Help designing a low-noise TIA

On a sunny day (Thu, 06 Jun 2019 07:55:37 -0700) it happened John Larkin wrote in :

Trannies are good!

My old favorite universal NPN BC109C hFE max 900 (good species) Zin 6kOhm (best species)

voltage gain: ((900 * (200E-9 / 6000)) * 500e3) / 200E-9 = 75000 for a 500 k collector resistor

But how far are we leakage limited for that value collector resistor? The collector cut off current at room temperature is 1 nA

1e-9 * 500e3 = 0.0005 = 500 uV

Your 200 nV amplified by 75000 is:

-> 200e-9 * 75000 = 0.015 = 15 mV

So we have .5 mV offset

OK than we increase the collector resistor to 10 M, a factor 20. Then we have 20 x 15 mV = 300 mV signal, usable. The offset is now 20 * .5 mV = 10 mV no problem on a 9V supply. The actual gain is 20 x 75000 = 1.5e6.

I am sure there are better trannies than that one. And you can always go darlington.

But I give you modern opamps can do a bit better, but so can adding an extra stage.

Max beta found 1000:

formatting link
\u03b2-in-bjt-current-amplification

Anyways. just for the sake of math,, I no longer have BC109-C or even any BC109 here.

We do not have a speed limitation I think, signals are slow.

Just dreaming.... :-)

Reply to
Jan Panteltje
Loading thread data ...

Sorry, Jan, but that's all silly.

Build one. Or even Spice one.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Hi all, Again, thanks for the great ideas. I will try and keep up with them all. The original Tolman-Stewart used this rotating coil and braking; they re-di d their experiment about a decade later with a constantly rotating system a nd then everything moved to ac and was easier. The recent papers from the

80s used the ac system. The problem is, the physics is harder, and we want this to be for undergrads. So we like the braking experiment better.

Re: noise reduction. We can try and reduce the noise by turning off monito rs and running at night. Ideally, we?d like to be able to run this as part of our undergraduate advanced lab, so making it an extremely low-n oise environment is not ideal. But we can try. We?ve actually tri ed mu-metal, and it is not great at keeping out the magnetic fields. We? ??re currently considering just adding a 60 Hz notch filter.

We haven?t yet seen the signal we want, precisely because right now the noise swamps the TIA, and brings it to the rails.

George, The dc offset when it is just sitting there usually sits around a v olt or so. We?re working with low-noise op-amps, and so they are n ot trimmable. We were using the LT1792 and that one we could trim and that worked well. We can and have tested the circuit with an actual honest-to- goodness current source, and with two coils, where we send a pulse into one coil and pick it up in the second coil. The first experiment works well, the second is inconclusive so far.

As for connections to the coils, to contact the rotating coil we use a rota ry feedthrough that one of my students found on the web:

formatting link
c.com/html/110.html . The stationary coil, we just bring the contacts out and onto the bench. None of the electronics is attached to the coils, it i s all on the bench. We connect our two coils in series when the wires come to the breadboard.

Jan Panteltje, I am intrigued by your suggestion, as we initially wanted to integrate the current pulse. Our thought was that integrating would allow us to integrate over the noise and be left with only the slower 0.5 s puls e. But in order to get a large current, the input impedance of the measure ment system has to be small. This is why the TIA, with its virtual ground at the negative input is nice, acts as zero input impedance for the current pulse. What would be the input impedance of your transistor circuit be?

Whit3rd, We started first using a charge amplifier exactly as you described , but without the switch. So to prevent pinning the op-amp we added a feed back resistor also. The system we had was very unstable, even for short pe riods of time. But we did not try it with a switch to keep the capacitor f or charging. You are right that the math is easy if we can integrate the c urrent to get the charge. Then we only need to know the starting speed and the total charge.

John Larkin, I have limited experience with a differential amplifier. Can those be built to reject the 60 Hz noise?

Reply to
ithacacollegephysics

I am puzzled by that. Have I missed something. I thought op-amps without offset adjust trim pins could usually still be trimmed by injecting a small trim bias into one of the inputs?

piglet

Reply to
piglet

trimmable....

We have not tried that. The LT1792 uses one of its extra pins with a varia ble resistor between one rail and ground to trim the offset. I haven't tri ed it into the input.

Reply to
ithacacollegephysics

Yes, the second coil co-located with the moving one can be differenced and the only output effect will be low (because of common-mode rejection ratio); if it's KNOWN to be 60 Hz noise, you can also gate your outputs at some multiple of 1/60 second (for 50 and 60 Hz, a multiple of 1/10 second is preferable) which should reject a large portion of that pickup. Your signal is DC, but the pickup goes positive and negative so an integer number of cycles sums to... zero.

