P10.9.2 Tune the CTLE Zero

MediumOptimizengspice 42 · serverReview the topic

Place the zero of a passive CTLE. The long channel's eye must open past 0.95 UI, with no peaking above 1.5 dB.

Circuit context: The digital blocks of a serializer/deserializer: line encoder and decoder, scrambler, CRC and error-correcting code. Keep the line balanced and busy, and detect or correct the errors a fast serial link makes.

After the long RC-ladder channel of P10.9 comes your passive CTLE, then a gain stage of 4. The CTLE is R1 with C1 across it, in series, and R2 to ground.

At low frequency the CTLE divides by R2 / (R1 + R2). Above its zero the capacitor bypasses R1 and the high frequencies pass. That tilt must undo the channel's.

Choose R1, C1 and R2. The eye must be at least 950 mV tall and 0.95 UI wide. The response of channel, CTLE and gain together must not peak above +1.5 dB between 10 MHz and 20 GHz.

What the Grader Measures

Your subcircuit ctle (ports in, out) under the grader's own bench:

CheckRule
Eye height≥ 950 mV
Eye width≥ 950 mUI
Peaking of channel, CTLE and gain≤ 1.5 dB
Full specification
The subcircuit the grader tests (keep the name and the ports)
.subckt ctle in out
R1 in out 300
C1 in out 0.5p
R2 out 0 100
.ends
CheckRuleHow the grader measures it
Eye height≥ 950 mVtran over 137 bits of PRBS7. v(rx), after the gain of 4, is folded on the 100 ps unit interval as in the lesson eye plots
Eye width≥ 0.95 UIthe same fold: the span of sampling instants at which every bit is decided right
Peaking≤ +1.5 dBac from 10 MHz to 20 GHz on a second copy of channel, CTLE and gain: the largest gain

The channel and the gain stage are the lesson's, with ideal sources. The limits are bench values.

Examples

Example 1

The Lesson's CTLE: 0.5 pF
Design
.subckt ctle in out
* R1 with C1 across it in series, R2 to ground; the grader follows it with a gain of 4
R1 in out 300
C1 in out 0.5p
R2 out 0 100
.ends
Measured by the grader
Eye height: 1.004 V (needs ≥ 950 mV)
Eye width: 900 mUI (needs ≥ 950 mUI)
Peaking of channel, CTLE and gain: 3.571 dB (needs ≤ 1.5 dB)

A tall eye, but the boost overshoots: the eye is too narrow and the response peaks at 3.6 dB.

Example 2

Zero Too High: 0.1 pF
Design
.subckt ctle in out
* R1 with C1 across it in series, R2 to ground; the grader follows it with a gain of 4
R1 in out 300
C1 in out 0.1p
R2 out 0 100
.ends
Measured by the grader
Eye height: 646.1 mV (needs ≥ 950 mV)
Eye width: 920 mUI (needs ≥ 950 mUI)
Peaking of channel, CTLE and gain: -0.00006216 dB (needs ≤ 1.5 dB)

The zero sits at 5.3 GHz, above the 5 GHz Nyquist frequency. The CTLE hardly boosts, and the ISI stays.

Example 3

A Flat Divider: No Capacitor
Design
.subckt ctle in out
R1 in out 300
R2 out 0 100
.ends
Measured by the grader
Eye height: 466.4 mV (needs ≥ 950 mV)
Eye width: 820 mUI (needs ≥ 950 mUI)
Peaking of channel, CTLE and gain: -0.00007663 dB (needs ≤ 1.5 dB)

With no zero the CTLE divides every frequency by 4, and the gain of 4 puts it back: the bare channel's eye.

Constraints and Contract

  • The netlist defines .subckt ctle in out ... .ends exactly once, with these ports in this order.
  • Inside ctle: resistors, capacitors and .param lines only.
  • Everything outside ctle is yours: the grader uses its own bench.

Hints

Hint 1

The zero sits at 1 / (2π R1 C1). With R1 = 300 Ω and C1 = 0.5 pF that is 1.06 GHz.

Hint 2

Too much boost overshoots the edges: the eye gets taller but narrower, and the AC run peaks. Too little leaves the ISI.

Hint 3

Keep R1 and R2 and move C1 alone. The lesson's 0.5 pF peaks at about +3.6 dB; a little more than half of it is enough.

What the grader checks: The grader places your subcircuit between the lesson's RC-ladder channel and its gain of 4. It folds v(rx) into an eye with the lessons' eye code and reads the peak of an AC run. Runs use ngspice 42, measured 2026-11-12. It proves this bench channel, not a real board, noise or crosstalk.

Run simulates your netlist as written, with your own analyses and measurements. Submit sends it to the grader, which keeps only the subcircuit ctle and makes 1 ngspice run. The numbers come from ngspice and the SKY130 models the site pins; they prove the measured cases, not every condition a chip meets.

See all 2 steps and the final capstone

Report a problem with this page

Your netlist SPICESaved locally
Ln 1, Col 1Tab indents · Ctrl/⌘ + Enter runs
Not run yet
Run your netlist to see its measurements and the ngspice log here.
ngspice 42 · server