Looking to have the group do a scrub on the last little project I built: A super-sensitive light sensor. The design "seems" to work, but I'm not sure that it can't be made better. The power source, listed as P1, is a PICOR Cool-Power Series Evaluation Board hosting their PI3111 DC-DC converter. The input to the PI3111 device is 24VDC. My concern is that there are so many grounds. Does anyone see anything glaring? Thanks!!!!!!

 

Nothing wrong with the schematic, merely amazing to get that quality at that impedance. Don't get it hot or the input bias current will have the output walk all over the place.

 

We can't advise you until you tell us what you are trying to measure.
Moonlight? Starlight? Photons? Continuous? Event Counting?

 

Look at the BPX61 sensor. 0.9 electrons per photon. With that, you could measure a candle at 2 miles.

 

We can't advise you until you tell us what you are trying to measure.
Moonlight? Starlight? Photons? Continuous? Event Counting?

My sincere apologies, would have helped to explain what I was using it for. Doh! I'm not really trying to measure anything, except the inherent nature of the detector. I'm using it as a switch. The environment for the device is inside a sealed compartment. When any light is detected by the device it "turns on" passes voltage in the range of 5VDC to 15VDC. It doesn't take much to set it off. It's really amazing to play with. The comment about heat is a compelling conversation; the PICOR PI3111 is a hot little sucker. I've had to make special design accommodations for it the make sure it sits against a side of the enclosure for a heat sink. Anyway, that is the use case, trigger.

 

The comment about heat is a compelling conversation

I feel it is not necessary to explain to you that input bias current in the op-amp can reach 600 pa at 70C and that could cause as much as 6 volts of error. The fact that you built a board which works correctly in the range of 10 to the 10th power indicates that I am just babbling at the moment.

 

I feel it is not necessary to explain to you that input bias current in the op-amp can reach 600 pa at 70C and that could cause as much as 6 volts of error. The fact that you built a board which works correctly in the range of 10 to the 10th power indicates that I am just babbling at the moment.

Don't assume anything, please. I'm not an electronics guy; as I stated in a previous post, I do this because I have to. My background is as a senior systems programmer/analyst, intelligence side of the house. When I need to do something, I try and do as much research as I can before I ask for help. But, inevitably, when it comes to electronics, I have to often ask for some help. So if you have any suggestions, please feel free.

 

So if you have any suggestions, please feel free.

Thank you for your courtesy, but I don't have any suggestions. More like admiration for a person who stretched the limits of practical application. I am educated in this subject. I have seen Teflon standoffs, air built circuits, circuits built on glass, and guard rings designed to make this sort of circuit work. The fact that you could do this is remarkable to the point that I wonder if a contributing factor is that nobody told you how difficult it is. You are working in a range where humidity is a factor and the circuit might not work properly in a month if you fail to protect it from contaminants in the air coating it. I suppose I just made two suggestions: Protect it from air born contaminants and be aware of humidity.

 

You have an ultra-sensitive light sensor feeding a charge coupled amplifier.
Why the need for ultra-sensitivity when a simple light-dependent resistor (LDR) and an ordinary transistor would do the same job?

 

Thank you for your courtesy, but I don't have any suggestions. More like admiration for a person who stretched the limits of practical application. I am educated in this subject. I have seen Teflon standoffs, air built circuits, circuits built on glass, and guard rings designed to make this sort of circuit work. The fact that you could do this is remarkable to the point that I wonder if a contributing factor is that nobody told you how difficult it is. You are working in a range where humidity is a factor and the circuit might not work properly in a month if you fail to protect it from contaminants in the air coating it. I suppose I just made two suggestions: Protect it from air born contaminants and be aware of humidity.

 

Maybe that explains something I've seen. I've had to build a new one if I ever touch the glass of the optical sensor. Once touched it you can't clean it to the point it works the same. Can you advise on that observation? Is there some residual biometrics that might be left on the glass? Or does rubbing it to clean it cause an issue? I've tried pure acetone as a wash and it helps a bit. But the only real solution is to completely, never, ever touch the glass. In fact, I don't touch the sensor or any part of it.

One other thing for anyone thinking of using this finicky little pain in the ass... solder pin2 of the TLC271 and the connection to the R1 and C1 so that they don't touch anything else, not even the PCB. Don't use SMD/SMT, use through hole and don't let them touch anything else. This I learned by trial and error. Looks funny as hell, but for reason it reduces "ripples" in the power output. Can you advise on that too by any chance?

 

You have an ultra-sensitive light sensor feeding a charge coupled amplifier.
Why the need for ultra-sensitivity when a simple light-dependent resistor (LDR) and an ordinary transistor would do the same job?

