Forums

Phototransistor fed to a transimpedance amplifier

Started by SilverLeo July 17, 2016
Hi all.

I'm running through some test of a phototransistor signal conditioner, and I'm facing with wild noise issues.
The conditioning is a basic TIA stage: http://i.imgur.com/PcgglKZ.png
Feedback resistor is 5K, feedback capacitor is 470pF, negative bias is -12V, and photocurrent is in the range 8uA-800uA. The bandwidth should be about 50kHz. Opamp is OPA2277UA.

Problem is that phototransistor if far away from TIA, with four meters of cable running along with power cables (valves, actuators, motors in the 5kW range). This causes tenth of uA of noise well inside the signal bandwith.

The same phototransistor, with same cabling, works great with the simple resistor between collector and supply rail: http://i.imgur.com/Y1C2vEA.png

TIA is better for me because it will keep phototransistor far away from saturation, but maybe the finite opamp bandwidth is actualy amplifying noise.
 
Hints?



Why not put the TIA at the sensor?

Putting long leads on the summing junction of an op amp is a guaranteed way of getting into trouble. 

With equal resistors and the same power rails, your TIA and load resistor circuits will saturate at about the same light level.

What's it for? 

Cheers

Phil Hobbs
In article <a9c78249-f82e-4e46-9df2-1e786196ea2b@googlegroups.com>, 
eugenio.newsgroup@gmail.com says...
> > Hi all. > > I'm running through some test of a phototransistor signal conditioner, and I'm facing with wild noise issues. > The conditioning is a basic TIA stage: http://i.imgur.com/PcgglKZ.png > Feedback resistor is 5K, feedback capacitor is 470pF, negative bias is -12V, and photocurrent is in the range 8uA-800uA. The bandwidth should be about 50kHz. Opamp is OPA2277UA. > > Problem is that phototransistor if far away from TIA, with four meters of cable running along with power cables (valves, actuators, motors in the 5kW range). This causes tenth of uA of noise well inside the signal bandwith. > > The same phototransistor, with same cabling, works great with the simple resistor between collector and supply rail: http://i.imgur.com/Y1C2vEA.png > > TIA is better for me because it will keep phototransistor far away from saturation, but maybe the finite opamp bandwidth is actualy amplifying noise. > > Hints?
shielded balanced pairs? Jamie
On 07/17/2016 01:32 PM, M Philbrook wrote:
> In article <a9c78249-f82e-4e46-9df2-1e786196ea2b@googlegroups.com>, > eugenio.newsgroup@gmail.com says... >> >> Hi all. >> >> I'm running through some test of a phototransistor signal >> conditioner, and I'm facing with wild noise issues. The >> conditioning is a basic TIA stage: http://i.imgur.com/PcgglKZ.png >> Feedback resistor is 5K, feedback capacitor is 470pF, negative bias >> is -12V, and photocurrent is in the range 8uA-800uA. The bandwidth >> should be about 50kHz. Opamp is OPA2277UA. >> >> Problem is that phototransistor if far away from TIA, with four >> meters of cable running along with power cables (valves, actuators, >> motors in the 5kW range). This causes tenth of uA of noise well >> inside the signal bandwith. >> >> The same phototransistor, with same cabling, works great with the >> simple resistor between collector and supply rail: >> http://i.imgur.com/Y1C2vEA.png >> >> TIA is better for me because it will keep phototransistor far away >> from saturation, but maybe the finite opamp bandwidth is actualy >> amplifying noise. >> >> Hints? > > shielded balanced pairs? > > Jamie >
It's a single-ended circuit. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
On Sunday, July 17, 2016 at 1:44:50 PM UTC-4, Phil Hobbs wrote:
> On 07/17/2016 01:32 PM, M Philbrook wrote: > > In article <a9c78249-f82e-4e46-9df2-1e786196ea2b@googlegroups.com>, > > eugenio.newsgroup@gmail.com says... > >> > >> Hi all. > >> > >> I'm running through some test of a phototransistor signal > >> conditioner, and I'm facing with wild noise issues. The > >> conditioning is a basic TIA stage: http://i.imgur.com/PcgglKZ.png > >> Feedback resistor is 5K, feedback capacitor is 470pF, negative bias > >> is -12V, and photocurrent is in the range 8uA-800uA. The bandwidth > >> should be about 50kHz. Opamp is OPA2277UA. > >> > >> Problem is that phototransistor if far away from TIA, with four > >> meters of cable running along with power cables (valves, actuators, > >> motors in the 5kW range). This causes tenth of uA of noise well > >> inside the signal bandwith. > >> > >> The same phototransistor, with same cabling, works great with the > >> simple resistor between collector and supply rail: > >> http://i.imgur.com/Y1C2vEA.png > >> > >> TIA is better for me because it will keep phototransistor far away > >> from saturation, but maybe the finite opamp bandwidth is actualy > >> amplifying noise. > >> > >> Hints? > > > > shielded balanced pairs? > > > > Jamie > > > > It's a single-ended circuit. > > Cheers > > Phil Hobbs > > -- > Dr Philip C D Hobbs > Principal Consultant > ElectroOptical Innovations LLC > Optics, Electro-optics, Photonics, Analog Electronics > > 160 North State Road #203 > Briarcliff Manor NY 10510 > > hobbs at electrooptical dot net > http://electrooptical.net
Doesn't have to be single ended. The OP could run the emitter to another TIA and then the two TIA outputs to a difference circuit. Need to split the bias across the TIAs. Bringing the TIA out to the photo-transistor is usually the better solution for noise but it's not always practical or necessary. Differential TIAs is a lot more circuitry and runs into calibration issues with balancing the two TIAs.
>Doesn't have to be single ended. The OP could run the emitter to >another TIA and then the two TIA outputs to a difference circuit. >Need to split the bias across the TIAs. &nbsp;Bringing the TIA out >to the photo-transistor is usually the better solution for noise >but it's not always practical or necessary. Differential TIAs is >a lot more circuitry and runs into calibration issues with balancing the two TIAs.
I'm not convinced that hanging a long cable on the summing junctions of _two_ TIAs is a big improvement. Cheers Phil Hobbs
On Sun, 17 Jul 2016 12:17:33 -0700 (PDT), Wanderer
<wanderer@dialup4less.com> wrote:

>On Sunday, July 17, 2016 at 1:44:50 PM UTC-4, Phil Hobbs wrote: >> On 07/17/2016 01:32 PM, M Philbrook wrote: >> > In article <a9c78249-f82e-4e46-9df2-1e786196ea2b@googlegroups.com>, >> > eugenio.newsgroup@gmail.com says... >> >> >> >> Hi all. >> >> >> >> I'm running through some test of a phototransistor signal >> >> conditioner, and I'm facing with wild noise issues. The >> >> conditioning is a basic TIA stage: http://i.imgur.com/PcgglKZ.png >> >> Feedback resistor is 5K, feedback capacitor is 470pF, negative bias >> >> is -12V, and photocurrent is in the range 8uA-800uA. The bandwidth >> >> should be about 50kHz. Opamp is OPA2277UA. >> >> >> >> Problem is that phototransistor if far away from TIA, with four >> >> meters of cable running along with power cables (valves, actuators, >> >> motors in the 5kW range). This causes tenth of uA of noise well >> >> inside the signal bandwith. >> >> >> >> The same phototransistor, with same cabling, works great with the >> >> simple resistor between collector and supply rail: >> >> http://i.imgur.com/Y1C2vEA.png >> >> >> >> TIA is better for me because it will keep phototransistor far away >> >> from saturation, but maybe the finite opamp bandwidth is actualy >> >> amplifying noise. >> >> >> >> Hints? >> > >> > shielded balanced pairs? >> > >> > Jamie >> > >> >> It's a single-ended circuit. >> >> Cheers >> >> Phil Hobbs >> >> -- >> Dr Philip C D Hobbs >> Principal Consultant >> ElectroOptical Innovations LLC >> Optics, Electro-optics, Photonics, Analog Electronics >> >> 160 North State Road #203 >> Briarcliff Manor NY 10510 >> >> hobbs at electrooptical dot net >> http://electrooptical.net > >Doesn't have to be single ended. The OP could run the emitter to another TIA and then the two TIA outputs to a difference circuit. Need to split the bias across the TIAs. Bringing the TIA out to the photo-transistor is usually the better solution for noise but it's not always practical or necessary. Differential TIAs is a lot more circuitry and runs into calibration issues with balancing the two TIAs.
Is there an active equivalent of a common-mode choke / balun? -- John Larkin Highland Technology, Inc lunatic fringe electronics
>Is there an active equivalent of a common-mode choke / balun?
Interesting question. The collector side of a cap multiplier looks like a big inductor, so if you built a pair of them that shared the same RC on their bases and had similar betas, that ought to work--the differential signal would get through without needing to move the RC. Cheers Phil Hobbs
Il giorno domenica 17 luglio 2016 09:56:50 UTC+2, SilverLeo ha scritto:

> cut
I'm replying to myself: I apologize, this seems the only action Google groups let me to do. So: interesting the cap multiplier approach; I'll do some math, maybe it'll be handy in the future. I cannot move the TIA close to the phototransistor: it's a retrofit for an old device for estimate the density of filled paper tubes (filters for very specific application) through the detection of diffused light. The more filled is the tube, the less is the light diffused. Phototransistor is buried inside a cast iron frame, far away from any conditioning electronic. The main reason that lead to the TIA approach was the fact that pototransistor with classic resistor on collector will fall into saturation (Vce near or below Vcesat), actually messing up density estimation. A TIA, as far as the opamp keeps doing its magic, will force Vce=Vbias, running the phototransistor deep in active region. This in theory: in real world application the TIA output shows 50mV of noise voltage, that does not correspond to a 10uA of current noise, because with the simple resistor there is no evidence of it. This noise comes out when some valves or motors are running. TIA input impedance should be zero (as stated in books) because of virtual ground at inverting input. But, because emitter is at -12V, the phototransitor will see a variable resistor matched to drop exactly 12V for any current. So, for "small" currents this resistor becomes big", and for big current this resistor turns "small". Is this right,somehow? I could add a resistor from inverting input to ground (10K range) to set a limit to input impedance, and a capacitor in parallel to that resistor (nF range) to bypass everything the opamp will not be able to follow due to bandwidth limitation. Useless or not? If I cannot overcome this issue I'll switch back to old, classi resistor on collector; but, it would be nice to understand what's going on.
On Monday, July 18, 2016 at 3:28:59 PM UTC-4, SilverLeo wrote:
> Il giorno domenica 17 luglio 2016 09:56:50 UTC+2, SilverLeo ha scritto: > > > cut > > I'm replying to myself: I apologize, this seems the only action Google groups let me to do. > > So: interesting the cap multiplier approach; I'll do some math, maybe it'll be handy in the future. > I cannot move the TIA close to the phototransistor: it's a retrofit for an old device for estimate the density of filled paper tubes (filters for very specific application) through the detection of diffused light. The more filled is the tube, the less is the light diffused. > Phototransistor is buried inside a cast iron frame, far away from any conditioning electronic. > > The main reason that lead to the TIA approach was the fact that pototransistor with classic resistor on collector will fall into saturation (Vce near or below Vcesat), actually messing up density estimation. > A TIA, as far as the opamp keeps doing its magic, will force Vce=Vbias, running the phototransistor deep in active region. > > This in theory: in real world application the TIA output shows 50mV of noise voltage, that does not correspond to a 10uA of current noise, because with the simple resistor there is no evidence of it. This noise comes out when some valves or motors are running. > > TIA input impedance should be zero (as stated in books) because of virtual ground at inverting input. But, because emitter is at -12V, the phototransitor will see a variable resistor matched to drop exactly 12V for any current. So, for "small" currents this resistor becomes big", and for big current this resistor turns "small". Is this right,somehow? > > I could add a resistor from inverting input to ground (10K range) to set a limit to input impedance, and a capacitor in parallel to that resistor (nF range) to bypass everything the opamp will not be able to follow due to bandwidth limitation. Useless or not? > > If I cannot overcome this issue I'll switch back to old, classi resistor on collector; but, it would be nice to understand what's going on.
As Phil said, it's most likely the capacitance to ground on the inverting input that is causing the noise. (The noise at the non-inverting input see's a the feedback resistor and cap to ground.... that gives gain that grows as the frequency increases... (cap impedance decreases.) if it gets too bad you've got an oscillator.) What's wrong with the resistor? Can you make the bias (collector) voltage higher? George H.