DC/DC converter in low-noise/photodiode application?

[Chris Carlen]

Hi:

I'm contemplating the design of an amplified photodiode instrument that
should be able to bias a photodiode for photoconductive operation, and
provide a positive voltage output of at least 0-10V into a high-Z load.
There would likely be three stages of active circuitry, a
transimpedance amp, an adjustable extra gain stage, and a
buffer/line-driver. Max bandwidth of about 10MHz. I won't be
optimizing the design for noise other than making sure power is clean,
the board layout is decent, mid-range parts with fairly low-noise as
well as low-power are chosen, and the circuitry is enclosed in a metal
can for shielding. This does not need to be a "shot-noise limited"
design, a task that is beyond my capabilities.

It would be preferable to power it with a single 9V battery. However,
to meet the specs, this would then require some DC/DC conversion to make
a split supply for ease of op-amp circuitry. Or perhaps I'd power
from 2x9V batts, and just use a switched-cap chip to generate -V for the
first transimpedance stage. That way the DC/DC could be less
complicated and not need to supply much power.

But can a DC/DC be incorporated in a way that will not substantially
increase the noise of the system? In other words, would you even
consider putting any form of DC/DC into a can with a photodiode amp, and
expect to avoid adding any noise to the system that would not have
originally been present?

If it hopeless to avoid adding noise by putting a DC/DC into the can
then I'll have to settle for a remote split DC supply fed by a cable.


Thanks for input.



--
Good day!

________________________________________
Christopher R. Carlen
Principal Laser&Electronics Technologist
Sandia National Laboratories CA USA
[email protected]
NOTE, delete texts: "RemoveThis" and
"BOGUS" from email address to reply.

 

[Phil Hobbs]

Hi:

I'm contemplating the design of an amplified photodiode instrument that
should be able to bias a photodiode for photoconductive operation, and
provide a positive voltage output of at least 0-10V into a high-Z load.
There would likely be three stages of active circuitry, a
transimpedance amp, an adjustable extra gain stage, and a
buffer/line-driver. Max bandwidth of about 10MHz. I won't be
optimizing the design for noise other than making sure power is clean,
the board layout is decent, mid-range parts with fairly low-noise as
well as low-power are chosen, and the circuitry is enclosed in a metal
can for shielding. This does not need to be a "shot-noise limited"
design, a task that is beyond my capabilities.

It would be preferable to power it with a single 9V battery. However,
to meet the specs, this would then require some DC/DC conversion to make
a split supply for ease of op-amp circuitry. Or perhaps I'd power from
2x9V batts, and just use a switched-cap chip to generate -V for the
first transimpedance stage. That way the DC/DC could be less
complicated and not need to supply much power.

But can a DC/DC be incorporated in a way that will not substantially
increase the noise of the system? In other words, would you even
consider putting any form of DC/DC into a can with a photodiode amp, and
expect to avoid adding any noise to the system that would not have
originally been present?

If it hopeless to avoid adding noise by putting a DC/DC into the can
then I'll have to settle for a remote split DC supply fed by a cable.


Thanks for input.

Sure you can. Avoid magnetics, though--a capacitor charge pump voltage
inverter followed by a capacitance multiplier will do great.

Cheers,

Phil Hobbs

 

[[email protected]]

It would be preferable to power it with a single 9V battery. However,
to meet the specs, this would then require some DC/DC conversion to make
a split supply for ease of op-amp circuitry. Or perhaps I'd power
from 2x9V batts, and just use a switched-cap chip to generate -V for the
first transimpedance stage. That way the DC/DC could be less
complicated and not need to supply much power.

You dont need any converters if your using batteries since they are a
floating supply you can define where the "0v" point is. For example if
you have 2 9v batts in series with a 12v zeneer to ground you get a
=12v, -6v supply.

 

[Joerg]

Hello Phil,

Sure you can. Avoid magnetics, though--a capacitor charge pump voltage
inverter followed by a capacitance multiplier will do great.

I use magnetics in there all the time and some switchers are right next
to sensitive receivers. Works fine if layout and design are 'RF proof'.
The old saying: Toroids don't talk much. So that's my core of choice,
usually.

Regards, Joerg

 

[John Larkin]

Hi:

I'm contemplating the design of an amplified photodiode instrument that
should be able to bias a photodiode for photoconductive operation, and
provide a positive voltage output of at least 0-10V into a high-Z load.
There would likely be three stages of active circuitry, a
transimpedance amp, an adjustable extra gain stage, and a
buffer/line-driver. Max bandwidth of about 10MHz. I won't be
optimizing the design for noise other than making sure power is clean,
the board layout is decent, mid-range parts with fairly low-noise as
well as low-power are chosen, and the circuitry is enclosed in a metal
can for shielding. This does not need to be a "shot-noise limited"
design, a task that is beyond my capabilities.

It would be preferable to power it with a single 9V battery. However,
to meet the specs, this would then require some DC/DC conversion to make
a split supply for ease of op-amp circuitry. Or perhaps I'd power
from 2x9V batts, and just use a switched-cap chip to generate -V for the
first transimpedance stage. That way the DC/DC could be less
complicated and not need to supply much power.

