Spice Lite

Spice Lite is a second, more electrical way of running a circuit. The standard engine treats every wire as a perfect logic level that arrives instantly. Spice Lite gives every wire a voltage and every input reads the voltage on its own pin, and it adds what a real bench has: time, voltages that charge and discharge, currents through every part and the limits of the outputs and power supplies that deliver them, and the resistance of the jumper wires. Wire something up wrong and it fails the way the real circuit would, and you can watch a capacitor charge until it crosses a gate's threshold.

It is called Spice Lite because it is not SPICE: there is no circuit-wide matrix and no manufacturer models. A plain "74LS00" still needs no manufacturer or part choosing. Everything comes from a few shared formulas (Ohm's law, the exponential charge curve) and a short table of numbers per logic family, taken from TI's datasheets, and one common set for the diodes and transistors.

Turning it on

Open Settings ▸ Spice Lite and set Spice Lite to On. The change takes effect the next time you press Run: a circuit that is already running keeps the engine it started with.

With it Off, nothing on this page applies: the standard engine runs exactly as described in Running a Simulation.

Every wire is a voltage

Each time the circuit settles, Spice Lite works out the voltage on every wire from what is on it: each chip output as the circuit it really is, each input's own current, and every resistor, LED, diode, transistor and switch between them. Then every input reads the voltage on its own pin:

• At or above its HIGH threshold (2 V for 74LS, 3.5 V for CD4000 at 5 V) it reads HIGH; at or below its LOW threshold (0.8 V and 1.5 V) it reads LOW. In between it is undefined, and the gate's output with it. A Schmitt-trigger input (74LS14, CD40106B, CD4093B) keeps what it last read until the voltage leaves its hysteresis.
• The level a wire shows is what the inputs on it read. Most of the time that is exactly what the standard engine shows; where the voltage says otherwise, the voltage wins.

What that changes:

• A divider reads its voltage. 10 kΩ to the supply and 1 kΩ to ground makes about 0.6 V at a 74LS input — LOW. The standard engine, with two resistors pulling against each other, can only say undefined. A potentiometer's wiper reads where you set it.
• A diode has a drop. A diode-AND (inputs on the cathodes, a pull-up on the joined anodes) gives a diode's 0.6 V above a LOW output, which reads LOW. An LED or a diode in a chip's supply drops its voltage too.
• A 74LS input pushes current out of its pin while it is held LOW (about 0.2 mA at 0.4 V). Pulled down through 1 kΩ it sits at about 0.25 V, a good LOW; through 10 kΩ it sits at 0.9 V, in its undefined band — the reason a 74LS input is never pulled down through 10 kΩ on a real bench. A CD4000 input draws nothing, so 10 kΩ is fine there. The example circuits pull 74LS inputs down through 1 kΩ for this reason.
• A 74LS HIGH is about 3.6 V, unloaded. That is plenty for another 74LS input, and only just enough for a CD4000 input on 5 V; on 12 V it is no HIGH at all.
• A loaded output sags. A CD4000 output at 5 V lighting an LED through 330 Ω sags to about 3.2 V — under its own family's HIGH threshold, so a CD4000 input on the same wire reads it as undefined. Through 1 kΩ it stays at 4.1 V. The CD4000 example circuits use 1 kΩ for their LEDs for this reason.
• Two outputs fighting meet where their strengths put them: a 74LS LOW beats a 74LS HIGH at about 0.75 V, so the inputs on that wire read LOW. The conflict is still reported, and both outputs carry the current (below).
• Open-collector outputs (74LS01, 74LS03, 74LS05, 74LS47, and the 74LS181's A=B) never drive HIGH: wired together with a pull-up they make a wired-AND, in both engines.
• A wire nothing holds — inputs alone, or a lead to an LED that is off — is floating, and reads as each family reads a floating input: HIGH for 74LS, undefined for CD4000.