I needed a trigger that was so sensitive to light that even a cigarette lighter would set it off. I tried several types of things. This was the best, most sensitive, and was it ever fast. Almost instantaneous reaction. By changing the R1 and C1 you can reduce the sensitivity. But that wasn't my objective. Hope that answers your question

 

Oh yes, one more thing... I use Silver based solder. It was a suggestion from another tech when he saw some of the signal variations. It appeared to work a bit and maybe counters the humidity/containments problem that was mentioned? I dunno, I played with this for nealy 3 weeks solid before I get it to work. Makes it a damn expensive toy.

 

Maybe rubbing the glass leaves a static charge. After all, you are detecting a very few electrons and they will leak off the glass slowly. Maybe time will be the cure. Leave it alone and come back tomorrow, or rinse it in deionized water and then wait until tomorrow.

solder pin2 of the TLC271 and the connection to the R1 and C1 so that they don't touch anything else,

That is called air mounting. It is even better than a Teflon stand off. Sorry you had to learn the hard way, but I did too. You have to wrap your head around picoamp leakage paths to get good at this.

I do wonder why the PI3111 is hot supplying such a low current to the load. Could you feed it less voltage? Are there loads you did not show? Are you using a high tech switching supply when an analog regulator (LM78L15) would be sufficient?

 

Maybe rubbing the glass leaves a static charge. After all, you are detecting a very few electrons and they will leak off the glass slowly. Maybe time will be the cure. Leave it alone and come back tomorrow, or rinse it in deionized water and then wait until tomorrow.

That is called air mounting. It is even better than a Teflon stand off. Sorry you had to learn the hard way, but I did too. You have to wrap your head around picoamp leakage paths to get good at this.

I do wonder why the PI3111 is hot supplying such a low current to the load. Could you feed it less voltage? Are there loads you did not show? Are you using a high tech switching supply when an analog regulator (LM78L15) would be sufficient?

To be honest, I was grabbing the most convenient DC/DC down converter I could find readily available. The source power is a fixed 24VDC power supply. The voltage range out I needed from the TLC271PC was the range I needed. It's definitely not optimized since the proto-board for the PI3111 is like 3"x3". Makes it clumsy looking.But that is why I put it on the forum, to let the smart people provide alternatives. I will read up on the LM78L15

 

5 ma x 9 volts (difference) is 0.045 watts dissipated in the regulator.
Even the maximum output of the chip of 30 ma + 5 ma for the chip would cause 0.315 watts of heat.
That has never caused a heat problem in my world.
Have you skipped installing capacitor(s) external to the regulator, thus causing it to oscillate?
Be aware that touching an oscilloscope probe to the regulator pin might stop the oscillation and present a false testimony.

 

5 ma x 9 volts (difference) is 0.045 watts dissipated in the regulator.
Even the maximum output of the chip of 30 ma + 5 ma for the chip would cause 0.315 watts of heat.
That has never caused a heat problem in my world.
Have you skipped installing capacitor(s) external to the regulator, thus causing it to oscillate?
Be aware that touching an oscilloscope probe to the regulator pin might stop the oscillation and present a false testimony.

I looked at the LM78L15. My question sort of devolves into one my earlier concerns, and that is so many grounds. The PI3111 proto-board gave me two inputs and two outputs, +/- 24VCD and +/- 15VDC. The LM78L15 adds another ground. Will there be problems having a single ground for all of the components in the schematic? Basic question I know, but one I always seem to end up stubbing a toe on.

 

Yes, we all know that GND means 0V.

Think of GND as a common reference point as analogous to zero-height sea level.
All circuits have to have this common reference point. There are exceptions which we need not get into at this point.

Yes, you ought to connect all your devices to a common GND point.

 

If you are trying to detect light from ultra-low level light source, you need to define two things:

1) What is the light level of the source to be detected?
2) What is the ambient light level?

In other words, you need to know the signal to noise ratio (SNR), or signal over background.

 

There is not enough resistance in the ground path on that small, low current board for the current to produce significant voltage changes when the load changes. The LM78L15 has one ground and you will need to give it one or two capacitors to keep it from oscillating. Still, the number of grounds is irrelevant unless you make quite an effort to make the ground traces long and narrow.

If you place the BPX61 directly across pins 2 and 3 of the amplifier, there is no ground circuit involved for an error to be amplified.
MrChips speaks of a, "star" ground where every part which needs a ground has its own conductor to one point, but you have already demonstrated that you have very few problems without knowing of this method. You can re-design it with a star ground shape, but I am hard pressed to understand why you need to fix what is not a problem.

MrChips also speaks of signal to noise ratio. I was under the impression that you are only discovering, "if" or, "if not". Zero or one. You might make this point clear to us.