Maker Pro
Maker Pro

Meter with a log scale

D

Dave

Jan 1, 1970
0
Is there a simple way of getting an analogue meter (50uA or so) to give
a deflection proportional to the log of the input? It need not be
precise at all.

I want to have a 'light level' meter, which will handle a large range of
light levels.

I was thinking that perhaps a forward biased diode in parallel with the
meter might do this, as for increasing input levels, more current will
go through the diode and less through the meter.

I expect there is a chip that will do this sort of thing, but I just
want a simple solution. The instrument has a A/D converter and can
collect the exact data into a file. I just want a simple analogue meter
to show things are not too bad, without having a computer program running.

If it can be done with a diode as I suspect, has anyone tried optimising
the layout of resistors that might be needed to give the most accurate
log response?
 
B

BobG

Jan 1, 1970
0
Regular old ohmmeters have a nonlinear scale... I think its a
reciprocal.... there is a bridge in the meter.
 
D

Dave

Jan 1, 1970
0
BobG said:
Regular old ohmmeters have a nonlinear scale... I think its a
reciprocal.... there is a bridge in the meter.
Does the reciprocal scale not result from simply the fact that the
current through a resistor is proportial to 1/R. I don't think you need
anything clever for that.
 
J

John Larkin

Jan 1, 1970
0
Is there a simple way of getting an analogue meter (50uA or so) to give
a deflection proportional to the log of the input? It need not be
precise at all.

I want to have a 'light level' meter, which will handle a large range of
light levels.

I was thinking that perhaps a forward biased diode in parallel with the
meter might do this, as for increasing input levels, more current will
go through the diode and less through the meter.

I expect there is a chip that will do this sort of thing, but I just
want a simple solution. The instrument has a A/D converter and can
collect the exact data into a file. I just want a simple analogue meter
to show things are not too bad, without having a computer program running.

If it can be done with a diode as I suspect, has anyone tried optimising
the layout of resistors that might be needed to give the most accurate
log response?

A silicon solar cell already has a log voltage output.

John
 
J

John Popelish

Jan 1, 1970
0
Dave said:
Is there a simple way of getting an analogue meter (50uA or so) to give
a deflection proportional to the log of the input? It need not be
precise at all.

I want to have a 'light level' meter, which will handle a large range of
light levels.

I was thinking that perhaps a forward biased diode in parallel with the
meter might do this, as for increasing input levels, more current will
go through the diode and less through the meter.

I expect there is a chip that will do this sort of thing, but I just
want a simple solution. The instrument has a A/D converter and can
collect the exact data into a file. I just want a simple analogue meter
to show things are not too bad, without having a computer program running.

If it can be done with a diode as I suspect, has anyone tried optimising
the layout of resistors that might be needed to give the most accurate
log response?

A forward biased (voltaic mode) photo diode produces a voltage
proportional to the light level, so you can use a linear meter to read
the self generated voltage. You can add a linear DC voltage amplifier
if you need more voltage to match the meter.
 
D

Dave

Jan 1, 1970
0
John said:
A silicon solar cell already has a log voltage output.

John

I can't take that approach.

I have (and must use, as light levels are very low) a silicon avalache
photodiode module from Hamamatsu. The output voltage is 7500 V / Watt of
light on the device. With further amplifiation, the voltage level should
be 1V or a little less, before feeding into an A/D.

The source is a 50mW laser at 780 nm, but this is attenuated a lot,
hence light levels are very low, and I need the sensitivity of the APD.
(I also need a bandwidth of DC-100MHz, which I'm not sure if solar cell
would provide anyway).

This part is just one part of a more complex project. I just wanted an
indication that the light level was about right, and was not too low, or
too high.
 
This is a well-known solution. My Ph.D. thesis includes a reference to
J.F.Gibbons and H.S. Horn's paper on that circuit in the IEEE
Transactions on Circuit Theory CT-11, from page 378 (1964). I didn't
use the circuit in my Ph.D. work, but I did put together a logging and
anti-logging circuits on the same basis at EMI around 1977. They worked
fine, but didn't do anything useful for our ultrasound images.

At Cambridge Instruments, the electron microscopes vacuum monitors used
much the same circuit to compress some five orders of magnitude onto
one analog meter.
 
J

John Larkin

Jan 1, 1970
0
I can't take that approach.

I have (and must use, as light levels are very low) a silicon avalache
photodiode module from Hamamatsu. The output voltage is 7500 V / Watt of
light on the device. With further amplifiation, the voltage level should
be 1V or a little less, before feeding into an A/D.

Wow, I couldn't *afford* and APD module from Hamamatsu!
The source is a 50mW laser at 780 nm, but this is attenuated a lot,
hence light levels are very low, and I need the sensitivity of the APD.
(I also need a bandwidth of DC-100MHz, which I'm not sure if solar cell
would provide anyway).