But can a DC/DC be incorporated in a way that will not substantially
increase the noise of the system? In other words, would you even
consider putting any form of DC/DC into a can with a photodiode amp, and
expect to avoid adding any noise to the system that would not have
originally been present?

If it hopeless to avoid adding noise by putting a DC/DC into the can
then I'll have to settle for a remote split DC supply fed by a cable.


Thanks for input.

This gadget...

http://www.highlandtechnology.com/DSS/J720-730DS.html

has a silicon pin diode feeding a couple of fast opamps, then a
comparator. It runs from external +12 volts. I used a National LM2664
to flip +6 to -5 to power the opamps, with an extra R-C on its input
and output for luck (which is where the extra volt is lost.) There's
no hint of dc/dc noise, and we're getting link jitters below 10 ps
RMS. Like your app, there's no attempt at shot-noise level performance
(sorry, Phil) as we have tons of light, nominally a milliwatt from the
laser end.

Go for it.

John

 

[Phil Hobbs]

Hello Phil,


I use magnetics in there all the time and some switchers are right next
to sensitive receivers. Works fine if layout and design are 'RF proof'.
The old saying: Toroids don't talk much. So that's my core of choice,
usually.

Well, I certainly defer to the your expertise if you've made this work
well--I have never actually designed a switcher with a toroidal
transformer or inductor, let alone put one near a front end amplifier.
I'm glad to know that it's possible.

What I was mainly thinking of was the garden variety low-power switcher,
using a solenoidal SMT inductor that sprays flux everywhere. The OP's
power requirement is very modest, and his level of expertise unknown (to
me at least), so a charge pump seems like good medicine. Oh, and he'll
need cap multipliers on both polarities--the charge pump will put fuzz
even on a battery.

Cheers,

Phil Hobbs

 

[Joerg]

Hello Phil,

Well, I certainly defer to the your expertise if you've made this work
well--I have never actually designed a switcher with a toroidal
transformer or inductor, let alone put one near a front end amplifier.
I'm glad to know that it's possible.

Toroids are wonderful. You just have to make sure that there is no
residual DC component that could cause core saturation because you can't
really have air gaps on the cheap ones. Saturation would turn the
switching device into a pretzel if you don't include a detector for that.

I have some designs that are around or above 10 years in production now.
The nice thing is that these sailed through EMC in a jiffy, clean like a
whistle.

What I was mainly thinking of was the garden variety low-power switcher,
using a solenoidal SMT inductor that sprays flux everywhere. The OP's
power requirement is very modest, and his level of expertise unknown (to
me at least), so a charge pump seems like good medicine. Oh, and he'll
need cap multipliers on both polarities--the charge pump will put fuzz
even on a battery.

Charge pumps usually also put fuzz on the signal path. The current peaks
can create some headaches there. For very low power and muffled with
resistors it could work.

Quite frankly, I'd chuck those 9V batteries. They can be a pain, buttons
ripping off, premature failures and so on. In Chris' case I'd take two
or three AA cells and create whatever voltage is needed with a switcher
or two.

Regards, Joerg

 

[Joerg]

Hello John,

This gadget...

http://www.highlandtechnology.com/DSS/J720-730DS.html

I really like the one in the round tube. Cool. Reminds me of the inline
hydrophone amps that cost a small fortune.

Regards, Joerg

 

[Chris Carlen]

This gadget...

http://www.highlandtechnology.com/DSS/J720-730DS.html

has a silicon pin diode feeding a couple of fast opamps, then a
comparator. It runs from external +12 volts. I used a National LM2664
to flip +6 to -5 to power the opamps, with an extra R-C on its input
and output for luck (which is where the extra volt is lost.) There's
no hint of dc/dc noise, and we're getting link jitters below 10 ps
RMS. Like your app, there's no attempt at shot-noise level performance
(sorry, Phil) as we have tons of light, nominally a milliwatt from the
laser end.

Go for it.

John

John, what are the prices for those gadgets?



--
Good day!

________________________________________
Christopher R. Carlen
Principal Laser&Electronics Technologist
Sandia National Laboratories CA USA
[email protected]
NOTE, delete texts: "RemoveThis" and
"BOGUS" from email address to reply.

 

[John Larkin]

John, what are the prices for those gadgets?

The laser-in-a-tube is $175 and the receiver is $375.

John

 

[John Larkin]

Hello John,


I really like the one in the round tube. Cool. Reminds me of the inline
hydrophone amps that cost a small fortune.

Regards, Joerg

The laser comes with a big square ST-connector housing which we
machine down, believe it or not. We have to machine the BNC, too.
There's a tiny PC board inside with a few resistors.

John

 

[Joerg]

Hello John,

The laser-in-a-tube is $175 and the receiver is $375.

Compared to the inline amps I am used to that sounds like a bargain. And
it's not that those amps have many more parts in them. Usually some
ordinary discretes. But it is the usual, very low qties.

Regards, Joerg