Electro Optical Gadgets

22 Jly 2007 - A few minutes using the TSL267 with the Battery Top Power Supply and the Printer Encoder Strip looped back on itself seems to confirm my idea that the slot opto interrupter is not working because the light source is closer to a point source than a collimated source. With a point source the light falling on the strip not only passes at 90 degrees, it also is at other angles, making it much harder to block the light. I ran this test in the day time, but room IR background saturates the TSL267 so need to wait till dark to really try it out.

Electro Optical Gadget TSL267 & 150 LPI strip

22 July 2007 9 pm - Used angle head Flashlight about 4 feet from TSL267 and it's working.
The output voltage goes between 0.67 and 2.63 for a Pk to Pk signal of about 1.9 volts. The speed I was sliding the encoder strip on itself resulted in a frequency of about 633 Hz.

So it's important that the light source is collimated and desirable that it has near IR output.
Blurry photo hand held auto time exposure.

Theoretically the wave form should be a triangle with points at the top and bottom and the scope image is close but the top and bottom are rounded so looks like a sine wave. That may be related to the sampling scope's low one shot bandwidth.

So to use the strip ambient light needs to be blocked and a collimated near IR source used.

The Light to Voltage family of TAOS parts uses a photo diode as a solar cell generating a current proportional to the light input. By connecting the photo diode output to the virtual ground negative input to an op amp the current gets transformed into an output voltage The negative feedback circuit consisting of a resistor in parallel with a cap allows trading gain and bandwidth to some extent. The rise time for these varies from a few to a few hundred microseconds.

Methods of using Photodiodes:

As Solar Cell - no bias -In this mode the output current is proportional to the light input. At a former residence I had a small solar panel (1x3 inches) flat on the roof driving an analog current meter as an indicator of the Sun's brightness. A voltage meter connected to a solar cell does not give you the same information. TAOS uses an op amp connected as a transimpedance amplifier to convert the current out of an unbiased photo diode (small solar cell) to a voltage from a low impedance in their Light to Voltage products. These have a dynamic range of many decades. There's a tradeoff between high gain for high sensitivity and lower gain for faster operation.
Reverse Biased - the leakage current is a measure of the light on the diode. The back bias lowers the capacitance allowing the diode to be faster. There are a lot of complications related to the presence of DC so this method is typically only used for things like fiber optic receivers where speed is very important.
Novel Method with high sensitivity - uses PICmicro controller to first back bias the LED to charge up it's capacitance (uses 2 I/O pins, cathode ground), then switches the pin connected to the cathode from ground to a digital input then grounds the other PIC pin. Now there's 5 volt input to the digital input. The digital input pin has an extremely high impedance. A timer is started and stopped when the input switches low (good to make this an interrupt pin). The time to discharge is a measure of the light on the LED. Can take a second when light level is 0.0001 lux).
The output current from the photo diode can be used to control the frequency of an oscillator. These devices have the effect of integrating the light level and so are slower in responding than a photo diode connected as a current source. This is what's done in the TAOS Light to Frequency converters.

Idea for Much Higher Resoltion Incremental Encoder

When you look at two identical Printer Encoder Strips as one is moved over the other the amount of light coming through varies from completely blocked off (if they are well aligned) to 50% of the light on the other side of the strip. So a linear light intensity sensor set so it's full scale output goes from black to 50% of the light source will have a sawtooth output as the strip is moved. Two of these sensors with a separation that's some integer of the pitch plus 1/4 pitch (i.e. a quadrature sensor arrangement) would allow not only much finer resolution but also direction of movement detection.

This method would not be good for fast turning motors but would be great for things like telescopes that move slowly. Note that the light intensity is directly proportional to the effective slit width so the voltage output from the light to voltage converter is a straight line function of the displacement of the two strips. Only an offset and scale factor correction need to be applied.

Instead of taking the photo diode output as a binary signal, process is through a transimpedance amplifier to get an analog signal. Use an A/D converter to read how much light is there. There will be some limit on how many bits can be added by I expect that 8 bits is not out of the question.

For most applications only the final position is important so the photo diode output can be split into two channels. One channel is for conventional digital counting and direction. During high slew moves the analog channel may or may not be able to keep up. So a speed based switch driven from the digital channel can turn off the analog output or for lower cost just ignore the lower significant digits during slews. This might be a problem when in a GOTO mode, so there should be two speeds used, the first for getting near the target and a slower speed used to creep up on the desired setting. I think this is how they now work, but would be required to use the higher resolution mode.

