- The current you need
- The overdrive you want
- The length you choose
- The voltage across it
Its exact widthIts gate voltage
Start from intent
Size from the currents and voltages your circuit needs, not from a guess.
SizeAt Why SizeAt
Not a better map, but a direct road from intent to size.
Chip performance depends strongly on transistor sizing. Yet for every designer, the path from intent to transistor size is still a loop of guesses.
01 A loop
Today, sizing is still a loop.
An engineer estimates an initial size, runs a simulation, analyzes the results, adjusts the dimensions, and repeats, again and again, until the circuit meets its specification.
The naive idea: why not simply try every size?
- 15 transistors
- × 2 width and length
- = 30 design variables
- 10 values each
One failure, and the loop starts again.
The specification must be met at every process corner, supply voltage and temperature. One failure, and the loop starts again.
Even the world's best analog engineers lose countless hours to this loop.
Every hour spent in this loop is an hour not spent on new theory, new architectures and new ideas.
1,000,000,000,000,000,000,000,000,000,000
1030 combinations. Far too many to simulate.
TT cold lowTT cold nomTT cold highTT room lowTT room nomTT room highTT hot lowTT hot nomTT hot highFF cold lowFF cold nomFF cold highFF room lowFF room nomFF room highFF hot lowFF hot nomFF hot highSS cold lowSS cold nomSS cold highSS room lowSS room nomSS room highSS hot lowSS hot nomSS hot highFS cold lowFS cold nomFS cold highFS room lowFS room nomFS room highFS hot lowFS hot nomFS hot highSF cold lowSF cold nomSF cold highSF room lowSF room nomSF room highSF hot lowSF hot nomSF hot high
Check failed at SS cold low supply · back to the guess
02 Why the loop exists
There is no direct path from intent to transistor size.
The conventional flow
W, LVGSimulateID = ?
First pick a size and bias voltages, then see what current flows.
How the designer thinks
Power budgetID per branchVDS headroomW, VG = ?
First split the power budget into branch currents and the supply into voltage headroom, then find the size and bias voltage that give each transistor its current and headroom.
Why do designers think current first?
In a branch, every transistor carries the same current. Current is the only thing common to all of them, so it is the natural place to start sizing.
1960sHand-derived equations
Gave designers intuition.
1980s–90sOptimizers
Let the computer search for sizes, but every search costs many simulations.
1996The gm/ID method
Brought sizing closer to a designer's intuition about efficiency.
Late 2010sMachine learning
Learns from thousands of past simulations, but needs a lot of data and computing power.
Each one was a real step forward, and each one made the loop shorter.
But the path from intent to size is still a long and winding road. Designers have better maps, but not a direct road.
03 What changes
So here is what changes for every designer.
Guess a size, simulate, check, adjust, and repeat.
State your intent as a design point, and receive the exact transistor that delivers it.
Getting from a target current to a final size and bias voltage still takes several steps.
Within seconds: its size, its bias voltage, and a tag with every electrical parameter.
Every time the specification changes or a corner fails, those steps begin again.
See how your chosen size behaves at any other corner or temperature. It's just one more request away.
Scripting to read charts and interpolate.
You no longer calculate or guess a size, because every answer comes complete.
Countless hours lost to the loop.
Those hours go back to new theory, new architectures and new ideas.
04 What you gain
This is more than a faster way to size a transistor. It changes the way designers work.
Hundreds of candidate transistors, from one request.
Every option at once
Hundreds of sizes in one request, each with its key electrical parameters, so you see the trade-offs before you choose.
Your processYour currentsYour voltages
A new processYour currentsYour voltagesNew sizes
Port or start fresh
On a new process, keep your currents and voltages and get the new sizes, or start a new design from scratch.
- New theory
- New architectures
- New ideas
More time to innovate
Less trial and error, more creative design.
Exact
Every answer is exact and ready to use. Just verify it once.
Real
Every value comes from the real transistor, the same reference the whole industry trusts, at sign-off accuracy.
Defensible
Every sizing decision traces back to one exact transistor and its tag, which gives you a clear, defensible answer to “Why this size?”
One simulation of the whole circuit confirms them all: every current is right on target, and every voltage is within half a millivolt.
So, what size should this transistor be? Now you get the answer within seconds.
Try it once, on your next circuit.
To collaborate or to learn more, write to contact@sizeat.com