SIMD1
From BEAM Robotics Wiki
Contents |
[edit] General
All circuits on this page © Wilf Rigter ; note that the majority of this explanatory text comes from either postings by, or correspondence with Wilf
In response to the shortcomings of the "classic" D1 solar engine, Wilf Rigter designed a nocturnal solar engine of his own. As various people posited various applications of the circuit, it soon fragmented into a whole family of circuit designs -- the SIMD1 (simplified D1) solar engines. Each of the SIMD1 designs is a bit different, and so each has its own advantages and disadvantages. To date, Wilf has dreamed up 4 variants of SIMD1 solar engines -- V0, V1, V2, and the SIMD1 / Solar Regulator. I'll attempt to document and explain the latest version of each in turn.
Note that there is a qualitative difference between the D1 solar engineand most of the SIMD1 circuits -- the D1 solar engine can provide significant current to power up a circuit like a pummer while most of the SIMD1 circuits provide a logic signal for control and indication. In other words, the D1 turns off everything connected to it while charging, but most SIMD1 variants simply provide a logic output that must be used to disable other parts of the circuit.
[edit] SIMD1 V0
SIMD1 V0 is the original SIMD1 design, and is based on a simple concept -- charging the storage capacitor through a Ge diode (for minimum forward voltage drop).
SIMD1 V0 Basic Circuit e = enable (active low); v = Vcc; g = Gnd
SIMD1 V0 Example Circuit LED load with gate driving enable
A resistor (here shown as 1MOhm, but can be as small as 100KOhm) is connected in parallel with the solar cell for two reasons:
- It provides much quicker turn-on when the Sun goes down. Note that a smaller value for the resistor will provide faster turn-on; the trade-off is that a smaller value for the resistor also makes the solar engine less efficient at charging the capacitor (since we're then losing more of our precious solar power through resistive losses in daylight). In any event, the SIMD1 V0 requires almost complete darkness in order to trigger.
- It acts as a "pull-down" resistor to provide a discrete indication of when things are dark enough for the load to come on. Essentially, this means that the solar cell is used as a light sensor (incidental to its function of charging the storage capacitor).
In the example circuit, the SIMD1 V0 is shown driving a single LED, which is of course a trivial example to demonstrate the basic solar engine circuit. The example takes advantage of the fact that HC logic is made with FETs so why bother using an additional discrete FET to drive a load? So here the V0 output is connected to a CMOS gate input. The Schmitt trigger is just an example of "the rest of the circuit" albeit a special case because of the fast switch-over due to input hysteresis. More importantly, the CMOS input threshold drops as Vcc drops which reduces the effect of the dropping capacitor voltage during discharge.
Normally the active high output from the inverter is used to enable the remainder of the CMOS gates which then perform some useful function. When the SIMD1 triggers at night, it snaps on and the output signal can be used to control the 5 remaining 74HC14 inverters connected in parallel with the outputs used as a "power" switch. Or the signal can be used as the tristate control of a 74HC240 or 74HC245 or be used as a PNC input for an [[Nv net,] etc. to apply the stored energy to the load.
A point to remember when adding on the rest of the circuit is that CMOS logic draws considerable supply leakage current when inputs are not at 0V or Vcc. That problem virtually disappears when Vcc <3V; as a result, this circuit is only useful with a single 2.5V capacitor and a lower voltage solar cell. Also note that the example circuit also has a somewhat high switching voltage when the difference between the solar cell voltage and the capacitor voltage is sufficient to turn on the CMOS gate. A 74HC14 switches at a difference voltage of 2/3 Vcc. In general, HC logic switches at 1/2 Vcc and HCT logic around 1.8V.
A variant of this circuit can also be used to provide a dark-driggered PNC. Note the 100 ohm positive feedback resistor loading the solar cell only after the SIMD1 turns on, this can be adjusted to optimize the SIMD1 turn off point. The output diode is connected in place of the usual PNC in an Nv net circuit.
[edit] SIMD1 V1
SIMD1 V1 is functionally the same circuit as SIMD1 V0 but provides an active high signal to the load when night (or at least darkness) has arrived.
SIMD1 V1 Basic Circuit e = enable (active high); v = Vcc; g = Gnd
SIMD1 V1 Example Circuit LED load with gate driving enable
As with the SIMD1 V0, a 100K - 1M resistor across the solar cell helps with turnoff, but the SIMD1 V1 requires almost complete darkness before switching.
In the example circuit, the SIMD1 V1 is again shown driving a single LED. This example (as with the V0 example) makes use of a 74*14 Schmitt inverter to switch power on within the load.
Normally the active high output from the inverter is used to enable the remainder of the CMOS gates which then perform some useful function. When the SIMD1 triggers at night, it snaps on and the output signal can be used to control the 5 remaining 74HC14 inverters connected in parallel with the outputs used as a "power" switch. Or the signal can be used as the tristate control of a 74HC240 or 74HC245 or be used as a PNC input for an Nv net etc. to apply the stored energy to the load.
A point to remember when adding on the rest of the circuit is that CMOS logic draws considerable supply leakage current when inputs are not at 0V or Vcc. That problem virtually disappears when Vcc <3V; as a result, this circuit is only useful with a single 2.5V capacitor and a lower voltage solar cell. Also note that the example circuit also has a somewhat high switching voltage when the difference between the solar cell voltage and the capacitor voltage is sufficient to turn on the CMOS gate. A 74HC14 switches at a difference voltage of 2/3 Vcc. In general, HC logic switches at 1/2 Vcc and HCT logic around 1.8V.
[edit] SIMD1 V2
Wilf's description of this design's birth: In the never-ending quest for simplicity even one extra diode is a price I loath to pay and to avoid using extra components it struck me that the the two diodes in SIMD1 V1 were connected like the two junctions in one transistor. Eureka! replace two diodes with one transistor to restore the original component. Now one whole transistor is more than the sum of its two diode junction parts and on closer examination, it became clear that I had fortuitously solved another problem. I wondered if the design could be made any simpler. Lo and behold SIMD1 V2 which uses one less component and doesn't require very dark conditions to turn on. The two 1N914 diodes in the earlier SIMD1 design can be replaced with a single transistor (i.e. the two diode junctions) but in the process, this revision has really enhanced the behavior.
SIMD1 V2 Basic Circuit (NPN) e = enable (active high); v = Vcc; g = Gnd
SIMD1 V2 Example Circuit LED load with gate driving enable
SIMD1 V2 Basic Circuit (PNP)
e = enable (active high); v = Vcc; g = Gnd
SIMD1 V2 Example Circuit 2
Driving LED flasher circuit
I was looking for a way to effectively speed up rate of change of the input voltage by amplifying it. When I subbed the transistor for the two diodes, I realized that the base emitter junction of the transistor was a replacement for the sensing diode, and the collector base junction provided isolation, BUT the gain of the transistor would increase the rate of voltage change on the base pin by a factor of 100 at the collector pin thereby eliminating the supply leakage problem. Since the 2N3904 replaced diodes (two in the original SIMD1 V0, one in later versions), no additional parts were required for this much improved performance.
This circuit was introduced as SIMD1 V2 first shown with a Schmitt trigger driving a LED but this trivial application also required a pull-up resistor on the NPN collector. Since the NPN is on during charging, the value of the pull-up resistor is made large to reduce supply current.
A second SIMD1 V2 is shown in a more useful application where the transistor (this time without a pull-up resistor) is used to hold off a CMOS oscillator during charging which when enabled at night drives an LED pump flasher. This version really highlights the simplicity of the SIMD1 V2 design which (as in this example) can also be implemented with a PNP transistor which provides an active high signal during charging.
The PNP, which is any high gain transistor including the 2N3906, is ON during charging as the solar cell pumps current into the storage capacitor through the base emitter junction. The transistor collector clamps the input of the typical oscillator shown to +V until the solar cell output drops below the voltage of the capacitor. In my prototype, the capacitor is a 1F 2.5V gold cap charged with a simulated 5.5V solar cell and the transistor will turn off when the charging voltage drops below the fully charged capacitor voltage (~2.4V). Since the PNP operates as a common base amplifier, the voltage gain at the collector will cause a rapid voltage transition through the linear region of the 74HC240. When the PNP turns off, the oscillator starts and causes the two LEDs at the output to alternately flash. The way the 47 uf cap discharges through each LED produces a small light explosion and after-image quite pleasing to the eye.
The SIMD1 V2 design can also replace the two diodes of the original 74HC14 design or control a whole 74HC240 chip if the collector has a 1M pull down resistor to 0V and is connected to the tristate enable pins. The SIMD1 V2 prototype shown was actively flashing the 2 LEDS at a 2 Hz rate for over 4 hours on a fully charged (2.4 V) cap.
[edit] SIMD1 V3
The SIMD1 V3 is very similar to the NPN flavor of SIMD1 V2, but with the addition of a FET on the output line. This makes life easier for its attached load (since there's no need here for something that can use an "enable line"), by providing power that's just "there" in the dark. Here the FET is used to sink load current from a pummer, motor, or whatever.
Turn-on is crisp, and the circuit switches rapidly when the light drops just a little. As with the other SIMD1 circuits, the 1M resistor across the solar cell can be changed to a smaller value (i.e. 100K) to adjust sensitivity and make it turn on even faster (with a corresponding loss in charge efficiency). I suspect that the FET could be replaced with a PNP transistor, but haven't tried this yet...
[edit] SIMD1 / Solar Regulator
The SIMD1 was enhanced in the Solar Regulator version which supplies a constant 2V after triggering and permits driving LEDs without using current limiting resistors with constant brightness. Note that you can use this Low Drop Out (LDO) linear regulator for various applications but like all linear regulators it is "lossy". This is not a problem for controlling LEDs since they would otherwise use "lossy" resistors but for motors it is a different story.
Try this new SIMD1 / Solar Voltage Regulator for use with blinking LED circuits (pummers). It turns on when it gets dark, just like a D1 but the output voltage is regulated to about 2V (depending on the reference LED Vf).
The SIMD1 / solar regulator circuit draws no current during charging and when turned on, it draws less than 100uA with a maximum 10ma output current. The regulator provides constant LED brightness during the discharge and turns off when 1F solar capacitor voltage drops below the LED turn on voltage. The LED used for voltage reference in the solar regulator feedback loop and the LEDs used for the flasher must be the same high efficiency type to match the forward voltage specs. This circuit is ideal for supplying voltage to a Bicore or 74HC14 LED flasher since it eliminates the LED current limiting resistors and greatly reduces current consumption of the HC flasher circuits.
One interesting alternative would be to substitute a 5V NiCad (4 cell) battery for the 1F supercap which acts to increase the storage capacity many fold for use with flag waver motors, pendulums, etc. With higher load current, the 100K resistor may be replaced with 20K for up to 50 ma output current. The quiescent current of the regulator remains very low and is proportional to the load current for high efficiency operation during discharge.
[edit] Charging
The solar cell charges a 1F capacitor through a 1N34A Germanium diode to a maximum voltage of 5.5V. While the charging current flows through the diode the voltage at the cathode (stripe) is about 100 mV negative with respect to the 0 V line. This negative voltage is applied through a 100K resistor to the base of a 2N3904 NPN transistor Q1 and holds that transistor off. This cuts off the base current for the 2N3906 transistor (PNP) Q2 and the output of the regulator will be zero volts.
[edit] Switching
Rapid switching is very important in this type of circuit because a circuit that is half on draws power, draining the capacitor, but performs no useful work. On the SIMD1/ Solar Regulator, the output snaps on and off.
At the end of the charging cycle, when the light on the solar cell decreases, the negative terminal of the solar cell starts to become more positive than the 0V line. The base of the NPN Q1 must be at about +500 mV (positive) with respect to the emitter which is connected to the 0 V line, before it turns on and turns on the rest of the regulator. That usually happens in the evening but can be simulated by cupping your hand in front of the solar cell.
When Q1 turns on, the PNP transistor 2N3906 - Q2 receives base current and it starts to turn on. The regulator output voltage at the collector of Q2 increases to about + 2V when the red LED starts to turn on and to supply current to the base of NPN transistor 2N3904 - Q3. When Q3 turns on it "robs" base current from Q2. This in turn controls the base current for Q2 and the regulator output will stabilize at +2 V. The 10K resistor from the regulator output to the negative terminal of the solar cell provides positive feedback to the regulator turn on by loading the solar cell down so that it's output voltage drops even more and the regulator "snaps" on. Note that the red LED is used for reference voltage only and does not actually light up.
[edit] Discharging
The output voltage at the collector of Q2 remains at 2V while the voltage on the main capacitor can range from a full charge at 5.5V to 2.1V at the end of the discharge cycle. If no load is attached to the regulator the capacitor voltage will drop very slowly because of leakage and a small amount of current required for the active regulator (<50 uA). When a HC chip like a 74HC240 or 74HC14 is powered from the 2V regulated output, the current for that chip is also very low.
If the HC chip has a LED connected to the output which the same type of LED that is used for reference, then the current will be limited by a small voltage drop on the HC driver output. Since the regulated voltage is constant the brightness of the LED is also constant.
When the voltage on the 1F cap drops below 2V, the regulator reference LED turns off and the base currents of Q1 and Q2 increase discharging the remaining charge on the cap and turning any attached circuit rapidly off. At some point the voltage of the solar cell even in dim light is higher than the remaining charge on the capacitor and if there is sufficient light (usually in the morning) the charging cycle repeats all over again. If the Sun is bright and the solar cell was shielded by your hand, then exposing the solar cell to the bright sunlight generates enough power to turn the regulator off and force the circuit back into the charging cycle.
The PowerSaver Flasher uses capacitive output coupling to produce brighter shorter flashes and has a much lower average current drain than standard bicore or 74HC14 flashers. The PS Flasher with one LED circuit (2 LEDs) runs all night from a 1F cap charged to 5.5V. Up to 12 LEDs can be controlled with one 74HC14 flasher and probably would run for 2 hours from a full charge. Use a range of timing resistors between 1M and 4.7M for each oscillator to give a random light show appearance.