The probe shows the voltage of every wire that has one beside its level (H · 3.60 V), and the current through the lead in the hole it is on. A floating wire shows its level alone.

Time and charging capacitors

In Spice Lite every settle pass takes one gate delay: about 10 ns for 74LS parts and about 125 ns for CD4000 parts at 5 V (faster at higher supply voltages, following the datasheet's 5 / 10 / 15 V figures). On a desk that mixes the two, the CMOS gates take their proper share longer to answer, and a glitch shorter than a gate's delay never makes it through that gate, as on a real chip.

A net with a capacitor on it becomes an analog node. Its voltage follows the real charge curve toward the voltage the circuit round it pulls it to, at the rate the circuit's resistance and the capacitance set (the RC time constant). The whole circuit round it counts: a chain of resistors, a switch or transistor, the resistance inside the chip output driving it, and the current an input on it draws. A CD4000 output charging a capacitor straight can only deliver its limited current (about 4 mA at 5 V), so the capacitor first charges in a straight line and then curves the rest of the way.

• Each input reads the node through its own threshold: a 74LS input at 1.4 V, a CD4000 input at half its supply. An RC delay feeding both families switches the 74LS gate first and the CMOS gate later.
• A crossing happens once. When the voltage passes an input's threshold, that input sees one clean change. The voltage keeps climbing, but nothing happens again until it falls back through the threshold.
• A Schmitt-trigger input has two thresholds. The 74LS14, CD40106B and CD4093B switch HIGH at a higher voltage than they switch back LOW (the 74LS14 at 1.6 V and 0.8 V). That gap is what makes the classic one-gate RC oscillator work: a resistor from the output back to the input and a capacitor from the input to ground swing between the two thresholds at a steady rate. Built round an ordinary inverter, the same circuit has nothing to swing between and is reported as oscillating.
• A 74LS input on a node pushes its own current into it below 1.3 V, so an RC into a 74LS input starts faster than its R·C alone, and a 74LS14 oscillator needs a resistor of a kilohm or so to swing at all (through 10 kΩ its own input holds the capacitor above its lower threshold).
• A chip output charges a node only as far as its HIGH. A 74LS output takes it to about 3.6 V; on a desk with a 12 V supply, a CD4000 gate on 12 V, which switches at 6 V, never sees it change: the reason a 5 V part can't drive 12 V CMOS directly.
• A node that will never get there is left alone. If the resistors only pull the node to 0.9 V, a 1.4 V threshold is never crossed, and the circuit doesn't wait for it.
• Power-up is real. An empty capacitor from a node to ground starts that node at 0 V. A capacitor to the supply starts it at the supply, which is how a power-on reset works.
• A capacitor keeps its charge. If a switch or button ties a node straight to a rail and is then released, the node starts from the voltage the rail left on its capacitor and drains from there, as a debounce or hold-up circuit does.
• A capacitor wired to nothing is ignored. One whose other lead sits in an empty column, or off the board, holds no charge and does not slow the node down.
• A step carries through a capacitor. A capacitor's charge can't change in an instant, so when the far side of one jumps, the node on this side jumps with it and then settles back through its resistors. A button feeding a 555's TRIG through a capacitor triggers it, and the two-inverter RC oscillators (the CD4060B's and CD4541B's) work the way their datasheets draw them, the junction kicked past the supply each time.
• The analyzer draws voltages. A channel on a logic analyzer is drawn as its wire's voltage, so a charging capacitor curves up toward its supply instead of stepping from LOW to HIGH, and a 74LS signal swings between about 0.2 V and 3.6 V. The channel's value reads in volts.

Slow curves cost nothing to run: Spice Lite calculates when a node will cross a threshold rather than stepping toward it, so a 10-second RC delay runs as cheaply as a microsecond one. If the app falls behind (a busy computer, or its window in the background), it catches up on the crossings it missed, in order.

