Need to drop 0.2 V with 2-ternimal series device

[Tim Hubberstey]

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

The closest thing I can find with similar discharge characteristics is a
silver oxide cell at 1.55 V. Compounding the problem is the fact that
the camera uses the metal body as the positive ground and getting access
to the on-off switch would require far more disassembly than I'm
comfortable with. Hence the need for a 2-terminal solution. The battery
check circuit in the camera considers a cell to be "good" when the
terminal voltage is between 1.27 and 1.35 V at 90 uA load.

A 3 V manganese dioxide lithium cell is also an option if paired with a
3-terminal positive-ground regulator with very low (<10 uA) quiescent
current.

The standard "solution" involves putting a low-current Schottky in
series with the cell, but reports are that this still causes significant
errors in the meter readings on this particular camera model.

The 5 parts limitation comes from the small amount of space available in
the base of the camera, in between various gears.

I don't usually do this type of design so I'm looking for some help.
Anyone got some ideas?

Thanks

 

[Robert Baer]

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

The closest thing I can find with similar discharge characteristics is a
silver oxide cell at 1.55 V. Compounding the problem is the fact that
the camera uses the metal body as the positive ground and getting access
to the on-off switch would require far more disassembly than I'm
comfortable with. Hence the need for a 2-terminal solution. The battery
check circuit in the camera considers a cell to be "good" when the
terminal voltage is between 1.27 and 1.35 V at 90 uA load.

A 3 V manganese dioxide lithium cell is also an option if paired with a
3-terminal positive-ground regulator with very low (<10 uA) quiescent
current.

The standard "solution" involves putting a low-current Schottky in
series with the cell, but reports are that this still causes significant
errors in the meter readings on this particular camera model.

The 5 parts limitation comes from the small amount of space available in
the base of the camera, in between various gears.

I don't usually do this type of design so I'm looking for some help.
Anyone got some ideas?

Thanks

Are you saying that the 1.55 V silver oxide cell messes up the
reading or functionality of the camera?
Other than a resistor (if current load is relatively constant), theer
is nothing that would give a 0.2V drop.
A schottky or germanium diode will give about 300-350mV drop,
depending on current and diode rating.
Now using a 3-terminal regulator with a 3V source seems to be a
solution, but you need one with two attributes: Low Drop-Out (or LDO in
the trades) *and* low drain.
Maxim advertises a shunt regulaor that supposedly works down to 1uA,
but they can be completely discounted, as Maxim/Dallas parts are vaporware.
However, Linear technology does have the LT1389 shunt regulator,
available at 1.25V, 2.5V and 4.096V, speced to operate from 0.8uA to
2mA; curves show operation down to around 0.4uA.
Then their LT1634 shunt regulator, available at 1.25V, 2.5V, 4.096V
and 5V, speced to operate from 10uA to 2mA.
Their LT1004 seems to have similar specs, but available only at 1.23V
and 2.50V.
The LM285-2.5 works from 20uA to 20mA, and there are adjustable
versions made by National Semi.
So, you could make your own 3-term regulator, as most of what is
cited above is available in SOT-23 or SOIC-8.
Idea: NPN pass transistor, base to cathode of adjustable, FB of
adjustable to tap of voltage divider: 1.25 meg from FB to ground (anode)
and 100K from FB to NPN emitter.
Now all one needs is base drive from NPN collector that will run the
regulator and the NPN.
Say one allows a LM285 regulator current of 10uA and a base drive of
1uA (yes, that is high - so the extra also goes thru the LM285).
The base will be 1.35V (ouput or emitter voltage) plus a Vbe of about
600mV, or about 1.95V; making the Vcb about 1.05V.
Obviously, no constant current device (JFET or DMOS) is available, so
a resistor "pullup" will have to do.
E = I * R or 1.05(V) = 11(uA) * R(megohms); or roughly 100K.
The resistors can be SMD 0805; easily available in 1% values to 10
megs; the NPN is available in SOT-23 so it can be done in a small
profile PCB which could be 20mils thick if desired.
That totals to the 5 parts max.

