V

So that means the shunt will produce 500mV full scale, and you therefore
need a voltage gain of 10.

> In principle, the lowest current that I care about in this application
> is 7 mA (1.75mV input voltage), but it would be nice to differentiate
> between 0 mA (no current) and 7 mA (idle current).

That's doable, but keep in mind that the opamp offset voltage is added
directly to the input voltage.  You either have to find an opamp with less
than 1.7mV offset (certainly available, especially if you don't need high
speed), or do some calibration to find the output voltage produced at 0
current and subtract it off from the readings.

> I'm using the MC34071A op-amp, which has a low input voltage offset
> (0.5mV typical), but the output starts out at about 180 mV and doesn't
> change until the input voltage reaches ~5 mV. After that point it works
> perfectly, multiplying the voltage by a factor of 10.

This sounds like your opamp isn't "rail to rail".  Even the ones that are
sometimes have artifacts close to the rails.  Give the opamp some headroom
to work with on both ends, and it will probably be fine.  In other words,
make the negative supply -1 or -2 volts, and the positive supply 6-7 volts
(or more, depending on what the opamp can take).  Watch out that a lot of
Microchip opamps can't take more than 6V, which makes them pretty useless in
most applications.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
Vitaliy wrote:
> I was under the impression that an op amp without an input offset
> voltage does not exist, I guess I was wrong?

Correct.  However you don't need 0 input offset, only somewhat less than
1.7mV.  2mV offset is commonly available, but below 1mV is available too.
It depends on the other parameters you need, like speed and cost.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
> I'm using the MC34071A op-amp, which has a low input voltage offset
> (0.5mV typical), but the output starts out at about 180 mV and doesn't
> change until the input voltage reaches ~5 mV. After that point it works
> perfectly, multiplying the voltage by a factor of 10.

I had a look in the National Semiconductor web site under Analogue/Low Power
and there are a number of op-amps with Common Mode to V- which is really
what you are looking for to get around this particular problem. I would
suggest looking at the LMC6084 for a quad version with similar offset
voltage to what you have.

Alan B. Pearce wrote:
> I had a look in the National Semiconductor web site under Analogue/Low
> Power and there are a number of op-amps with Common Mode to V- which is
> really what you are looking for to get around this particular problem.

That's important, but the easy part.  He also needs the opamp output to go
to the low rail.  While many are specified to do that, there are gotchas if
you actually try to take them up on it.  For example, I've seen supposed
"rail to rail" opamp output jump to 700mV as soon as either input was above
700mV, even though the differential input voltage should have made the
output go to 0.  Common mode range extending to the low rail seems to
usually work as specified, but be very suspicious of any opamp claiming
output to the low rail.

This sort of stuff is not specified in the data sheet, and in fact the data
sheet gives you indications that these things shouldn't happen.  This is an
area where you have to test yourself.  That's why I recommended using a
negative supply.  All you need is a diode drop or two below ground and a
"rail to rail" opamp will work fine.

You can get this for nearly free if you've got a buck switcher in there.
Put a parallel diode and cap in series with the main switcher diode and
voila, you've got a -700mV supply.  In case you think this is just
theorectical, I've actually done this on the ProProg and it works great.  In
that case I used two diodes, so I have roughly a -1.5V supply.  This is only
used for the negative supply of the opamps for exactly the reason Vitaliy is
runnin into.  For the cost of two diodes and a cap it allows me to use cheap
LM324s instead of something more fancy that might not work at the limits
anyway.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
> -----Original Message-----
> From: piclist-bouncesmit.edu On Behalf Of Vitaliy
> Sent: Monday, November 21, 2005 3:59 AM
>
> Hi List,
>
> I have a project where I'm trying to sense current going through a
<snip>
> I'm using the MC34071A op-amp, which has a low input voltage offset
> (0.5mV typical), but the output starts out at about 180 mV and
> doesn't change until the input voltage reaches ~5 mV. After that
> point it works perfectly, multiplying the voltage by a factor of
> 10.

You are hitting up against the output voltage swing specification. The low
side swing (VOL) of the MCMC34071A is 0.1V Typical, 0.3V Maximum with a 2K
load on the output. You are seeing 0.18V which is in spec but towards the
maximum side. If your output load resistance is greater than 2K then
lowering that to 2K will likely get you closer to the typical value of 0.1V.

To get closer to zero with zero input you either need an op-amp with a lower
VOL or use a negative supply.

Also note that the VOH specification is 4.0V typical and 3.7V minimum.

Paul

>
> I have to admit that op amps are not my forte, therefore I would
> greatly appreciate any help.
>
> Best regards,
>
> Vitaliy

>Alan B. Pearce wrote:
>> I had a look in the National Semiconductor web site

>That's important, but the easy part.  He also needs
>the opamp output to go to the low rail.
...
>That's why I recommended using a negative supply.
>All you need is a diode drop or two below ground
>and a "rail to rail" opamp will work fine.
>
>You can get this for nearly free if you've got a
>buck switcher in there.