The reasonable priced angle encoders now have 500 slots per turn (0.72 deg) and the minimum step size you can get is 1/4 pitch (10.8 min angle).
But this might be changed to 2.5 sec angle using the analog method.

When the number of bits in the D/A converter gets high enough variations in the pitch of the strip or wheel will start to show up. A way around that is to use the index mark to allow calibration of a complete circle by using an external index head to set the position and build a table in EEPROM.

Slot Opto Interrupter

Clock Escape Wheel

Printer Encoder Strip

Electronic Gold Mine G15602. A 13" long x 0.237" (330 mm x 6 mm) with alternating clear and dark bars at 150 line pairs per inch (5.9 lp/mm). The dark lines are 0.157" long ( 4 mm). There is a slot at each end, both angles at 45 degrees and parallel to each other. I think this was used on a printer to locate the print head. The obsolete HP Deskjet 3810/3820 Printer series has manuals with product number C8952A and the encoder strip is marked "C8952A-80005".

At 150 lp/inch one line pair takes up 0.006666" (0.17 mm( which is smaller than the 0.5 mm slit in the above Sharp slot type opto interrupter, but that shouldn't matter since when two strips are in the gap and perfectly aligned but out of phase such the bars one one strip are aligned with the spaces on the other no light should get through. And this is what happens. The data sheet for the Sharp sensor mentiones that the sensor slot is vertical, i.e. aligned with the long axis of the gap but I don't think that's much of an issue.

It's impossible to hold a folded encoder strip in your hands and adjust it to be black for more than about 1/4". The problem comes when the alignment of the two strips is off a little the you get a Moiré pattern (Wiki). The effect can be calculated (Wiki). For two strips with the same pitch the big distance on the pattern between dark lines is D = p/a where a is in radians and D and p are in the same units. So for p = 0.0066666" and a = 1 degree or 0.017453 radians, D= 0.38"

angle deg (radian)	radian	D =
0.029 (1.7 arc min)	0.000513	13"
0.25	0.004363	1.5"
0.5	0.008727	0.76"
1	0.017453	0.38"
2	0.034907	0.19"
4	0.069813	0.095"
19.5 deg	0.34	0.5 mm

If the angle between the two strips was within 1.7 arc min the whole 13" would appear to be black.
The bottom line is that it's not trivial to make a linear encoder that works.
If the two strips are aligned to within 19 deg across the 0.5 mm wide slit then the brightness will be a function of the linear offset between the two strips which will vary as the phase of the pitch.

I've heard that the commercial linear encoders use two sensors for the "A" phase and two for the "B" phase where the two sensors for the same letter are out of phase so one is dark and the other is light. Theoretically only one A and one B are needed, but by using complementary sensors it works better. With the folded strip this would be an "A" only incremental sensor, i.e. no direction information.

TAOS Light to Frequency Converter

10 Hz IR Avalanche Pulser

Light to Audio

LM3909 LED Blinker

Unihedron Nu-B Light Source

Solar

Solar House Number

Solar Garden Light


Black battery holder, photo diode, w LEDs green is the back side of the solar panel. This may be a sililar circuit dirgram	PCB

After reading an interesting analysis of two different Solar Garden Lights I got one to see the high technology they contain. The interesting ones have a single cell battery to store the power so they include a Switch Mode Boost Power Supply to get enough voltage to drive a LED.

This one has a metal cylindrical frame about 5" dia x 6" high. A square solar panel a little smaller than 2.5" on a side drives the electronics. Three AAA Ni-MH cells store the power. A photo diode senses if it's night. Two 5 mm LEDs shine down and some light would fall in a circle around the fixture and what light hits the bottom reflector is spread horizontally. The batteries were dead (this is a used discount store unit) and are now on the C401FS charger, then to the C9000 for discharge analysis and cycling. Since there's a three cell battery no need for the SMPS.

The batteries showed fully charged in less than 10 minutes. They are rated at 1.2 Volts and 600 mAh. Discharging them showed about 33 mAh capacity, so they now are on the Break-in mode of the C9000. After the Break-In the cells now have 279, 229 and 308 mAh capacity almost 10x what they were down to after sitting. But it's still half the label capacity.

After installing the batteries the light did not work. The voltage at the PCB was low. Rubbing the battery terminals on my pants and after installing rotating the cell in the battery holder fixed that. Now when the photo diode is put in the dark the two LEDs turn on. Used Radio Shack Lube Gel (Silicon Grease) on both ends of each battery. You might think that's an insulator and would stop good contact, but that's not the case. There's enough spring pressure for the metal to push the weak grease aside and make contact. But Lube Gel is made without any entrapped oxygen so no air can now get to the joint to allow it to corrode.