 

[Joop]

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

The closest thing I can find with similar discharge characteristics is a
silver oxide cell at 1.55 V. Compounding the problem is the fact that
the camera uses the metal body as the positive ground and getting access
to the on-off switch would require far more disassembly than I'm
comfortable with. Hence the need for a 2-terminal solution. The battery
check circuit in the camera considers a cell to be "good" when the
terminal voltage is between 1.27 and 1.35 V at 90 uA load.

A 3 V manganese dioxide lithium cell is also an option if paired with a
3-terminal positive-ground regulator with very low (<10 uA) quiescent
current.

The standard "solution" involves putting a low-current Schottky in
series with the cell, but reports are that this still causes significant
errors in the meter readings on this particular camera model.

The 5 parts limitation comes from the small amount of space available in
the base of the camera, in between various gears.

I don't usually do this type of design so I'm looking for some help.
Anyone got some ideas?

Thanks

Perhaps you should try an LM10 opamp with 200mV voltage reference.
Specs say it could work down to 1.1V. The 90uA output might keep the
the saturation enough below 0.2V.

Of course you must be able to live with the quiescent current of about
300uA. But if it is similar to my old camera, then there is a button
press needed to operate the light meter. So it is not on very often.

Joop

 

[Joop]

Perhaps you should try an LM10 opamp with 200mV voltage reference.
Specs say it could work down to 1.1V. The 90uA output might keep the
the saturation enough below 0.2V.

Of course you must be able to live with the quiescent current of about
300uA. But if it is similar to my old camera, then there is a button
press needed to operate the light meter. So it is not on very often.

Joop

I just read your post more carefully and see the requirement of having
the positive side to the camera body.
Making it a 1.35V regulator might not work without external
components. Right now I can't think of a simple solution for that.

But making it a 0.2V regulator is simple. Just connect the reference
opamp output to the its input. When your battery gets more empty, then
the opamp output drop voltage relative to the positive rail. This
setup always 'takes 0.2V off', but requires no external components.

Joop

 

[Frank Bemelman]

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

The closest thing I can find with similar discharge characteristics is a
silver oxide cell at 1.55 V. Compounding the problem is the fact that
the camera uses the metal body as the positive ground and getting access
to the on-off switch would require far more disassembly than I'm
comfortable with. Hence the need for a 2-terminal solution. The battery
check circuit in the camera considers a cell to be "good" when the
terminal voltage is between 1.27 and 1.35 V at 90 uA load.

A 3 V manganese dioxide lithium cell is also an option if paired with a
3-terminal positive-ground regulator with very low (<10 uA) quiescent
current.

The standard "solution" involves putting a low-current Schottky in
series with the cell, but reports are that this still causes significant
errors in the meter readings on this particular camera model.

The 5 parts limitation comes from the small amount of space available in
the base of the camera, in between various gears.

I don't usually do this type of design so I'm looking for some help.
Anyone got some ideas?

Could you use a supercap and charge it to 1.35V ?

 

[Frank Bemelman]

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

Does it affect the light meter much? If only 10%, does it make a
noticable difference?. Shutter and diapragm settings are made in
large steps, and even one (half) step off isn't often a problem.
Perhaps there is a calibration pot, that you could adjust.

[snip]

 

[Martin Riddle]

The mercury cells were very stable over their life span.
The selenium photo detectors used this 'calibrated voltage' to give stable results.
The idea is to mimic the mercury cells with the replacement battery, ahla 0.200 v drop.

Maybe a 3v or 1.5v battery and a small switching regulator to 1.35v will work. The problem is where to put it.

Cheers

Hi all,

I've got a problem that I've been wracking my brain over for a while
now. I need a small (under 5 parts) 2-terminal circuit to drop 0.2 V at
up to 100 uA.