IIRC he said that there was a 6.3V supply, so he probably needs only a
1N4001 in the negative lead to get a slightly negative voltage.

Alan (guess who has done this before ... ;) )

Check out the LM10 from National Semi.

On 11/21/05, Alan B. Pearce <A.B.Pearcerl.ac.uk> wrote:
> >Alan B. Pearce wrote:
> >> I had a look in the National Semiconductor web site
>
> >That's important, but the easy part.  He also needs
> >the opamp output to go to the low rail.
> ...
> >That's why I recommended using a negative supply.
> >All you need is a diode drop or two below ground
> >and a "rail to rail" opamp will work fine.
> >
> >You can get this for nearly free if you've got a
> >buck switcher in there.
>
> IIRC he said that there was a 6.3V supply, so he probably needs only a
> 1N4001 in the negative lead to get a slightly negative voltage.
>

Neil Baylis wrote:
> Check out the LM10 from National Semi.

How's that supposed to help?  That's a voltage reference.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
It's an Op Amp and a buffered reference. You can use the reference to
establish a pseudo split supply.

Neil

On 11/21/05, Olin Lathrop <olin_piclistembedinc.com> wrote:
> Neil Baylis wrote:
> > Check out the LM10 from National Semi.
>
> How's that supposed to help?  That's a voltage reference.
>
> ******************************************************************
> Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
> consultant in 2004 program year.  http://www.embedinc.com/products
> -

> How's that supposed to help?  That's a voltage reference.

it's a voltage referemce + one free opamp. but that opamp does not
strike me as particularly well suited to this application.

> ******************************************************************
> Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
> consultant in 2004 program year.  http://www.embedinc.com/products
> --

Jinx wrote:
> I asked for 2 samples of each from Maxim and got them.
> Unknown to me, my customer had asked for 10 of each, and
> to my great surprise actually got them. So my sensor convertors
> cost nothing

This is (hopefully) not a one-off project. :) And even for the
prototypes, I would need at least 21 (3x7).

Best regards,

Vitaliy

> From: peisermaridgid.com
> How about Analog Device's AD8551 (or 8554 for quad). Has Vos 1uV and
> can swing to within 10mV of ground. Note 10mV is the specified max,
> typical is 1mV (but we all know better than to use the typical
> figures, don't we?)

This would definitely work, but it is a bit on the expensive side. Is
there something similar for < US\$1?

> If you truly want to go to 0V, you will need to power the OpAmp
> with both positive and negative supplies.

Really, all I care about is being able to tell the difference between
0mA and 7mA (0mV and 1.75mV), and get correct readings (Vin x10) from
7ma to 1500ma. So it seems that the negative supply is not necessary.

Best regards,

Vitaliy

2005\11\21@125205 by
Olin wrote:
> So that means the shunt will produce 500mV full scale, and you therefore
> need a voltage gain of 10.

Absolutely correct.

[snip]
> You either have to find an opamp with less
> than 1.7mV offset (certainly available, especially if you don't need high
> speed), or do some calibration to find the output voltage produced at 0
> current and subtract it off from the readings.

I certainly don't need high speed, the application involves monitoring
current consumed by servo motors.

> This sounds like your opamp isn't "rail to rail".

You're right, it is not. :(

> Even the ones that are
> sometimes have artifacts close to the rails.  Give the opamp some headroom
> to work with on both ends, and it will probably be fine.  In other words,
> make the negative supply -1 or -2 volts, and the positive supply 6-7 volts
> (or more, depending on what the opamp can take).

I will definitely give that a try.

Best regards,

Vitaliy

Paul wrote:
> You are hitting up against the output voltage swing specification. The low
> side swing (VOL) of the MCMC34071A is 0.1V Typical, 0.3V Maximum with a 2K
> load on the output. You are seeing 0.18V which is in spec but towards the
> maximum side. If your output load resistance is greater than 2K then
> lowering that to 2K will likely get you closer to the typical value of 0.1V.

feeling that it would help a lot. You are right - I forgot about the
VOL, my impression was that the reason I got positive voltage on the
output was because of the input voltage offset.

[snip]
> Also note that the VOH specification is 4.0V typical and 3.7V minimum.

True, for a 5V supply. I am using 6.3V to power the op-amp.

Best regards,

Vitaliy

> Really, all I care about is being able to tell the difference between
> 0mA and 7mA (0mV and 1.75mV), and get correct readings (Vin x10) from
> 7ma to 1500ma. So it seems that the negative supply is not necessary.

So you want it cheap, easy, and correct? You can surely get any 2 ....