An oscillation faster than the desk can show (anything above 1 kHz, the fastest clock on offer) is recognised once it has gone round twice the same way. From then on it is drawn at 1 kHz with its duty cycle kept, as the standard engine draws a fast 555, while time itself runs at its true rate: a timer's readout shows the true frequency, and a CD4060B or CD4541B counting its own oscillator counts every real cycle. Anything that changes the circuit (a reset, an edited value, a supply moving) ends the fast drawing, and the oscillator is worked out afresh. An RC round an ordinary (non-Schmitt) inverter has no second threshold to swing to: it chatters at its one threshold and is reported as oscillating.

Timers

In Spice Lite a timer is the circuit inside it, not a formula. Each one is built from its datasheet's own block diagram: comparators reading the voltages on its pins against their references, the package's own resistors, a discharge transistor, and the flip-flops behind them. Nothing in it computes a period. The period is whatever your resistor and capacitor actually do, so the circuits the formulas never covered work too.

• The 555: three 5 kΩ resistors from VCC to ground inside it set its two trip points at ⅔ and ⅓ of the supply (CONT is the top tap), the threshold and trigger comparators set and reset its latch, and DISCH is a real open-collector transistor. The familiar 0.693 is really ln 2 and 1.1 is ln 3, so the usual circuits run within a percent or two of the formula; the capacitor starts empty, so an astable's first HIGH is about 1.6 times as long as the rest. A voltage on CONT moves both trip points, RA built from two resistors in series works, a trigger through a capacitor fires it, RESET reads its own 0.7 V threshold, and a DISCH asked to sink more than its 200 mA rating says so.
• The CD4047B: R and C are driven in opposition from its oscillator's input on RC COMMON, which switches at half the supply, exactly the swing its datasheet's own design formulas describe: 4.40·RC astable, 2.48·RC for a one-shot, a retrigger running it on by whole 2.2·RC periods.
• The CD4098B, CD4528B and CD4538B: a trigger turns on the transistor that empties Cx, and the pulse ends when Rx has charged it back up to the upper reference. Their datasheets give only the formula (½·RC, 0.2·RC·ln VDD, RC), so the references are worked back from it (the lower at 5 % of the supply), and at 100 kΩ and 100 nF every width comes out within about a percent of the sheet (a CD4528B at 5 V runs 1.2 % long). A small Rx with a large Cx strays further, because the discharge transistor's own resistance starts to count: a CD4538B at 10 kΩ and 1 µF runs 2.3 % long. A retrigger empties the capacitor again, and RESET holds it empty. The discharge transistor is held to its datasheet's least Rx — 5 kΩ for the CD4098B, 4 kΩ for the CD4538B (the CD14538B sheet), and 5 kΩ for the CD4528B (the smallest its sheet tests at; it states no minimum). What it carries while it holds the capacitor — RESET LOW against Rx — is checked against what that least Rx would let through (the transistor's own resistance included, so an Rx exactly at the minimum is fine): a smaller Rx gives a Timing resistor too small warning, and past 100 mW in the transistor the brown smoke. The burst of current as it empties Cx at a trigger is not checked: real silicon does the same, and the datasheets bound it by the largest Cx instead (100 µF for the CD4098B and CD4538B, which a larger capacitor is warned about).
• The CD4060B and CD4541B: Rx, Cx and Rs on one junction, as in their datasheets. The junction is kicked past the supply each time an output switches, and Rs keeps the input's protection diodes off it, which is why the datasheets ask for it: leave Rs out and the period is visibly shorter. Both run at about 2.23·Rx·Cx with Rs at twice Rx, the 4060's 2.2 and the 4541's 2.3 on their sheets (so the 4060 runs about 1.5 % slow of its formula and the 4541 about 3 % fast of its own). A CD4541B with AUTO RESET enabled (pin 5 wired LOW) draws the extra supply current its datasheet gives for it — 7 µA at 5 V, 30 µA at 10 V, 80 µA at 15 V. The formula leaves out the chip's own outputs, each about 400 Ω at 5 V, which sit in series with Rx and Cx: harmless with Rx at 10 kΩ and up, but at Rx = 1 kΩ a CD4060B's period comes out about 23 % longer than 2.2·Rx·Cx.