The situation is that I have an old but high quality camera that used a
1.35 V mercury cell for the light meter. These cells are no longer
available due to a world-wide ban on mercury cells. I had a few extras
stashed away but those are now gone.

Does it affect the light meter much? If only 10%, does it make a
noticable difference?. Shutter and diapragm settings are made in
large steps, and even one (half) step off isn't often a problem.
Perhaps there is a calibration pot, that you could adjust.

[snip]

 

[Mark]

if the calibration of the camera is off an f stop or 2 with the higher
voltage battery, just trick the camera by setting the film speed a
little off.

if your pictures are coming out too bright, set the film speed a little
faster than it really is and the camera will adjust to let in less
light.

or vice versa as the case may be

Mark

 

[Tim Hubberstey]

Are you saying that the 1.55 V silver oxide cell messes up the reading
or functionality of the camera?

It throws the meter off by about 1.5 stops. The camera is totally
mechanical, except for the meter.

Other than a resistor (if current load is relatively constant), theer
is nothing that would give a 0.2V drop.
A schottky or germanium diode will give about 300-350mV drop,
depending on current and diode rating.
Now using a 3-terminal regulator with a 3V source seems to be a
solution, but you need one with two attributes: Low Drop-Out (or LDO in
the trades) *and* low drain.
Maxim advertises a shunt regulaor that supposedly works down to 1uA,
but they can be completely discounted, as Maxim/Dallas parts are vaporware.
However, Linear technology does have the LT1389 shunt regulator,
available at 1.25V, 2.5V and 4.096V, speced to operate from 0.8uA to
2mA; curves show operation down to around 0.4uA.
Then their LT1634 shunt regulator, available at 1.25V, 2.5V, 4.096V
and 5V, speced to operate from 10uA to 2mA.
Their LT1004 seems to have similar specs, but available only at 1.23V
and 2.50V.
The LM285-2.5 works from 20uA to 20mA, and there are adjustable
versions made by National Semi.
So, you could make your own 3-term regulator, as most of what is cited
above is available in SOT-23 or SOIC-8.
Idea: NPN pass transistor, base to cathode of adjustable, FB of
adjustable to tap of voltage divider: 1.25 meg from FB to ground (anode)
and 100K from FB to NPN emitter.
Now all one needs is base drive from NPN collector that will run the
regulator and the NPN.
Say one allows a LM285 regulator current of 10uA and a base drive of
1uA (yes, that is high - so the extra also goes thru the LM285).
The base will be 1.35V (ouput or emitter voltage) plus a Vbe of about
600mV, or about 1.95V; making the Vcb about 1.05V.
Obviously, no constant current device (JFET or DMOS) is available, so
a resistor "pullup" will have to do.
E = I * R or 1.05(V) = 11(uA) * R(megohms); or roughly 100K.
The resistors can be SMD 0805; easily available in 1% values to 10
megs; the NPN is available in SOT-23 so it can be done in a small
profile PCB which could be 20mils thick if desired.
That totals to the 5 parts max.

Thanks, Robert!

I didn't use your circuit but you did give me an idea and the lead on
the LT1389. This is what I ended up with:

+---+-----+-------+--------+
| | | | .|.
=== | | | | |
Camera | | | | |1M
Body | | | '-'
| |LT1389 | |
| _|_/ +---|--------+
| / A | | .|.
--- | | | | |
3V - | | | | |68k
LiMnO2 | | | |\| '-'
| | +-|-\ | to
| | | >-------+---> Meter
| +-----|+/
| | |/| LT1494
| .-. |
| | | |
| | |1M2 |
| '-' |
| | |
+-----+-------+

(created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

LTspice says it draws 3.8 uA quiescent, so my battery should last a
couple of years. I may do a bit more tweaking on the resistor values
after further analysis but I think that this is a pretty good starting
point.

Any obvious problems?

Thanks.