OK, dump the diode current into a good silicon diode and measure the
diode voltage drop with a high-impedance opamp, or put the diode in
the opamp feedback. The b-e junction of a small-signal transistor is
an excellent diode down to very low currents.


John
 
And if you drive the current into the collector, while maintaining the
collector at much the same voltage as the emitter, it works even
better. See the Gibbons and Horn paper I cited earlier in the thread.
 
J

Joerg

Jan 1, 1970
0
Hello Bill,
This is a well-known solution. My Ph.D. thesis includes a reference to
J.F.Gibbons and H.S. Horn's paper on that circuit in the IEEE
Transactions on Circuit Theory CT-11, from page 378 (1964). I didn't
use the circuit in my Ph.D. work, but I did put together a logging and
anti-logging circuits on the same basis at EMI around 1977. They worked
fine, but didn't do anything useful for our ultrasound images.

In medical ultrasound you really need log detection though. There are
some nice log amps but their prices have gone up quite a bit so often it
becomes necessary to "roll your own". That also solves potential single
source issues.

Regards, Joerg
 
J

Joerg

Jan 1, 1970
0
Hello Dave,
This part is just one part of a more complex project. I just wanted an
indication that the light level was about right, and was not too low, or
too high.

If cost isn't critical and here I mean one or two $20 bills look at the
Analog Devices log amps. AD606 and others. During layout be careful that
the input never "sees" the output but that goes for every fast log amp.

Regards, Joerg
 
M

Mike Monett

Jan 1, 1970
0
Joerg wrote:

[...]
If cost isn't critical and here I mean one or two $20 bills look at the
Analog Devices log amps. AD606 and others. During layout be careful that
the input never "sees" the output but that goes for every fast log amp.

Regards, Joerg

Heh - as long as we are getting into that price range, there's maybe good
reason to look further. Analog methods drift, aren't too accurate, and get
slooooow at low signal levels.

A perfect log converter (cost be dammed) would be a precision A/D, PROM lookup
table to convert to log, followed by a suitable D/A.

This would have no trim adjustments, very tiny tempco, a constant conversion
rate independant of signal level, and no long-term drift. Perfect for the
photographer who has everything.

Mike Monett
 
J

Joerg

Jan 1, 1970
0
Hello Mike,
Heh - as long as we are getting into that price range, there's maybe good
reason to look further. Analog methods drift, aren't too accurate, and get
slooooow at low signal levels.

Not necessarily. I have used an AD log amp in a design a while ago. IIRC
it was the AD630 and two of them in series to achieve the necessary
dynamic range. Nice wide BW down to low levels and very precise.
However, isolating input from output in such a high gain setup is no
small feat.
A perfect log converter (cost be dammed) would be a precision A/D, PROM lookup
table to convert to log, followed by a suitable D/A.

This would have no trim adjustments, very tiny tempco, a constant conversion
rate independant of signal level, and no long-term drift. Perfect for the
photographer who has everything.

That works nicely but when you need a dynamic range north of 80dB and
several MHz of bandwidth it can really become cost prohibitive. The
other downside I have experienced here was that top notch AD converters
do not nearly have the production life of a typical log amp. Innovation
happens so fast that it may be only a few years and you'll have to
redesign the AD section which pretty much means the whole board.

Regards, Joerg
 
R

Robert Baer

Jan 1, 1970
0
BobG said:
Regular old ohmmeters have a nonlinear scale... I think its a
reciprocal.... there is a bridge in the meter.
No there isn't.
The basic circuit is extremely complex: from one lead to a battery to
a selected resistor, to the meter and to the other lead.
 
M

Mike Monett

Jan 1, 1970
0
Joerg said:
Hello Mike,
Not necessarily. I have used an AD log amp in a design a while
ago. IIRC it was the AD630 and two of them in series to achieve
the necessary dynamic range. Nice wide BW down to low levels and
very precise. However, isolating input from output in such a high
gain setup is no small feat.

I thought the AD630 was a lock-in amplifier. Maybe it was the AD640?

Yes, high gain and wide bandwidth are a recipe for late nights:)
That works nicely but when you need a dynamic range north of 80dB
and several MHz of bandwidth it can really become cost
prohibitive.

I thought the OP's app was a light sensor for photography.
The other downside I have experienced here was that top notch AD
converters do not nearly have the production life of a typical log
amp. Innovation happens so fast that it may be only a few years
and you'll have to redesign the AD section which pretty much means
the whole board.

That's called a "lifetime buy". The new devices usually have better
specs and cost less, and you may end up redesigning anyway to keep
up with the competition or the customer's requirements may change.
That's called a "sales opportunity".