A running timer's readout shows what it measured: the frequency between its last two rising edges, or the length of its last pulse. The Properties card's Timing row still gives the datasheet's figure, worked out from the values you set. Under Spice Lite a timer never complains that it doesn't recognise its wiring: it does whatever its pins make it do. It still says when a pin it needs is left unwired, or a terminal that must be grounded isn't (a CD4528B's T1): it cannot time at all then. The probe shows the real voltage on every timing pin (THRES and TRIG, CONT, RX CX, RC COMMON, an oscillator's junction).

The CD4000 timers' outputs, and the 555's, are as strong as their datasheets say, and are judged by the same current limits as any other output.

Current and fan-out

Every input wired to an output draws a little current, as its own circuit does: a 74LS input pushes about 0.2 mA out of a pin held LOW (it is a resistor and a diode to its supply inside), and nothing once its pin is past about 1.3 V; a CMOS input, or a memory's or processor's, draws nothing at all. So an output with many inputs on it is pulled away from its level by them, and the voltage it ends up at is the one every input on it reads.

• A Brownout warning names an output whose load holds its net where an input on it no longer reads the level the output is driving — though it would at the output's own unloaded voltage. It says the pin, the level and the voltage the net is held at. Too many inputs is one cause; a resistor to ground that is too small for a HIGH (a 74LS HIGH into 100 Ω sits at about 1.6 V) is another. A brownout never damages anything: it is a circuit that does not work, not a part in danger.
• An output that never reaches an input's threshold even unloaded — a 74LS HIGH's 3.6 V into a CMOS input on 12 V — is a level mismatch, not a brownout: that input simply reads it as unknown.

A CD4000 output at 5 V holds one 74LS input LOW at about 0.1 V, two at 0.2 V, five at 0.4 V and six at 0.45 V — all LOW to them, typically, though the datasheet guarantees only one (which is what the standard engine's fan-out warning goes by; under Spice Lite the voltage replaces it). The CD4049UB and CD4050B buffers, whose LOW is about five times stronger, are what belongs between the families in a design that has to work with every part off the shelf.

Every part of a family shares that family's numbers: a 74LS244 bus driver drives as every other 74LS output does, so a circuit behaves the same whichever maker's 74LS parts you picture on the bench. An output that is switched off (a tri-state output not enabled) drives nothing and is no part of a brownout.

Shorts and overloaded pins

What damages an output is the current through it, which Spice Lite knows for every output pin, whatever it is driving, and holds each to what its family is made for:

Past the smoke limit the chip is drawn burnt with brown smoke rather than grey, drives nothing for the rest of the run, and a Brown smoke! notice says why.

A 74LS HIGH shorted to ground carries about 30 mA (a warning); a LOW shorted to the supply about 190 mA (brown smoke). Two 74LS outputs fighting each other carry about 24 mA. An ordinary CD4000 output at 5 V cannot pass enough current to hurt itself; at 15 V a short lets out the smoke. The CMOS parts built to drive more — the CD4049UB/CD4050B buffers' LOW, the CD4511B's segment drivers — are held to the same 100 mW per output transistor their datasheets give every B-series output, so a buffer's LOW shorted to a 15 V supply smokes too. The NE555's output is held to its own sheet's 200 mA.

An analog switch channel (CD4066B, CD4051B/52B/53B) is rated for 10 mA: past it a Switch overloaded warning, and past 25 mA brown smoke. A CD4066B channel switched on straight across a 5 V supply carries 10.6 mA.

A transistor is held to the common limits of its kind:

A transistor past its smoke limit is said with a warning that a real one would have failed, but it carries on conducting: it has no supply pins, so nothing on the desk can be switched off for it.