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Neil Baylis wrote:
> It's an Op Amp and a buffered reference. You can use the reference to
> establish a pseudo split supply.

I still don't see how that solves the OP's problem, since lack of reference
voltage between his power and ground isn't the issue.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
spam@maksimov.org wrote:
>> If you truly want to go to 0V, you will need to power the OpAmp
>> with both positive and negative supplies.
>
> Really, all I care about is being able to tell the difference between
> 0mA and 7mA (0mV and 1.75mV), and get correct readings (Vin x10) from
> 7ma to 1500ma. So it seems that the negative supply is not necessary.

Whoever said "0V" should have said "close to 0V".  1.7mV is close to 0V for
this purpose.  After all, it's smaller than the offset voltage of many
jellybean opamps.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
olin_piclist@embedinc.com wrote:
> spammaksimov.org wrote:
>>> If you truly want to go to 0V, you will need to power the OpAmp with
>>> both positive and negative supplies.
>>
>> Really, all I care about is being able to tell the difference between
>> 0mA and 7mA (0mV and 1.75mV), and get correct readings (Vin x10) from
>> 7ma to 1500ma. So it seems that the negative supply is not necessary.
>
> Whoever said "0V" should have said "close to 0V".  1.7mV is close to
> 0V for this purpose.  After all, it's smaller than the offset voltage
> of many jellybean opamps.

That's true. But as I understand the original problem, the 1.75mV
includes the effect of the low-side voltage swing (Vol), because the OP
wants to distinguish between no current and 7mA. Unless I missed
something, a negative supply is probably the easiest way to accomplish
this. As a previous email pointed out, while the MC34071A has Vos of
500uV, its maximum Vol is 300mV. Even the AD8551 I mentioned is only
guaranteed to 10mV. As long as Vol isn't guaranteed to be less than
the 1.75mV requirement, you're going to need a negative supply.
> -----Original Message-----
> From: piclist-bouncesmit.edu On Behalf Of Vitaliy
> Sent: Monday, November 21, 2005 1:00 PM
>
> [snip]
> > Also note that the VOH specification is 4.0V typical and 3.7V minimum.
>
> True, for a 5V supply. I am using 6.3V to power the op-amp.

That's still cutting it close for a volume design if you want a guaranteed
5V output from the op-amp. Assuming the VOH worst case minimum is no worse
at VCC = 6.3V than it is at VCC = 5.0V, then you will have no safety margin
at all.

Paul

>
> Best regards,
>
> Vitaliy

Wouter,

> So you want it cheap, easy, and correct? You can surely get any 2 ....

I'd like to have a well balanced mixture of the three, with emphasis on
"cheap". :-)

Best regards,

Vitaliy

Hello Olin,

>> Really, all I care about is being able to tell the difference between
>> 0mA and 7mA (0mV and 1.75mV), and get correct readings (Vin x10) from
>> 7ma to 1500ma. So it seems that the negative supply is not necessary.
>
> Whoever said "0V" should have said "close to 0V".  1.7mV is close to 0V
> for
> this purpose.  After all, it's smaller than the offset voltage of many
> jellybean opamps.

It was I who said "0V", but IMHO I made it clear from the beginning that I
don't mean exactly 0V.

I have a feeling that my biggest mistake was not loading the op-amp. I will
try connecting a 2k resistor to the output per Paul's suggestion. If that
does not solve the problem, I will follow your advice and use a negative
power supply.

Thank you all for your help, I will post the solution once I know what it
is.

Best regards,

Vitaliy

>> You are hitting up against the output voltage swing specification.
>> The low
>> side swing (VOL) of the MCMC34071A is 0.1V Typical, 0.3V Maximum
>> with a 2K
>> load on the output. You are seeing 0.18V which is in spec but
>> towards the
>> maximum side. If your output load resistance is greater than 2K
>> then
>> lowering that to 2K will likely get you closer to the typical value
>> of 0.1V.

If you can add a "whisper" of negative supply to opamps like this then
you can swing to true ground. Even a few tenths of a volt will do.
This can be accomplished with eg a simple diode pump driven from just
about any available signal.

Another "naughty" but potentially (pun vaguely intended) trick is to
place a Schottky diode in the ground lead of the power supply.
Decouple this well (large cap across the diode and also a small
ceramic). Take negative opamp supply rail from diode cathode and
system ground from diode anode. This gives you a Schottky diode
voltage drop (typically a few tenths of a volt) as the opamp negative
supply. Voila - true ground output. This method has its potential
"problems" but may well be perfect in your application. This of course
reduces the general supply rail (Vdd-Vss) by the diode drop. eg from
5v to 4.8v or whatever. depending on your regulator you may be able to
increase it by the same amount (regulator divider change with eg
LM317) or add another Schottky in the regulator ground return (care
needed) OR the reduced supply may still be in spec and acceptable.
This is a method that must be applied with understanding BUT can
produce magical results. (Any technology ...). (An 'ordinary' silicon
diode will also work but produces a much greater voltage drop).

