Electronics > Projects, Designs, and Technical Stuff
Photodiodes saturating in ambient light
<< < (6/10) > >>
tggzzz:

--- Quote from: SparkyFX on May 12, 2020, 12:44:38 pm ---You could try a polarizing filter in front of the receiver, those filter out ambient stray light very efficient. I can't tell if all lasers are polarized in general or you need to arrange the laser in a specific orientation, could you work with that?

--- End quote ---

A neutral density filter with an attenuation of 10dB will filter out exactly as much ambient light as a 10dB polarising filter. It will only make a large difference if the ambient light is polarised in one appropriately - which is unlikely.
aussie_laser_dude:

--- Quote --- You could try a polarizing filter in front of the receiver
--- End quote ---

Polarisers are a nice idea, yes lasers are highly polarised and ambient light usually is not. Unfortunately it's not easy to implement in this particular design where polarisation angle and detector orientation are constantly changing. No big loss though, an optical bandpass filter and a high pass ac filter will give pretty impressive results. Simple photodiodes with an ac filter without optical filtering is already giving very clean waveforms as seen in the pics, they're just too small (~5mV100mV) for the adc to read, needs amplification.

  I'm a noob at circuit design, can get my head around bjt / op amp part selection but can't understand mathematically or intuitively how to select the right transformer and load resistor?
twospoons:

--- Quote from: aussie_laser_dude on May 12, 2020, 10:21:46 am ---Could I get some help with understanding twospoons small pulse transformer photovoltaic circuit in "reply 16". Transformers scare me, I don't understand transformer details very well.


I'm a noob at circuit design, can get my head around bjt / op amp part selection but can't understand mathematically or intuitively how to select the right transformer and load resistor?

--- End quote ---
This is a great opportunity for self-learning, and those should never be passed by  ;D

1. Think about this as a current transformer and everything becomes much clearer. In fact its really important to think of this whole thing in terms of a signal current.
2. 1:1 will work just fine. Construction should be done using bifilar winding for best coupling. The core can be very small - but should ideally have an airgap to avoid saturation. The core I used was a 6mm potcore - really tiny.
3. You really want to feed into a low impedance to get the bandwidth up, think about amplifying current, rather than voltage. The conversion to a voltage signal can be done later. The transimpedance circuit I posted does this.
4. Driving into a low impedance means the voltage across the the diode changes very little, which is good because you are then not trying to charge the self capacitance of the diode, which would otherwise reduce the signal.
5. Choose a reasonably high frequency ferrite for the core. Something intended for pulse transformers, or RF inductors or filters. Something like 4C65 or 4B2 would probably work well (Ferroxcube types - there will be equivalents from other manufacturers

As I mentioned in the first post - ignore the resistor values posted, they are just placeholder values and are not correct.

If you haven't already, its worth studying the photodiode equivalent circuit:


T3sl4co1l:

--- Quote from: aussie_laser_dude on May 12, 2020, 10:21:46 am ---Could I get some help with understanding twospoons small pulse transformer photovoltaic circuit in "reply 16". Transformers scare me, I don't understand transformer details very well.

My understanding is the transformer will have a limited bandwidth, I have no idea how to calculate transformer properties and load resistor R1 that will give a bandwidth of say >3 MHz for the photovoltaic photodiode.

Could R1 be changed from 1 Kohm to 10 Kohm without ruining the circuit bandwidth? My reasoning is that the large DC current from the photodiode won't be passing through the R1 resistor, so we can use a higher value resistor to get a higher potential difference. I'm worried this will sacrifice the bandwidth of the transformer or photovoltaic photodiode, but I'm too dumb to understand how to figure this out. I have no idea of the math. If anyone has a good link that explains these properties of transformers I'd love to have a look.

Also, could a 1:4 transformer be used, ie. 1mV AC in and 4mV ac out on other side? Would a transformer like this ruin the bandwidth somehow? Is it better to stick to 1:1 transformers?

Thanks for any help

--- End quote ---

No, actually you probably want a lower impedance to keep bandwidth higher.

Transformers also have a characteristic impedance, which should be matched for best results.  It may not need to be -- say if you have enough excess bandwidth that you're limited by other constraints in the circuit, like PD capacitance.

Roughly speaking, you can calculate the bandwidth of the photodiode by its capacitance, and the load resistance:
F_H = 1 / (2 pi R C)
This is a lowpass cutoff, so that you can't expect to get much gain at frequencies above this (indeed, the gain will be dropping ~asymptotically with rising frequency, i.e. -20dB/dec), and gain will be flat below.