 

[Tim Hubberstey]

if the calibration of the camera is off an f stop or 2 with the higher
voltage battery, just trick the camera by setting the film speed a
little off.

if your pictures are coming out too bright, set the film speed a little
faster than it really is and the camera will adjust to let in less
light.

or vice versa as the case may be

That would probably work but I'd rather "do it right" since I'm going to
loan the camera to my niece (a definite non-techie) once I get a new
digital SLR. Plus, the higher voltage would invalidate the battery check
setting, and I'm also not sure what non-linearities (if any) a higher
voltage would introduce.

All in all, supplying the correct voltage is the best solution.

 

[Mark]

good luck

Mark

 

[John Fields]

Thanks, Robert!

I didn't use your circuit but you did give me an idea and the lead on
the LT1389. This is what I ended up with:

+---+-----+-------+--------+
| | | | .|.
=== | | | | |
Camera | | | | |1M
Body | | | '-'
| |LT1389 | |
| _|_/ +---|--------+
| / A | | .|.
--- | | | | |
3V - | | | | |68k
LiMnO2 | | | |\| '-'
| | +-|-\ | to
| | | >-------+---> Meter
| +-----|+/
| | |/| LT1494
| .-. |
| | | |
| | |1M2 |
| '-' |
| | |
+-----+-------+

---
Shouldn't it be wired like this?


+-------+----------+-- Camera body
| | |
| [1M] |
|+ | | LT1494
[BAT] +---------|+\
| |K | >--+-- To meter
| [LT1389] +--|-/ |
| | | | [68K]
| | | | |
+-------+-[1M]-+---|----+
| |
+------------------+

 

[Tim Hubberstey]

Shouldn't it be wired like this?

+-------+----------+-- Camera body
| | |
| [1M] |
|+ | | LT1494
[BAT] +---------|+\
| |K | >--+-- To meter
| [LT1389] +--|-/ |
| | | | [68K]
| | | | |
+-------+-[1M]-+---|----+
| |
+------------------+

No, this circuit sets the "to meter" point at 1.35 V above the negative
rail. The output needs to be referenced to the positive rail (body).
Thanks for checking it over, though.

 

[Mac]

It throws the meter off by about 1.5 stops. The camera is totally
mechanical, except for the meter.


Thanks, Robert!

I didn't use your circuit but you did give me an idea and the lead on
the LT1389. This is what I ended up with:

+---+-----+-------+--------+
| | | | .|.
=== | | | | |
Camera | | | | |1M
Body | | | '-'
| |LT1389 | |
| _|_/ +---|--------+
| / A | | .|.
--- | | | | |
3V - | | | | |68k
LiMnO2 | | | |\| '-'
| | +-|-\ | to
| | | >-------+---> Meter
| +-----|+/
| | |/| LT1494
| .-. |
| | | |
| | |1M2 |
| '-' |
| | |
+-----+-------+

(created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

LTspice says it draws 3.8 uA quiescent, so my battery should last a
couple of years. I may do a bit more tweaking on the resistor values
after further analysis but I think that this is a pretty good starting
point.

Any obvious problems?

Thanks.

I don't have any nano-power design experience either, but it looks to me
like your circuit has a shot at working. I think better values for the
resistors are 1 meg and 80.6k, though.

Of course you will use 1% tolerance resistors, which will give you a
variance of around 1% in the output voltage. If you want to, you could use
a 100k trimpot instead of the 80.6k resistor. Then you can trim to the
desired voltage. Or you could use your 68k resistor, then add a 20k
trimpot, but then the parts count will be up to 6 parts.

I also haven't checked the deviations you will get due to leakage currents
at the inputs, but the 1494 seems to have pretty good specs in that
regard. If you use a trimpot, this probably won't matter anyway.

I was going to recommend the lm185 (or lm285), but this circuit seems to
have that one beat in terms of current consumption. But if you did use the
lm185, the part count might be as low as 3 parts. The current consumption
would be right around 10 uA.

regards,
Mac

 

[Robert Baer]

It throws the meter off by about 1.5 stops. The camera is totally
mechanical, except for the meter.



Thanks, Robert!