Inputs have limits too:

• A 74LS input over 7 V lets out the smoke — a CD4000 output on 12 V wired into one, for instance.
• A CD4000 input held beyond its own supply (more than 0.5 V above its VDD, or below ground) conducts through its protection diode, which is warned about, and lets out the smoke past 10 mA. The CD4049UB and CD4050B have no diode to their supply, which is what lets them take a higher voltage than they run on: they are the level shifters. A memory's, processor's or peripheral's input has the same two diodes and the same limits, and so does an analog switch's control pin and a CD4007UB's gate.
• A CD4000 input left in its undefined band draws about 0.5 mA from its own supply while it is there, as both of its input transistors are part-way on. (A CD4007UB's gate is a bare transistor's, and draws nothing of its own: what its pair conducts is the current.)

A short through a transistor or an analog switch — one switched on with its two ends on opposite supplies — is reported as a Short circuit only when the current really flowing through it is 100 mA or more, or the supply it is on has hit its current limit. The standard engine reports every such join; under Spice Lite a transistor whose base is fed through 10 MΩ, passing microamps, is no short. A short with no part in it (two supplies wired together) is always reported.

Stop restores every chip that let out its smoke.

Power supplies

A power supply brick gains a Current limit in its Properties (100 mA to 5 A; 1 A unless you change it). While Spice Lite is running, the brick shows the current being drawn under its voltage.

• Below its limit, a supply holds its set voltage.
• Past its limit, the voltage droops in proportion. Asking 200 mA of a 100 mA supply gives you about half the voltage. The readout turns amber and shows both the voltage and the current.
• A supply shorted + to − (its + wired to a ground) is held at its limit and droops to a fraction of a volt, with a Short circuit notice; every chip on it is underpowered.
• A chip whose supply droops below its rating is underpowered (an amber dot; hover it for the voltage it sees), exactly as if you had set the supply too low. Its load still counts while it is: a chip whose own outputs drag its supply down stays down until you change something, rather than flickering on and off.

The supply's load is every chip's own supply current (1.6 mA for a 74LS part, almost nothing for CD4000), plus everything the voltages make flow: the current through every resistor, LED, diode, transistor and switch from the supply, and out of every chip output into whatever it drives (see LEDs, below): a red LED behind a 330 Ω resistor on 5 V draws about 9.4 mA, and one wired straight across the supply draws everything it can until it burns. Current a chip sinks is routed back through its own ground pin.

Chips powered off the rails

A chip whose supply pins are not wired straight to the rails — fed through a resistor, a diode, a transistor, or from another chip's output — runs at the voltage that actually reaches its pins. Its own supply current is a load on whatever feeds it, and so is everything its outputs drive: an LED lit from one of its outputs is fed through its VCC pin, so a 74LS04 fed through 47 Ω sags to about 4.6 V lighting one — underpowered, and still lighting it. A 74LS chip fed through a diode gets about 4.4 V and is underpowered (it needs 4.75 V); a CD4000 chip, which runs from 3 V, works on the same feed, and its HIGH outputs are the 4.4 V it is running on. Its outputs and inputs are measured from its own pins: a chip whose ground is raised drives its LOW at that ground, and a chip across two supplies' rails (VDD on 12 V, VSS on the 5 V rail) runs on the 7 V between them, its LOW at 5 V and its HIGH at 12 V. The standard engine sees only a supply pin that is not on a supply.

Diodes and transistors

Every diode and transistor is one of a single common kind, whatever part number you give it:

• A diode conducts from 0.6 V; a Zener also conducts backwards at its Zener voltage, so it clamps and regulates. One that carries more than its junction can take burns, like an LED.
• A bipolar transistor (NPN, PNP) has gain: its base conducts from 0.65 V, and its collector carries up to 100 times the base current — no more than the circuit lets through, in which case it is saturated, at 0.2 V. A base fed through 10 MΩ barely turns it on, where the standard engine treats any HIGH base as a closed switch.
• A MOSFET turns on from a 2 V gate threshold, measured from its source, and is fully on (1 Ω) 2 V past it — so a 5 V gate turns it fully on. Its gate draws nothing and keeps the voltage it was last driven to when left floating; the transistor's lamp rings amber while it does. The CD4007UB's six transistors are MOSFETs of the same kind.
• An analog switch (CD4066B, CD4051B/52B/53B) closes and opens by the voltage on its control pin, read through its thresholds like any input, and its channel passes whatever voltage is on it through its on-resistance.