So you need to keep the soldering iron plugged in, and have Eagle
and LTSpice running on the desktop:)
Regards, Joerg

Mike Monett
 
J

Joerg

Jan 1, 1970
0
Hello Mike,
I thought the AD630 was a lock-in amplifier. Maybe it was the AD640?

You are right, I believe it was the AD640. The 630 is a commutating
mod/demod.
I thought the OP's app was a light sensor for photography.

Yes but the commercial chips are usually high BW so some care is
necessary. Many people who use them in apps where only a few Hz are
needed tend to overlook that they are dealing with the temperament of a
Ferrari when using commercial log amps.
That's called a "lifetime buy". The new devices usually have better
specs and cost less, and you may end up redesigning anyway to keep
up with the competition or the customer's requirements may change.
That's called a "sales opportunity".

CFOs and corporate boards hate lifetime buys and the resulting surge in
inventory. In the med world many designs just don't need any improvement
over years, often over much more than a decade. Just look at the xray
machines of many dentists. These can be 20+ years old and work just fine
for them. Sometimes it's the same with automotive, for example the Ford
F150.
So you need to keep the soldering iron plugged in, and have Eagle
and LTSpice running on the desktop:)

Interesting. That's the same combination I use. I decided to switch from
OrCad to Eagle a half year ago. I still treasure the cloth covered
manuals of the old PSpice but now prefer LTSpice.

Regards, Joerg
 
M

Mike Monett

Jan 1, 1970
0
Hello Joerg,

[...]
In the med world many designs just don't need any improvement
over years, often over much more than a decade. Just look at the xray
machines of many dentists. These can be 20+ years old and work just fine
for them.

OT, but I just had two wisdom teeth pulled. The dentist said he wanted to
take X-Rays and asked me if it was OK. I figured he was concerned about
possible radiation damage, and said sure.

You know how they used to give you a film plate to hold in your mouth, then
bring the X-Ray head next to the tooth? This was completely different.

He told me to get up out of the chair and took me to another room the size
of a closet. The nurse put my chin on an adjustable rest and went into
another room. When she pushed the button, a mechanism on either side of my
head started rotating. When it stopped, she took me back to the other room.

A minute later, the dentist came back with a small viewer. It held a long
negative that showed a 360 degree view of my entire mouth and teeth.

I have never seen such a clear X-Ray. It made it easy to see the damage was
unrepairable and the teeth had to go. The next time I have to visit a
dentist, I will make sure he uses a machine like this.

This is an example of how fresh designs can vastly outperform older
technology.

[...]
 
J

Joerg

Jan 1, 1970
0
Hello Mike,
OT, but I just had two wisdom teeth pulled. The dentist said he wanted to
take X-Rays and asked me if it was OK. I figured he was concerned about
possible radiation damage, and said sure.

You know how they used to give you a film plate to hold in your mouth, then
bring the X-Ray head next to the tooth? This was completely different.

He told me to get up out of the chair and took me to another room the size
of a closet. The nurse put my chin on an adjustable rest and went into
another room. When she pushed the button, a mechanism on either side of my
head started rotating. When it stopped, she took me back to the other room.

A minute later, the dentist came back with a small viewer. It held a long
negative that showed a 360 degree view of my entire mouth and teeth.

Wow. That is high tech. Mine still uses the "hold this plate, please"
technology but he said it's fine for him. With root canals he has some
modern gizmo to see that the nerve is really dead.
I have never seen such a clear X-Ray. It made it easy to see the damage was
unrepairable and the teeth had to go. The next time I have to visit a
dentist, I will make sure he uses a machine like this.

This is an example of how fresh designs can vastly outperform older
technology.

It sure does. The only question is, what did he charge you for this
x-ray? Mine charges $20.

Regards, Joerg
 
M

Mike Monett

Jan 1, 1970
0
Joerg wrote:

Me wrote:

[...]
It sure does. The only question is, what did he charge you for this
x-ray? Mine charges $20.

Regards, Joerg

I don't know. The whole thing was paid by a medical plan and I never saw
the bill. It couldn't be too much - this is a small town in Canada with
mostly retired people on a pension.

Even if it costs a bit more, it sure is worth it. The amazing thing is
you get the entire picture all at once. I could see the beginning of
infection on the roots of some teeth on the other side, so I know I am in
for a return visit.

I found a site that shows what it looks like. It's called Panorex and
gives a panoramic view to show the general condition of all the teeth. It
is usually taken every 5-7 years. Other types of X-Rays are needed for
more specific diagnosis. (The panoramic image is much larger and better
than shown here:)

http://www.dentalgentlecare.com/necessary_x-rays.htm

Mike Monett
 
Top