Phototransistor versus Photodiode

[techoyaki]

Hi everyone,

I'm planning to create an IR TX/RX system and had some questions regarding the RX side where I'm trying to evaluate between IR phototransistor and IR photodiode.

From my basic understanding, a phototransistor is an 'all-in-one' package where it can both receive and amplify (with correct biasing with some impedance load) in a single package and can be directly connected to the ADC. In contrast, a photodiode requires an amplifier stage before the ADC but the biggest advantage of a photodiode is that it is much faster and can handle larger frequencies.

Are these the only major characteristics to consider when choosing between them (i.e., total number of components vs. frequency)?

Thanks for your help!

[peter-h]

Basically yes. It is a normal tradeoff between the CTR (current transfer ratio) and the speed.

Secondary factors may be stuff like max slew rate of the isolation voltage but that will be more package dependent.

[CatalinaWOW]

That is basically it.  You will find that in many applications you will need an amplifier for either choice.  And usually a filter to strip off the dc from ambient light.  You will also want to understand things like the viewing cone and other optical questions that are device, not technology dependent.

[TimFox]

An important detail about phototransistors is that the effective capacitance seen by the photocurrent is increased by the Miller effect (common to all triodes) when the collector terminates in a load resistor.
To reduce that effect, the collector can be terminated in a low impedance, such as the virtual ground of a shunt-feedback current-input amplifier or the emitter of a cascode connection to a second transistor.
The photodiode (diode) does not have that feature, but has much less output current than the transistor.

[Conrad Hoffman]

I think phototransistors might be noisier, but check that.

[CatalinaWOW]

I think phototransistors might be noisier, but check that.

They will be noisier under identical conditions by the square root of beta due to the background current which is usually the dominant noise.  That is more than made up for by the signal gain, beta.  Under dark conditions the relevant noise will be from the leakage current (2qi noise), which will be highly device dependent.    There is huge variation in the area of diode detectors, and the larger ones have correspondingly larger leakage.

[floobydust]

OP what is the carrier frequency for your IR system? You modulate the beam for immunity to sunlight, room lighting etc.
Phototransistors are very slow, if you wanted use a say 38kHz carrier like IR remotes use, that would not work. You would use a photodiode and amplifier, better as a module like TSOP38 so you don't have to reinvent the receiver.

[LaserSteve]

Enjoy some P.C.D. Hobbs.
Please see attached. His website has more.
Phil was once a good friend.

He has published some nice ways of compensating for the Miller Effect and noise issues.

Steve

[mikeselectricstuff]

PT's are best where you have quite a lot of light, e.g. photo-interruptor type  scenarios, and can produce a direct logic-compatible output with just a pullup resistor. PDs are more for lower-light levels, and for measuring light levels, and usually need some kind of amplification.

for higher-speed data type things, the issue with a PT is that as data rates increase, you need to reduce the pullup resistor value, which increases the amount of light needed. For close-coupled links up to a few kbaud ( e.g. an isolated data link on a DMM), they are a nice cheap & simple solution, but once you need either higher speeds or longer distances, a PD+amplifier will be needed.

For low data rates over longer distances, TV IR remote receivers are good, as they have an integrated amplifier & data slicer, and output a digital signal.
IRDA receivers/transceivers can also be a nice solution, but are getting increasingly harder to find as irda is basically dead.