Common-mode rejection, like matching of the coils, has errors (resistor value errors, in the case of the diff amp), but unlike coils-in-series connection, it is possible to trim the coil difference if you feed the two coils to different input pins. It might be reasonable to gate-alike two charge amplifiers, one for the driven coil and one for the static coil, and difference-amplify the outputs. Charge amplifier gain depends on capacitor value, that'd be an obvious place to gain-change for balance of the two coils. results,

Your apparatus also has to reject AM radio input, as well as the more audible 60 Hz artifacts. I'm guessing that any nearby switchmode power (cellphones?) is deadly, too.

Reply to
whit3rd

Oh, I always figured he (Matt) had the comp coil in series. If subtracting afterwards then yeah you have to match both signal chains. (10% capacitors might ruin your whole day.)

I should see how much of the 60 Hz and other crud is canceled in our EF-NMR.

George H. It might be reasonable

Reply to
George Herold

Our compensating coil is in series with our rotating (signal) coil. But we could easily take them apart and put the two signals into a difference amplifier to subtract the signals afterwards.

There's plenty of 60 Hz noise in the room, as commenters have pointed out, so if I can get a variable capacitor I can easily tune the compensating coil to zero out the signal from the rotating coil, to be left with only the signal.

I don't have a good feel for the electronics: would that be more feasible than the transistor method from Jan?

Reply to
ithacacollegephysics

On Jun 6, 2019, snipped-for-privacy@gmail.com wrote (in article):

of

mu-metal,

currently

the

There is a standard trick for compensating integrators for opamp offset induced drift:

"Integrating Fluxmeter with Input Current Compensation to Cancel Drift", Thomas G. Chadbourne, US Patent 3,978,399 issued 31 August 1976.

This may be useful.

Joe Gwinn .

Reply to
Joseph Gwinn

On Thursday, June 6, 2019 at 12:48:53 PM UTC-4, snipped-for-privacy@gmail.com wro te:

The original Tolman-Stewart used this rotating coil and braking; they re- did their experiment about a decade later with a constantly rotating system and then everything moved to ac and was easier. The recent papers from th e 80s used the ac system. The problem is, the physics is harder, and we wa nt this to be for undergrads. So we like the braking experiment better.

tors and running at night. Ideally, we?d like to be able to run th is as part of our undergraduate advanced lab, so making it an extremely low

-noise environment is not ideal. But we can try. We?ve actually t ried mu-metal, and it is not great at keeping out the magnetic fields. We ?re currently considering just adding a 60 Hz notch filter.

ow the noise swamps the TIA, and brings it to the rails.

volt or so. We?re working with low-noise op-amps, and so they are not trimmable. We were using the LT1792 and that one we could trim and th at worked well. We can and have tested the circuit with an actual honest-t o-goodness current source, and with two coils, where we send a pulse into o ne coil and pick it up in the second coil. The first experiment works well , the second is inconclusive so far.

tary feedthrough that one of my students found on the web:

formatting link
tac.com/html/110.html . The stationary coil, we just bring the contacts ou t and onto the bench. None of the electronics is attached to the coils, it is all on the bench. We connect our two coils in series when the wires co me to the breadboard.

Huh, have other people used these (rotating contacts) for low noise measurements? Is there a spec? I'd want to try and measure the noise, low currents might be different than high.. (mA) currents.

George H.

to integrate the current pulse. Our thought was that integrating would all ow us to integrate over the noise and be left with only the slower 0.5 s pu lse. But in order to get a large current, the input impedance of the measu rement system has to be small. This is why the TIA, with its virtual groun d at the negative input is nice, acts as zero input impedance for the curre nt pulse. What would be the input impedance of your transistor circuit be?

ed, but without the switch. So to prevent pinning the op-amp we added a fe edback resistor also. The system we had was very unstable, even for short periods of time. But we did not try it with a switch to keep the capacitor for charging. You are right that the math is easy if we can integrate the current to get the charge. Then we only need to know the starting speed a nd the total charge.

n those be built to reject the 60 Hz noise?

Reply to
George Herold

Breadboard? At 200 nV?

The mercotac things use "liquid metal" contacts. What metal is liquid??

I'd expect some healthy thermoelectrics.

If the 60 Hz is from a real magnetic field thing, no. If it's electrostatic common-mode voltage, yes. 60 Hz can be filtered out as long as it's not so big as to drive things nonlinear.

I think your rig needs an electrostatic shield. I've seen foamy sheet stuff covered with aluminum foil.

In most rooms, the 60 Hz e-field is huge. Plus lots of other stuff.

Why not amplify and digitize the outputs of both coils?

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

On a sunny day (Thu, 6 Jun 2019 09:48:48 -0700 (PDT)) it happened snipped-for-privacy@gmail.com wrote in :

It looks like at least several kOhm, bit higher than I expected. Still it would experiment with it. But others may have more advanced^H^H^H^H^H^H^H^Hcomplicated solutions.