This is a good reason to employ PD bias: the capacitance drops maybe 3x or more under bias, therefore increasing bandwidth by as much.  (Again, if you don't need it, it's okay not to, as in twospoons's case.)

The transformer, conversely, has a highpass cutoff:
F_L = R / (2 pi L)

F_H and F_L seem to be named backwards, huh?  Well, H is the higher (upper) cutoff, and L the lower.  F_H > F_L.  The upper cutoff has a lowpass characteristic, and the lower has a highpass characteristic.  "Pass" vs. "cut" terminology I guess.

The transformer also has another kind of cutoff: the series attenuating effect of leakage inductance (LL), and the parallel shunting effect of parasitic capacitance (Cp).  More generally, it has impedance and length -- it's wound from wire, after all, and wire has impedance and [electrical] length -- delay.  This too is a lowpass characteristic, so manifests as another F_H in the system.  You generally want this higher than needed; and, this shouldn't be hard to achieve at these modest frequencies.

Example: suppose your system resistance is 1k as shown (note that he just put in default values to show the connection, with no intent of these being real values!).  Suppose your photodiode is, oh I don't know, 100pF say.  100pF and 1kohm rolls off at F_H = 1.59MHz.

To get a bandwidth from say 10kHz to 1.59MHz, you need a transformer that gives F_L at most that, and that has small enough LL and Cp that its cutoff exceeds 1.59MHz.  This needs,
L = 15.9mH
LL << 100uH
Cp << 100pF

As it turns out, LL will be the easy one; more likely ~1uH will be seen.  Cp however is easily taken up by a few meters of wire windings, so you will prefer a hi-mu core (typically ferrite or nanocrystalline) with a fairly thick cross section, so as to maximize winding inductance (L) while minimizing wire length (which corresponds to LL and Cp).  Thick toroids, pot cores and other shapes are typical options.  Twisted pair wire is probably fine.

Note that a transformer isn't needed at all, if you don't need isolation!  You can simply use an inductor of, right about 15.9mH, in parallel with the photodiode, to set the high pass cutoff without needing a transformer at all.


Step-up:
Certainly, you can get ratios from a transformer!  Or again, if isolation isn't necessary, it can be an autoformer, which is nice to save a few turns.

We're still limited by the same resistance * capacitance at the photodiode.  Transformers with high impedance secondaries also get much harder to construct -- impedance goes as turns ratio squared, but Cp doesn't drop, indeed it tends to keep going up (with wire length).  So the transformer bandwidth gets worse as you go in this direction.  More than a few kohms at this bandwidth is actually challenging.

To what end, do you need the impedance matched, anyway?  Surely if the impedance is simply low to begin with, you would use a lower impedance amplifier as well?  Say, an op-amp with rather beefy inputs that gives refreshingly low e_n (nV/rtHz) but annoyingly high i_n (~pA/rtHz?), or a discrete low-noise amplifier or purchased module that is best suited to quite low impedances, say 50 ohms.  Indeed you might get better performance (lower noise and higher bandwidth) this way, than with a transformer ratio up to a more jellybean (TL072??) amplifier.

The gold standard for photodiodes of course is a low noise transimpedance amplifier; this is just an amplifier configuration that best suits the diode's properties.  It's best used when placed as near to the diode as possible, i.e., with no transformer or connecting cables delaying its response.

I guess that low noise isn't actually a high priority here, as your signal is fairly strong (a direct laser beam I guess?), and you don't seem to care about ambient noise sources; or maybe you just aren't to the stage where it matters yet.  (Sunlight may be fairly stable, but is subject to fading*, and ambient sources (fluorescents, LEDs; even incandescents to a lesser degree!) often fluctuate at mains frequency.)

*Fading is the radio-frequency effect of speckling or twinkling transmissions and reflections.  In the shortwave band it manifests as ionospheric layers moving about, and what's fading is the signal, which comes and goes over time, sometimes pretty deeply (+/- 10dB or more).  The same thing works optically with diffraction (twinkling stars!), speckle (coherent (laser) reflection from disorderly surfaces?) and so on.  Plus the more obvious influences to sunlight: clouds casting shadows, cars shining reflections through windows, etc.

Tim
twospoons:

--- Quote from: aussie_laser_dude on May 12, 2020, 08:28:01 pm ---
Polarisers are a nice idea, yes lasers are highly polarised and ambient light usually is not. Unfortunately it's not easy to implement in this particular design where polarisation angle and detector orientation are constantly changing.

--- End quote ---

The solution there is a circular polarizer, which should work well given the narrow linewidth of a typical laser. Just remember that reflection will reverse the handedness of the polarization.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod