Long-tailed JFET pair question

[Joerg]

Hello Phil,


[ ... ]

If you have 300 uA to work with, life is easy. A 200-ohm load will get
you to the shot noise limit--so something like a Mini Circuits T4-1
transformer running into a garden variety 50-ohm amplifier (with no load
resistor) will be the bee's knees from a SNR point of view. Gain
stability will probably be the limitation then.

Maybe I am too obsessed with noise here then. Gain stability won't be
such a problem because it's only changes we have to detect. The absolute
current value doesn't have to be logged since that is handled in the DC
section.

A cascode stage running into the summing junction of a current-feedback
amp would be another approach. That AD847 doesn't look as though it has
the GBW for a 100-MHz application.

It's the OPA847. That thing can do a GHz. Basically the new ones are all
too fast for this application. But that could be tuned down via Cf.

 

[Winfield Hill]

Phil Hobbs wrote...

Re-reading, you mean.

Wow, they must have got some more VC money if they can afford you. I hope
he still owns some stock by the time they're over the hump. Interesting
gizmo, that. (For everyone else: Stheno Corp. is a startup that has an
interesting technology for measuring chirality in biological molecules, by
way of very small amounts of optical activity. It ought to separate the
NIH from a small boatload of money if they can get it into production.)

We at the Rowland Institute may have wrapped up much of that area,
and for non-commercial open, widespread public use. For example,
see where Fresnel's 181-year-old prediction is finally full-filled,
Phys. Rev. Lett. 97, 173002 (2006), http://focus.aps.org/story/v18/st14
http://link.aps.org/abstract/PRL/v97/e173002. Note, Peer Fischer's
technique can work well for very small sample volumes. And there's
much more of that coming from his Rowland lab. Y'all watch out now!

 

[Phil Hobbs]

Maybe I am too obsessed with noise here then. Gain stability won't be
such a problem because it's only changes we have to detect. The absolute
current value doesn't have to be logged since that is handled in the DC
section.



It's the OPA847. That thing can do a GHz. Basically the new ones are all
too fast for this application. But that could be tuned down via Cf.

Whoops, right. I knew that.

With a 300 kelvin load resistor, dropping 200 mV gets you to 1 dB above the
shot noise--and it's the AC resistance times the DC photocurrent that's
relevant, if you're using a transformer. Above there, the NF is pretty well
inversely proportional to voltage drop--400 mV gets you 0.5 dB, and so on.

Cheers,

Phil Hobbs

 

[Robert Latest]

On Mon, 27 Nov 2006 17:58:05 GMT,

In case this is for a photodiode: FETs are often only used to decouple
the high photodiode capacitance from the TIA that follows at the drain
node of the FET. Buys bandwidth.

There is no following TIA. The drains go straight into the +/- inputs of
the following opamp. Feedback is via high-value resistor back to the summing
junction (the non-grounded FET gate), so the whole thing is a TIA.

robert

 

[Robert Latest]

On Mon, 27 Nov 2006 18:44:21 +0100,
Fred Bartoli wrote

Robert Latest a écrit :

But at 2mA/V you'll still at least 1.5K/2K drain resistors for the JFETs
to noise to be dominant.

For the TIA you opened, 220R resistors means:
- either that the noise performances aren't optimal
- either that the JFETs have higher gm, which isn't optimal either.

You confirmed analytically which had been my hunch supported by some
wild guessing. For the OpAmp following the JFETs I had thought about
using an LT1097 which isn't so great noise-wise but can be overcompensated
externally in case my combined loop gain causes stability trouble.

robert

 

[Tony Williams]

Out of curisoity I opened up a commercially available TIA to see
how they did it and if I could learn from their design. I found
a dual JFET running with Ids=3.5mA and Ugs=-1.7 V. It's probably
something like a 2N5912 (it's in a 7-leg TO78 can, part #
scraped off). The odd thing is: the drain resistors are only 220
ohms off a 9V rail. The parts (at least the ones I can find)
that fit this Vgs/Ids combo have a few mS transconductance which
would get the input stage's gain barely over unity. Parts with
significantly higher gm run at tens of mA in this gate voltage
range.

I used to work for a Co doing discrete opamps,
(mumble, mumble) years ago. The jfet opamps
were not my designs, but afair the drain loads
were always pnp current mirrors of some sort.

 

[Joerg]

Hello Phil,

Whoops, right. I knew that.

With a 300 kelvin load resistor, dropping 200 mV gets you to 1 dB above
the shot noise--and it's the AC resistance times the DC photocurrent
that's relevant, if you're using a transformer. Above there, the NF is
pretty well inversely proportional to voltage drop--400 mV gets you 0.5
dB, and so on.

The only fly in the ointment is probably going to be the DFB module. The
datasheets on those are pretty skimpy. Beside the RIN all they tell
you is that the line width is 10MHz max. That's a whole lot and who
knows what kind of noise is in there. AM noise can be handled, if
necessary with a differential amp where the MPD gets fed into the other
input. But that won't help with phase noise :-(

 

[John Larkin]

The other approach is to pick a low-noise 50-ohm amplifier and connect the PD
right to the input--I've done that very successfully with 60-cent Mini
Circuits amps. Since you're AC-coupling anyway, a transformer is another
approach.

The Sirenza SiGe parts are very quiet. One of them actually has a
50-ohm input impedance, a rarity among 50-ohm mmics. The MiniCircuits
ERA-series tends to run in the high 30's. Depends on output loading
and operating current, of course.

