I was watching a video by devttys0 where he describes a Photodiode/Transimpedance Amplifier.

I was intrigued by a couple of elements in the design and I decided to do a simulation to try to understand what was going on. In particular I had never seen a diode connected JFET used as a limiter, and I'm still uncertain about how the reference circuit maintains such a low positive level.



Enjoy

 

I had never seen a diode connected JFET used as a limiter

That's a new one for me too. I've just had a play and found that certain Schottky diodes have a similar soft limiting effect at very low (sub-1u) currents.

 

That's a new one for me too. I've just had a play and found that certain Schottky diodes have a similar soft limiting effect at very low (sub-1u) currents.

A supposed benefit of the diode connected JFET is very low reverse leakage. Haven't tested that with the model, but it is on the todo list.

 

The guy in the video says he chose diode-connected JFETs instead of ordinary silicon diodes because of their lower reverse leakage; and indeed, the specified Igss of an MMBF4117 is about 1000X lower than the Ir of a diode such as a 1N4148 (10 pA vs. 25 nA).

BUT: the diodes in this circuit are operated in forward bias, not reverse, so I don't see where the JFET's lower reverse leakage would confer any benefit. Am I missing something?

 

The guy in the video says he chose diode-connected JFETs instead of ordinary silicon diodes because of their lower reverse leakage; and indeed, the specified Igss of an MMBF4117 is about 1000X lower than the Ir of a diode such as a 1N4148 (10 pA vs. 25 nA).

BUT: the diodes in this circuit are operated in forward bias, not reverse, so I don't see where the JFET's lower reverse leakage would confer any benefit. Am I missing something?

I don't know for sure, but at very low output voltages are we sure all the diode connected FETs are all forward biased? It sounds like he is trying to insure that no current from the input can bypass the 10Meg feedback resistor.

 

Perhaps a low reverse leakage also means the gate diode will operate ideally at a lower forward current than a standard diode.

 

Perhaps a low reverse leakage also means the gate diode will operate ideally at a lower forward current than a standard diode.

I wondered about that, too. But if that was his rationale I would have expected him to say so; instead, he specifically cited the lower reverse leakage of the JFETs.

 

Mea culpa! Mea Maxima Culpa! There is a mistake on the schematic for the Transimpedance Amplifier. The net label on the right hand side of R2 was supposed to be "V+". Somehow(?) it ended up as "+V", which exists in no other place on the drawing. Since one end of R2 was floating, this meant that the reference voltage on the non-inverting input was about 187 nV instead of the intended 228 mV. Now the universe is back in balance. I should have trusted my instincts before posting, instead of believing in, dare I speak the word -- magic.

Sorry

 

i searched the Google for iR blaster development (in order to spot the navigation rules for the zoo of encodings avail.) - the only thing that included such was a generic match http://ww1.microchip.com/downloads/en/DeviceDoc/21627c.pdf

? why :
-- there's no point of evaluating any characteristics of this amp. unless we know the signal(+env.-l bg.-noise) parameters and the TM-protocol(or variety) used - none of which is specified by the video
___________________
random links about VL-/iR-comm
https://www.ijeat.org/wp-content/uploads/papers/v8i6/F8523088619.pdf
https://www.ti.com/lit/ds/symlink/log101.pdf

 

i searched the Google for iR blaster development (in order to spot the navigation rules for the zoo of encodings avail.) - the only thing that included such was a generic match http://ww1.microchip.com/downloads/en/DeviceDoc/21627c.pdf

? why :
-- there's no point of evaluating any characteristics of this amp. unless we know the signal(+env.-l bg.-noise) parameters and the TM-protocol(or variety) used - none of which is specified by the video

I was primarily interested in two things:

1. The use of diode connected JFETs as a soft limiter
2. What I though was an unusual bias circuit. Not in the arrangement of components, but in what I thought was the claim about getting very close to GND. Turns out the only thing unusual about it was the mistake I made in drafting the circuit. Took me the better part of a day going back over the things I thought I knew and eliminating possibilities until I stumbled on the solution. There are very few better things to do these days.

I never did look for the video made by his friend who actually used the board. If I stumble across it, I'll post a link.

The suitability of this particular amplifier for other applications is on the todo list. I suspect some of the things you would look for would be in the datasheet, and looking at it is also on the todo list.

SIDE NOTE: I like to add things to my SPICE libraries just for grins and giggles. You never know when something might come in handy. I can almost do it with one eye open and my right hand behind my back.

 

the circuit with iC-s is at 14:35 (must have a deeper look what it does)

 

the circuit with iC-s is at 14:35 (must have a deeper look what it does)

Yes, I only did the amplifier part and ignored the comparator part.

 

I viewed the remainder of the video and near the end he talks again about the choice of JFETs rather than ordinary diodes; he goes on again about the need for low reverse leakage current but never discusses any possible corollaries to reverse leakage there might be in the forward conduction mode. So it sounds to me like his concern is for reverse leakage, period.

It kinda looks to me like he simply latched onto "low reverse leakage" as a potential issue and never stopped to ask himself whether those diodes/JFETs were ever actually reverse biased.

 

i made some spice test with the "simplified mathematical" Ph-D model
it did not end up with any "WOW!-s" ... the simple transistor amp. seems to perform better (maybe not as economical as ®)
/// it should be LOG-Amp however ... with DC decoupler . . . . +depending on "many" the comparator must be a "true" autobiasing ... the 3-rd fig.
NOTE :: the V1 (node : "E") sets the mW/cm² stimulus the Duty is 20% for 67kHz (was 33% for higher fq.-s ...)
NOTE.2 :: cracked the ® to make it more stable for the LT-Spice and for the random P-Spice Op-Amp models ...
.
.
.
.
.
... just checked -- the video never put's this developed gizmo into a real life test of any kind ??? unless the below 1kHz (@ the beginning) was the cell phone capture by it ?

 

@ci139 I didn't understand much of what you were trying to say, and your pictures chopped off part of the circuit

(maybe not as economical as ®) <-- what does this mean?
/// it should be LOG-Amp however ... with DC decoupler . . . . +depending on "many" the comparator must be a "true" autobiasing ... the 3-rd fig. <-- what does this mean?
cracked the ® to make it more stable for the LT-Spice and for the random P-Spice Op-Amp models ... <-- what does this mean?

 

means.1 : the Li-coin gizmo (the detector circuit in the video e.g. "the ®") uses below 1mA (my transistor alternative uses about 4mA)

means.2 : basically the ln() of the photodiode input gives you better coverage over different distances if you scale it to suit the comparator input , however -- the comparator may develop the DC accumulation/-drift on it's inputs depending on the pulse and packet wave-form variations , so you need to "force" the threshold to the median of the signal

means.3 : if you use mathematical device models (or P-spice device models) in LT-spice , then they tend to response fast - which means - they tend to have a small amplitude HF noise component , but they also may spike at large amplitudes - so - if you don't have time to convert them to have a more mild transient response - you can (as an alternative) modify the circuit so it better suppresses any fast events (e.g. "crack the circuit")

((i'm getting tired of not finding any smart phone iR comm captures))
but somewhat related to the Ph-D detection in the video in the OP

 

Now I understand. Thanks for the clarification. Just for the record, I've never believed that simulation was a substitute for building real circuits, I do find it more efficient for investigating things I've never worked with before. I agree with "no WOWs", just some things I had not seen before. Interesting to me, maybe not so much for others. Thanks for the contribution.

I did spend a great deal of time writing firmware to decode the NEC protocol for a family of Wadia Remotes used in their product line.