A transistor's lamp on the desk lights while it conducts.

LEDs

With Spice Lite on, an LED carries the current its circuit really pushes through it, worked out from its colour's datasheet, and what that current does to it is what it would do on a bench. The standard engine only asks whether something limits the current; Spice Lite asks how many milliamps.

These are ordinary 5 mm LEDs. Every segment of a display and every bar of a bar graph is taken as an LED of its colour.

• Nothing flows below the knee. A red LED starts conducting at about 1.8 V, a blue or white one at about 2.8 V; above that the voltage climbs slowly with the current. A blue LED behind 100 Ω on a 3 V supply barely glows (1.6 mA); on 5 V it is bright (18 mA).
• Brightness follows the current. An LED is drawn at full brightness at the current its datasheet quotes its brightness at (10 mA, or 20 mA for blue and white), dimmer below it, and with a wider glow above it. Below 50 µA it is dark.
• Past its rating it is overdriven. It lights brightly and a warning names the current against its rating: a red LED behind 220 Ω on 9 V takes 31 mA. A real one would dim and fail early.
• Past its maximum junction temperature it burns. The datasheet gives how hot the junction runs per watt; once the heat the current puts in would take the junction past its maximum, the LED burns out (a red cross and smoke) and stops conducting for the rest of the run. Segments sharing one resistor with it then share its current too. Stopping the simulation restores it. The package taking a few seconds to warm up is not modelled: it burns at once.
• Backwards, it is rated for 5 V. More than that across it the wrong way gives a warning.
• The probe reads its current. Point the probe at an LED's leg — or at the resistor beside it, or the chip output driving it — to see the milliamps through that lead. No part shows a current on the desk; only a power supply shows its draw.

A chip output is not a perfect source. Each output is modelled as its datasheet says it behaves:

• A 74LS output driving HIGH is about 3.6 V behind 120 Ω, so an LED wired straight from it to ground lights at about 14 mA and survives. Driving LOW it has almost nothing in the way, so an LED from the supply straight into it takes over 80 mA and burns.
• A CD4000 output is a small transistor that can only pass so much: about 4 mA at 5 V, 16 mA at 10 V and 28 mA at 15 V. At 5 V that is a resistor in all but name.
• The NE555's output is good for 200 mA, and burns an LED wired straight to it. The CD4511B's segment outputs, and the CD4049UB and CD4050B's sinks, are stronger than an ordinary CD4000 output, as their datasheets say.

Character LCDs

The LCD modules' backlight is an LED of the module's colour between A and K, behind the 100 Ω resistor the common 1602A and 2004A boards carry for it. Wired to the supply it lights and draws its current (about 18 mA for a blue module on 5 V, 28 mA for a green one); unwired, the panel is dark. Its rating is the module maker's, so it never warns or burns.

V0 sets the contrast. The glass is driven by the voltage between VDD and V0: full contrast from 3 V (the controller's minimum) up, fading below that, and blank with V0 at the supply or left unconnected. The usual 10 kΩ potentiometer from the supply to ground, its wiper on V0, works as it does on a bench. With Spice Lite off, the backlight and contrast are cosmetic as before.

Wire resistance

Every jumper wire is treated as 24 AWG copper at its real length, about 0.084 Ω per metre. Current from a supply is routed through the wiring to each chip, and a chip loses the voltage dropped across the wires its current shares with others. A chip at the end of a long chain of jumpers sees a little less than the chip next to the supply.