I didn't use your circuit but you did give me an idea and the lead on
the LT1389. This is what I ended up with:

+---+-----+-------+--------+
| | | | .|.
=== | | | | |
Camera | | | | |1M
Body | | | '-'
| |LT1389 | |
| _|_/ +---|--------+
| / A | | .|.
--- | | | | |
3V - | | | | |68k
LiMnO2 | | | |\| '-'
| | +-|-\ | to
| | | >-------+---> Meter
| +-----|+/
| | |/| LT1494
| .-. |
| | | |
| | |1M2 |
| '-' |
| | |
+-----+-------+

(created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

LTspice says it draws 3.8 uA quiescent, so my battery should last a
couple of years. I may do a bit more tweaking on the resistor values
after further analysis but I think that this is a pretty good starting
point.

Any obvious problems?

Thanks.

No fiddling needed.
Instead of 1 meg, use an 0805 1.24 meg 1%; Digikey cut tape price is
an outrageous $0.80; and thus instead of 68K use an 0805 100 K 1%; same
price.
If the reference was exactly 1.24V, then the op-amp would be at 1.35V
within 1% (the resistors in practice are inside the limits of the spec,
so the average error of the output voltage works out to within 1%.
Since the reference is 1.25V "exactly", the derived voltage will be
slightly higher than 1.35V, and will be well within 1.4% (worst case)
and probably better than 1%.
Then, might as well change that 1.2 meg and use 1.24 meg 1% from the
same cut tape to save money.
Lay out a PCB and have a bunch made (Express PCB or similar), assemble
them yourself and sell the solution!

 

[Winfield Hill]

Tim Hubberstey wrote...

Thanks, Robert!

I didn't use your circuit but you did give me an idea and the lead
on the LT1389. This is what I ended up with:

+---+-----+-------+--------+
| | | | .|.
=== | | | | |
Camera | | | | |1M
Body | | | '-'
| |LT1389 | |
| _|_/ +---|--------+
| / A | | .|.
--- | | | | |
3V - | | | | |68k
LiMnO2 | | | |\| '-'
| | +-|-\ | to
| | | >-------+---> Meter
| +-----|+/
| | |/| LT1494
| .-. |
| | | |
| | |1M2 |
| '-' |
| | |
+-----+-------+

I'd use 1% resistors. If you move the 1.2M and change its value,
you can make the 1.25V reference current independent of the battery
voltage by running it from the 1.35V output. Then if you select a
different opamp, you can run it from your 1.55V silver oxide cell.

.. +---+-----+-------+--------+----
.. | | | | |
.. === | |LT1389 | 1.00M
.. Camera | \_|_ | |
.. Body | /_\ ,--- |--------+
.. _|_ | | | |
.. - | | _| 80.2k
.. 1.55V | | '--|- \ | to
.. | | | >---+--+---> Meter
.. | +-----|+_/ | -1.35V
.. | | | |
.. '---- |-------' |
.. '-- 68k ------'
.. 1.5uA -->

 

[Winfield Hill]

Winfield Hill wrote...

I'd use 1% resistors. If you move the 1.2M and change its value,
you can make the 1.25V reference current independent of the battery
voltage by running it from the 1.35V output. Then if you select a
different opamp, you can run it from your 1.55V silver oxide cell.

. +---+-----+-------+--------+----
. | | | | |
. === | |LT1389 | 1.00M
. Camera | \_|_ | |
. Body | /_\ ,--- |--------+
. _|_ | | | |
. - | | _| 80.2k
. 1.55V | | '--|- \ | to
. | | | >---+--+---> Meter
. | +-----|+_/ | -1.35V
. | | | |
. '---- |-------' |
. '-- 68k ------'
. 1.5uA -->

BTW, in this circuit the opamp shouldn't necessarily be a RRIO type.
In operation the input and outputs are both near the negative rail,
an easy opamp spec (even tho low-power 1.5V operation is not). But
to insure startup, you want an opamp whose input-stage works properly
with CM voltages closer to the + rail than the output is at startup,
or that pushes the output negative until the input CM is working.
If you turn the circuit upside-down, and are working at higher power
levels, many classic single-supply opamps meet these unusual specs.
I'm not sure if any offered opamps will work in my 1.55V circuit...