John

 

[Rich Grise]

I have used them in the past. Was ok for not too difficult jobs but when
in need for someone with skills to solder 0402 parts and hotrod RF
transistors the results were mixed. Hiring engineers that way did not
work at all.

I've been a temp, on almost exactly the same type of work, except they
found out I was a tech, and started using me for tech work, at temp pay.
(well, $8.00/hr is infinitely better than $0.00/hr., so I shouldn't
snivel too much. )

Tell the agency exactly what you've said here, and someone will show up.

And you really don't need to know any electronics to do teensy soldering
work - all you need is good eyes and a steady hand.

Good Luck!
Rich

 

[Joerg]

I've been a temp, on almost exactly the same type of work, except they
found out I was a tech, and started using me for tech work, at temp pay.
(well, $8.00/hr is infinitely better than $0.00/hr., so I shouldn't
snivel too much. )

Tell the agency exactly what you've said here, and someone will show up.

Thing is the temp agencies usually don't understand. They don't know
what 0402 really means. I've had folks who were claimed to be able to
handle "fine stuff" and it didn't work at all.

And you really don't need to know any electronics to do teensy soldering
work - all you need is good eyes and a steady hand.

Err, it does require a basic understanding of ESD. IOW some training.

Good Luck!

Thanks, we'll need it.

 

[Michael A. Terrell]

Thing is the temp agencies usually don't understand. They don't know
what 0402 really means. I've had folks who were claimed to be able to
handle "fine stuff" and it didn't work at all.


Err, it does require a basic understanding of ESD. IOW some training.


Thanks, we'll need it.

Too bad that you're on the other side of the country, and that my
eyesight is getting so bad. I did a lot of SMD rework under a stereo
microscope. I routinely worked with 1206, 0805 and 0402 components,
along with ICs and other SMD components.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Phil Hobbs]

Joerg wrote:

The only fly in the ointment is probably going to be the DFB module. The
datasheets on those are pretty skimpy. Beside the RIN all they tell you
is that the line width is 10MHz max. That's a whole lot and who knows
what kind of noise is in there. AM noise can be handled, if necessary
with a differential amp where the MPD gets fed into the other input. But
that won't help with phase noise :-(

Phase noise nearly always dominates the line width of lasers, because
the total pump current sets an upper limit on the output power. I
gather you're building an interferometer, otherwise the frequency noise
isn't a problem.

Cheers,

Phil Hobbs

 

[Joerg]

Joerg wrote:



Phase noise nearly always dominates the line width of lasers, because
the total pump current sets an upper limit on the output power. I
gather you're building an interferometer, otherwise the frequency noise
isn't a problem.

Something like that, and phase noise is going to be critical. My gut
feeling is that it will be the limiting factor here. It'll be interesting.

The strangest thing for an EE to work with optical engineers is that
they think in wavelengths, we think in frequencies. Almost like driving
in England or Ireland. Had a PID regulator backwards this morning but
caught it.

 

[Phil Hobbs]

Something like that, and phase noise is going to be critical. My gut
feeling is that it will be the limiting factor here. It'll be interesting.

The strangest thing for an EE to work with optical engineers is that
they think in wavelengths, we think in frequencies. Almost like driving
in England or Ireland. Had a PID regulator backwards this morning but
caught it.

Physicists think in frequency, but use the opposite Fourier transform
sign convention. It's almost as bad.

Diode lasers are also exquisitely sensitive to back-reflections--those
noise specs probably assume a test setup with a two-stage Faraday
isolator. Anything worse than about a 60 dB return loss will make that
puppy howl.

One good thing about DFBs is that you can FM them very well and very
fast, so if you can arrange the measurement so that the signal comes in
offset in frequency, you can filter out most of the junk. What remains
is multiplicative phase noise. Assuming the signal large compared with
the shot noise, the SNR will be constant with signal level, rather than
disappearing into the grass.

A typical method for doing this is sinusoidal FM, such that (Delta
f)(t_path_difference) = 2.405 radians, the first Bessel null. About
there, the first and second harmonics are equal in size and together
amount to about 85% of the total electrical power.

Cheers,

Phil Hobbs

 

[Joerg]

Physicists think in frequency, but use the opposite Fourier transform
sign convention. It's almost as bad.

Diode lasers are also exquisitely sensitive to back-reflections--those
noise specs probably assume a test setup with a two-stage Faraday
isolator. Anything worse than about a 60 dB return loss will make that
puppy howl.

Yes, that issue has me worried a bit. Also, I have read that pigtailed
versions make life a bit easier for reflections.

One good thing about DFBs is that you can FM them very well and very
fast, so if you can arrange the measurement so that the signal comes in
offset in frequency, you can filter out most of the junk. What remains
is multiplicative phase noise. Assuming the signal large compared with
the shot noise, the SNR will be constant with signal level, rather than
disappearing into the grass.

A typical method for doing this is sinusoidal FM, such that (Delta
f)(t_path_difference) = 2.405 radians, the first Bessel null. About
there, the first and second harmonics are equal in size and together
amount to about 85% of the total electrical power.

Thanks, Phil. I will keep that in mind. Right now it's all new and I am
not sure whether we can do FM, considering the signal BW of up to around
100MHz. But it would be no problem to provide a really wide bandwidth
receiver so we can play some nifty filtering/demodulating tricks
afterwards. At least then I'd be back in the RF world, my home turf ;-)