Working Solar Garden Light

Put outside in a spot that gets some Sun, but only an hour or two per day around Xmas. Light came one at dusk (noticed it was on at 5 pm) and at 7 am the next morning it's still on.
If each LED was running at 3.3 Volts and 10 ma that would be 20 ma total and if the battery capacity was 600 mah the max run time would be 30 hours. Off at about 7:13 at dawn probably not because the batteries have run down.

The job of the circuit would be to turn the LEDs on and off based on how dark it is outside. It's hard on Connect the solar panel to the battery when there's charge to be had in the most efficient manner possible an. Three Ni-MH cells and the white LEDs are a good match. For example it's desirable to stop the discharge of Ni-MH cells at about 1 volt per cell, or in this case at about 3 volts. The LED forward voltage is that of a diode so as the battery voltage goes down as it becomes discharged the LED current also decreases.

My guess is that the number of hours of light depends on how much solar energy gets put into the battery. During the winter it's only going to get a couple of hours of sun and I've not put in into the ground plumb, but rather with the top pointing kind of toward the sun. If this solar panel is like the one below for the solar fountain pump (35 mw/sq in) then it's good for about [35 * 2.5 * 2.5 =] 220 mw for some number of hours. The voltage at the end of charging three Ni-MH cells is about 4.2 Volts so in the ideal case 220 mw could provide 50 ma of charge current and so it would take 12 hours to fill the battery to 600 mah or considering the battery efficiency more like 18 hours. Some type of power point charge controller that matches the panel to the battery would help get more charge into the battery.

It's clear that in the winter time the battery is going to be operated most of the time at empty with only a small amount of charge and back to empty. Being on all of last night was because I charged the battery.

It may be that under these conditions the newer Ni-MH cells that have extended shelf life would improve the performance. This is just a guess based on the idea that self discharge may be more important when the battery is not fully charged.

A low resistance super capacitor probably would be a better energy storage method. The idea is that they should have better efficiency that a battery.
27 Dec 2007 light was on again last night about 28 hours total. It's cloudy today.

1 Jan 2008 - the light has been on every time that it could be that I've checked. I.e. it's on a little past 5 at night and till 7 something am. There's no way it's getting enough direct Sun light to charge the batteries and my initial charage would have worn off by now, so the solar panels must generate a small current just from the sky light.

Second Solar Garden Light

New solar Garden Light w/ Dead Batteries

This light also uses 3 AAA Ni-MH cells and after charging them the capacity was very poor (around 30 mAh instead of the label 750 mAh). After using the break-in function on the Maha C9000 the capacity was 420, 451 and 706 mAh which is too dissimilar to use in a series pack. The problem being that when the lowest capacity cell is discharged the other cells will power it in the reverse direction which rapidly ruins that cell and can cause it to vent. That's probably the cause of venting in the Solar House number.

This appeared to be a new unit. It had a red flag, like is used on aircraft "Remove Before Flight", only this one says "Remove This Shipping Tab Before Use" and has one end inside the battery compartment seperating a battery terminal from the battery holder contact thus opencircuiting the battery pack. Also there are clear plastic protectors on the four solar panels with the legend "Note, Please peel off this protecting film before use."

Photo taken after removing the top from the rest of the light.
You can see that there are solar panels each about 2 1/8" x 3/4". There's also a photo diode on the top. Two 5 mm plastic LEDs for the light and three AAA batteries for power storage. I'm guessing the time and temperature profiles of storage or poor initial quality of the batteries or both resulted in bad batteries when in as new condition.

Solar Garden Light or Marker Light Patents

4486820 Lighting equipment with a solar cell, Y. Baba (Kyoto Ceramic Co), Dec 4, 1984, 362/183; 362/157; 362/190; 362/276; 362/394; 362/395; 362/431; 362/802; 361/171; 136/244 -

a 4 W fluorescent lamp can be lit for 6 hours with the use of a solar cell of 16 V, 12 W and a battery of 12 V, 90 AH.