The effect is honest but small: a 10 cm jumper is under 0.01 Ω, so it takes a heavy load to see it. Drops under a millivolt are ignored.

Decoupling capacitors

Each time a chip's output switches, it draws a brief extra spike of current to charge what it drives. When many outputs switch at once, those spikes add up, and on a supply that is already close to its limit they can tip it over. Spice Lite warns with Supply spike.

A capacitor wired across a chip's supply (VCC to ground, the usual 100 nF beside every chip) supplies those spikes itself, and the supply never sees them. Spice Lite does not know where along the rails a capacitor sits or what its value is: one capacitor anywhere across the rails a chip's supply pins are on decouples that chip. A capacitor from a signal net to ground is a timing capacitor, not a decoupling one, and doesn't count.

The Spice Lite settings

Settings ▸ Spice Lite holds:

• Spice Lite: On or Off.

• Settle gap: how close a capacitor's voltage has to get to its final value before the probe and the analyzer stop redrawing it (an input watching it is told the moment it crosses, whatever this says). 1 % is about five time constants. It never holds the circuit up.

• A TTL | CMOS strip showing the families your tray shows (and any family the open project uses). Under each are its datasheet source, a Reset to defaults button, and an Advanced section with every number Spice Lite uses for that family:

  • Gate delay: how long an output takes to follow its inputs.
  • Output source and sink current: the strength of the family's output — the current it delivers at its datasheet's test voltage. Twice the figure is an output twice as strong (half the resistance behind it, and twice the current a CMOS output saturates at).
  • Input current, LOW: what a 74LS input pushes out of a pin held LOW (the figure is the datasheet's maximum, and the input pushes about half of it at 0.4 V, typically). A CMOS input's is too small to measure, and adds nothing unless you set it to something that is.
  • Supply current per chip, input LOW and HIGH thresholds and the switching load (what each output charges when it switches).

  CMOS values are stated at 5 V: the gate delay and the input thresholds follow each chip's supply, and the currents and the switching load are used as their 5 V figures at any supply. A part in no family (the memories, processors and peripherals) is no family's: its outputs are a common rail-to-rail stage, and its inputs read at the 74LS thresholds.

Every change, the numbers included, applies at the next Run: a circuit that is running keeps the settings it started with. Your changes apply to every project on this computer. Each field has a range — hover over it to see it — wide enough for any logic part (a gate delay of up to 10 µs, a threshold of up to 5 V): a number outside it is set to the nearest end of it, so 20,000 ns becomes 10,000 ns and 0 becomes the smallest value allowed. A field that won't read at all, or a LOW threshold at or above the HIGH one, turns red, keeps the previous value, and says why underneath. A very short gate delay is taken as no more than about 64 times faster than the slowest gate on the desk, so a run never crawls.

What stays the same

• Exports: the KiCad and Digital exports ignore Spice Lite entirely and use your parts' real values.
• The AI builder and the example circuits are proven on the standard engine.
• The standard engine is untouched: with Spice Lite off, every circuit behaves exactly as before.

What Spice Lite does not model

To keep it light, some things are left out deliberately:

• a capacitor's far side moving smoothly (only its jumps carry through): a capacitor between two nodes that are both charging only carries the steps each takes;
• inductors ramping their current (an inductor is still a wire);
• the timers' comparator references beyond what their datasheets say (the CD4098B, CD4528B and CD4538B are worked back from their formulas) and their internal propagation delays;
• two unrelated oscillations both faster than the desk can show: only a circuit that repeats as a whole is drawn at 1 kHz, so two together are reported as oscillating;
• a CD4047B's special RC COMMON protection diodes (its datasheet's formulas assume the full swing past the supply, and so does Spice Lite);
• two power supplies wired onto the same rail: the load is booked to the first of them only, and neither shares it with the other;
• current sharing across parallel wires, and the resistance of the breadboard's own contacts;
• heat, beyond an LED's or diode's own junction; signal reflections on long wires; and crosstalk between wires.