Not sure about TIA, TIA makes sense if the current response of your coil is more linear than the voltage response:

formatting link
Because you are basically shifting some electrons does this make sense in this case? And after all it is an artificial created low impedance created by feedback.

Some philosophy... Storm coming ;-)

Anyways the suggestion of a deserted island with .. should help against the mains hum and cars.

Reply to
Jan Panteltje

Wrong question. Amplify, certainly, but... digitize? To what end?

Or, do you mean apply a digital voltmeter?

If only a computer screen can display the results, the physical effect hasn't really been convincingly demonstrated to the classroom.

Reply to
whit3rd

Hi all again.

Joseph Gwinn, I will take a look at the patent to see how they compensated for the op-amp drift.

Regarding the rotary connectors, there is not a lot of noise information on their website. We moved to a liquid metal rotary feedthrough (the liquid metal is mercury) to remove problems with noise that comes from brushes. I don?t know what the noise specs are. We are set up to run it all using wire that twist up, but that is hard to run tests like that. You get one try, and then have to replace the wires.

Jan Panteltje, several kOhm of input impedance will reduce our expected cur rent by a factor of ten or more. So I am not sure your design will work. We?d drop our expected current down to 100 pA, or less.

John Larkin, we did have the rig inside mu-metal for a while, but that was hard to maintain and did very little for the static (Earth?s) field that we also have to get rid of. (If you don?t get rid of the Ear th?s field, when the coil is spinning the centrifugal force makes t he coil slightly bigger, and then when you brake it it gets smaller, leadin g to a signal that looks like the one we are trying to see.) So we abandon ed that. But we haven?t tried electrostatically shielding the enti re setup. We put our breadboard into a metal box to reduce noise ? are you suggesting we box up the whole thing into an electrostatic box? W e had a real one of those when I was in grad school, a screened room. Is t hat necessary for the coils as well?

Ideally, I?d like to be able to see the integrated pulse or the act ual V vs t of the pulse on the oscilloscope screen for the students to be a ble to see it before digitizing it. But if I have to digitize and then pla y with the signal, I will.

Reply to
ithacacollegephysics

On a sunny day (Fri, 7 Jun 2019 03:59:16 -0700 (PDT)) it happened snipped-for-privacy@gmail.com wrote in :

This is the interesting part. That 100 pA (1e-10) will flow into the base junction of that transistor and if it is a high beta type here for the simplicity of numbers not hFE 900 but 1000, result in a collector current of 1e-7, or .1 uA. In a 10 MOhm collector resistor that will give 1 V (1e-7 * 1e7).

There seems to be a basic misunderstanding about how opamps work, opamp output is the result of an input voltage (or current) being amplified, in the TIA case that voltage is then feedback via a resistor to bring the input back to zero, ALMOST zero. This gives the illusion of a low input impedance. In fact the real opamp impedance may be very high. There is a delay, so overshoot at the output, as it takes TIME to amplify the signal. What did you gain? speed ? But speed in this context is very very low.... The bit of overshoot gives better speed impression.

It is the same as the old logic feedback counter that as kid I could not figure out, until I realized it can only work because logic gates have a delay.

sigh

This is apart from issues such as linearity caused by the nice omhs feedback resistor of course. :-)

Reply to
Jan Panteltje

To see if the signal is real and not an artifact. To *see* what the artifacts are. To filter and signal process with matlab or something. To signal-average many runs. To subtract out the stationary coil signal with tunable math.

No, too slow to show the waveforms.

Don't integrate! Don't use a charge amp!

And to demonstrate a deceleration-rate-dependant pulse. And to illustrate the basics of signal processing, which does tend to use computers nowadays.

Why would a properly shaped pulse on a screen be less convincing than a static voltage at an opamp output? I wouldn't accept that in this case.

This is a great demo experiment, because there are so many hazards. It's not going to be a great demo if it doesn't ever work.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

This looks to be an extraordinally difficult experiment.

Magnetic shielding is probably not practical here. The compensating and helmholtz coils can help with mag fields. If electrostatic coupling is squirting noise into the system, it's easy to test that: temporarily shield the area with grounded aluminum foil or something and see if it makes any difference. Also look at the amp output with a scope when nothing is spinning.

I will say, for the last time, that the basic physics makes voltage, and measusing current was an artifact of hundred-year-old instrumentation limits, and the TIA is a bad idea.

Good luck getting this to work.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Yeah, well, you forgot about the h_oe output conductance, which isn't zero. There is effectively a resistor of a few 10 kOhm in parallel with your 10 MOhm collector resistor, limiting the gain. Somebody already told you.

Jeroen Belleman

Reply to
Jeroen Belleman

And some basic misunderstandings of how transistors work. Higher beta does not increase the voltage gain of the usual transistor amplifier.

200 nV will not push 100 pA into the base of a transistor whose collector current is around 1 uA. The impedance looking into the base will be 10s of megohms.
--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Spice does transistors pretty well.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.