RM

At 11:48 AM 11/22/2005 +1300, you wrote:
{Quote hidden}

The 7660 is also a good solution in some cases. Microchip makes a version,
only two external capacitors (can be ceramic SMT or electrolytic), and it
is not too expensive (43 cents US in 100's from Digikey).

Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speffinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff

peiserma@ridgid.com wrote:
> As long as Vol isn't guaranteed to be less than
> the 1.75mV requirement, you're going to need a negative supply.

No, the Vol needs to be somewhat less than the 1.75mV times the gain, which
is 10.  Vol max therefore needs to be no higher than 10mV or so.

******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
Russel wrote:

>
> >Another "naughty" but potentially (pun vaguely intended) trick is to
> place
> >a Schottky diode in the ground lead of the power supply. Decouple this
> >well (large cap across the diode and also a small ceramic). Take negative
> >opamp supply rail from diode cathode and system ground from diode anode.
> >This gives you a Schottky diode voltage drop (typically a few tenths of a
> >volt) as the opamp negative supply. Voila - true ground output.

Yes, this is the solution I was suggesting by using the LM 10. It will give
you a very tight low impedance supply about 200 mV above ground. Of course,
then you must reference your measurement to that phony ground, which is not
always possible. It's best if you use an A/D that has Vref+ and Vref-
inputs. Some PIC A/D channels work this way, I believe.

Neil
Going a fair way back down the discussion

>Something like the MAX471 would have been ideal,
>if I could control the relationship between the
>current and the output voltage.

Look at the Linear Technology LTC6101 and related chips. It took me a fair
while to remember just what the chip number was, but knew I had seen it
somewhere.

Russell McMahon wrote:

>If you can add a "whisper" of negative supply to opamps like this then
>you can swing to true ground. Even a few tenths of a volt will do.
>This can be accomplished with eg a simple diode pump driven from just
>about any available signal.

How about the LTC1150 with it's own internal oscillator output?

Brian.

olin_piclist@embedinc.com wrote:
> peisermaridgid.com wrote:
>> As long as Vol isn't guaranteed to be less than the 1.75mV
>> requirement, you're going to need a negative supply.
>
> No, the Vol needs to be somewhat less than the 1.75mV times the gain,
> which is 10.  Vol max therefore needs to be no higher than 10mV or
> so.

Duh! of course. Thanks for taking the time to point that out.

There is no need for negative voltage or a fancy opamp. Since your range
is low, the zero current output need not be zero volts. In other words,
a simple opamp bias circuit will provide a relatively stable (varies
with supply voltage and may need zero calibration in eeprom at unit
setup time) voltage for the 'zero' and increase in current will cause
this to raise further, all within range of the opamp and of the a/d. A
usual LM358 can do it fine.

Peter
Paul,

>> True, for a 5V supply. I am using 6.3V to power the op-amp.
>
> That's still cutting it close for a volume design if you want a guaranteed
> 5V output from the op-amp. Assuming the VOH worst case minimum is no worse
> at VCC = 6.3V than it is at VCC = 5.0V, then you will have no safety
> margin
> at all.

I understand your concern, however like I mentioned earlier the maximum
voltage produced experimentally was 0.375V, and 0.5V on the input is
theoretically impossible. And if the current exceeds 1.5A (0.375V) all I
really need to know is that it happened, I don't really care if it was 2.1A
or 3A.

Best regards,

Vitaliy

Russell wrote:
> If you can add a "whisper" of negative supply to opamps like this then
> you can swing to true ground. Even a few tenths of a volt will do.
> This can be accomplished with eg a simple diode pump driven from just
> about any available signal.
>
> Another "naughty" but potentially (pun vaguely intended) trick is to
> place a Schottky diode in the ground lead of the power supply.
> Decouple this well (large cap across the diode and also a small
> ceramic). Take negative opamp supply rail from diode cathode and
> system ground from diode anode. This gives you a Schottky diode
> voltage drop (typically a few tenths of a volt) as the opamp negative
> supply. Voila - true ground output. This method has its potential
> "problems" but may well be perfect in your application. This of course
[snip]

Thanks, I will definitely keep this trick in mind!

Best regards,

Vitaliy

To sum it up:

I am using the MC34071A driver with a 1K load resistance, and using the
1N5822 (Shottky) diode to raise the ground reference +0.3V. This gives me a
negative 0.3V supply for the op amp.

The output voltage is Vin x 10 + 2mV, over the entire range. The +2mV error
is perfectly acceptable in this application, even at the low end, since I
only needed to be able to tell the difference between 0 and 7mA (0V and
1.75mV).

Thank you very much for your help.

Best regards,

Vitaliy

{Original Message removed}

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