 

[Tim Hubberstey]

I'd use 1% resistors. If you move the 1.2M and change its value,
you can make the 1.25V reference current independent of the battery
voltage by running it from the 1.35V output. Then if you select a
different opamp, you can run it from your 1.55V silver oxide cell.

. +---+-----+-------+--------+----
. | | | | |
. === | |LT1389 | 1.00M
. Camera | \_|_ | |
. Body | /_\ ,--- |--------+
. _|_ | | | |
. - | | _| 80.2k
. 1.55V | | '--|- \ | to
. | | | >---+--+---> Meter
. | +-----|+_/ | -1.35V
. | | | |
. '---- |-------' |
. '-- 68k ------'
. 1.5uA -->

Looks interesting... but won't it need a large-value resistor to the
anode of the LT1389 to get it to start up? Looks to me like it would
come up with the output = 0 and just stay there. Or could we rely on
noise to give it enough of a "push" to get it started?

 

[Winfield Hill]

Tim Hubberstey wrote...

Looks interesting... but won't it need a large-value resistor to the
anode of the LT1389 to get it to start up? Looks to me like it would
come up with the output = 0 and just stay there. Or could we rely on
noise to give it enough of a "push" to get it started?

I posted a note concerning startup. This type of bootstrap circuit
start up fine if the opamp's output has more BJT saturation voltage
than the input's offset (divided by the feedback-resistor ratio), so
when the power is applied, the + input sees a bit more voltage than
the - input and drives the output further toward the turned-on state.

There are two problematic issues. First, the feedback resistors to
common tend to reduce the BJT saturation voltage. Using high-value
resistors helps; e.g., here they draw a current of 30nA for say 30mV
of output-transistor saturation. We note that 30nA is probably much
less than the base-drive current for the PNP output transistor.

Second, when the output is only modestly-higher than the reference,
the "offset voltage divided by the feedback-resistor ratio" parameter
becomes painful. Here we would divide our estimated 30mV by 12.5 to
get a 2.5mV max offset-voltage spec, which might be hard to meet.

 

[Winfield Hill]

Winfield Hill wrote...

Tim Hubberstey wrote...

I posted a note concerning startup. This type of bootstrap circuit
start up fine if the opamp's output has more BJT saturation voltage
than the input's offset (divided by the feedback-resistor ratio), so
when the power is applied, the + input sees a bit more voltage than
the - input and drives the output further toward the turned-on state.

There are two problematic issues. First, the feedback resistors to
common tend to reduce the BJT saturation voltage. Using high-value
resistors helps; e.g., here they draw a current of 30nA for say 30mV
of output-transistor saturation. We note that 30nA is probably much
less than the base-drive current for the PNP output transistor.

Second, when the output is only modestly-higher than the reference,
the "offset voltage divided by the feedback-resistor ratio" parameter
becomes painful. Here we would divide our estimated 30mV by 12.5 to
get a 2.5mV max offset-voltage spec, which might be hard to meet.

At which point Tim's suggested resistor is the way to go.

.. +---+-----+-------+---------+----
.. | | | | |
.. === | |LT1389 | 1.00M
.. Camera | \_|_ | |
.. Body | /_\ ,--- |---------+
.. _|_ | | | |
.. - | | _| 80.2k
.. 1.55V | | '--|- \ | to
.. | | | >-----+-+---> Meter
.. | +--+--|+_/ | -1.35V
.. | | | | |
.. | 10M '--- |- 68k -'
.. | | |
.. '-----+-------'

Assuming a good-quality reference, this two-resistor bias scheme has
little advantage over one resistor to the battery, unless a variety of
battery voltages is expected and power conservation is very important.