4816970 Solar powered light Mar 28, 1989
5065291 Marking Light, J.S. Frost et al (Atlantic Richfield Co), Nov 12, 1991, 362/183; 362/431; 362/31; 362/800; 362/145; 136/291 - small solar panel, minimal one transistor circuit and LED. - The battery voltage must exceed the LED operating voltage and the solar panel voltage must exceed the battery voltage. simple cleaver circuit.
5221891 Control circuit for a solar-powered rechargeable power source and load, R.W. Janda, Jun 22, 1993, 323/350; 323/906; 320/21; 320/61; 362/183 - 2.7V, 180 ma solar panel, 2 x SubC Ni-Cad cells, 3 transistor control circuit, #1767 2.3 volt incandescent lamp patents 5086267 5041952 use a very similar circuit
5984570 Self energized automatic surface marker Nov 16, 1999
6013985 Sealed solar-powered light assembly, David R. Green (Carmanah Tech), Jan 11, 2000, Jan 11, 2000, 315/149; 315/159; 362/183; 362/800 - two timers, voltage reg, cur lim resistors on multiple LEDs - not too efficient
6406163 Solar cell lighting fixture integrated with heat sink, Tai-Her Yang, Jun 18, 2002, 362/183; 362/374; 362/276 - diminishing effect of solar heat on batteries and electronics
6573659 Solar-powered light assembly with automatic light control, Ion Toma (Carmanah Tech), Jun 3, 2003, 315/149; 362/372 - uses micro controller to dim light to allow it to stay on all night based on charge obtained the prior day.
6729742 Solar lamp for outdoor use, W. Wismeth, May 4, 2004, 362/183; 362/153.1; 362/431; 136/206 - seperate light sensor diode & solar panels facing different directions

Small Solar Panels

My wife wanted to try out the Harbor Freight 91962 Floating Solar Fountain Pump but it came DOA. After going thought the RMA procedure they said there was no need to return the dead one so I opened it up in the hope of recovering the solar panels, which worked. The probable reason for the DOA is very poor soldering at the joint between the panel wires, the pump wires and a 1N4739A 9.1 Volt Zener diode connected cathode to red wires (positive), i.e. normally back biased by would limit spikes from motor that might exceed the solar panel breakdown voltage.

The two 6" x 4¼" panels are connected in parallel and put out about 11.7 Volts in direct sun. Short circuit current of 165 ma, but those are not at the same time. So the power out is going to be less than 2 Watts. The Harbor Freight 41144 5 Watt Solar Battery Charger is 18" x 12.5" or about 3.3 Watts / square foot. At that rate these two panels would be about 1.2 Watts.

Pump Testing

The pump while pumping water draws:

# AA Batt	H₂O Ht"¹	mA	V	Ohms	W
4	1	220	3.95	18	0.9
5	2	240	6.5	27	1.6
6	4	250	7.6	30	1.9
7	8	213	8^?	38	1.7^?
8	16	240	10	42	2.4

Note 1 estimated, not measured

Load Testing Solar Panels

By measuring the voltage across a load resistor the power can be computed as( V * V) / R.

Load Ohms	50 W Desk Lamp Volts	50 W D.L. mW	Oct¹ noon Volts	Oct noon mW
10	0.099	0.98	1.5	225
20	0.042	0.088	ng	ng
47	0.83	14.7	6.4	871
57	x	x	7	860
67	x	x	7.3	798

Note 1: Not really direct Sun, some tree filtering. October has less Sun energy than July.
Looks like 35 mw/sq in. [900 mw / (6 * 4.25)]

5040726 Solar energy powered water fountain , A.T. Dimitri, Aug 20, 1991, 239/17; 239/18; 239/20; 239/22 - seperate solar panel
6435422 Floating Fountain, Mark Wutschik, Aug 20, 2002, 239/23; 239/18 -this floating fountain

	This is a laser module that projects a "+" image instead of just a spot. 635 nm (red) and 5 mw power level. Max input is 3.2 V. It draws 33 ma at 3.0 volts.
	The "X" is poduced by the parts shown at the left. A lens and molded plastic line generator, one half of which has lines at 90 degrees to the lines on the other half. It needs to be adjusted by screwing in or out to balance the brightness of lines.
	With the optics (lens and line generator removed the beam looks like this. When 5" away from the paper the beam is 1" x 5". With the laser off, looking into the hole where the light exits you can see what looks liks the edge of a metal disk partially blocking the exit hole. This is the light pattern from a raw laser diode. The emitting geometry is a thin line and at right angles to this the beam spreads the most.
	You can see the disk that's blocking about half of the exit hole.

Light	TLS245 IR Hz	TLS237 Visible Hz
dark space	<2	<1
indoors bright sunny day outside, no lights on	20 k	160 k
Quartz Halogen desk lamp on low power 18" away	305 k	573 k
Quartz Halogen desk lamp on high power	403 k	na
Outside in direct Sun	na	na