Optocoupler CTR, SRAM reset restore, voltage-limited current source, macOS paste by esaruoho · Pull Request #241 · pfalstad/circuitjs1

esaruoho · 2026-03-02T11:03:46Z

Summary

Rebased onto v3-dev (replaces #223 which targeted master).

Adds adjustable CTR (current transfer ratio) for OptocouplerElm
Adds "Restore Contents on Reset" option for SRAMElm (snapshots initial state)
Adds voltage-limited current source (FLAG_VOLTAGE_LIMIT on CurrentElm)
Fixes macOS Cmd+V paste in Electron app

Fixes #150, #184, #149, #138.

Changes from master version

Extended v3-dev's dumpXml()/undumpXml() in CurrentElm and SRAMElm to serialize new fields
SRAMElm integrated with v3-dev's parseContentsString() helper

Test plan

Place Optocoupler → edit → verify CTR adjustment works
Place SRAM → load data → enable Restore on Reset → reset → verify data restored
Place current source → enable voltage limiting → verify current reduces at voltage limit
In Electron on macOS → Cmd+V → verify paste works

🤖 Generated with Claude Code

…urce, macOS paste fix - Configurable optocoupler CTR (pfalstad#150): Add CTR % parameter to edit dialog, scale CCCS transfer expression by CTR multiplier. Default 100%. - SRAM restore on reset (pfalstad#184): Add "Restore Contents on Reset" checkbox. Stores initial map snapshot, restores it on simulation reset. - Voltage-limited current source (pfalstad#149): Add compliance voltage mode. When enabled, current source becomes open circuit beyond max voltage. - macOS Electron paste fix (pfalstad#138): Replace Menu.setApplicationMenu(false) with proper Edit menu so Cmd+C/V/X/A work in text fields. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

@MatNieuw

When loading SRAM contents from a binary file, the filename is now captured and displayed next to the Load button as "Loaded: filename.bin". This helps users with multiple SRAMs identify which file was loaded into each one, addressing feedback from @MatNieuw on issue pfalstad#184. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

esaruoho · 2026-03-17T20:49:47Z

@pfalstad any takes on this?

pfalstad · 2026-03-18T01:37:28Z

I tried to do the limited current source thing once and got constant convergence failures. Looks like your version has the same issue. Try opening all the switches. Also I wouldn't have a separate checkbox for the voltage limiting. I would just use a default of 0 meaning unlimited.

<cir f="1" ts="0.000005" ic="10.20027730826997" cb="50" pb="50" vr="5" mts="5e-11">
  <w x="112 32 208 32" f="2"/>
  <w x="208 32 304 32"/>
  <w x="304 32 400 32"/>
  <s x="208 32 208 112" p="1"/>
  <s x="304 32 304 112"/>
  <r x="208 112 208 176" r="100"/>
  <r x="304 112 304 176" r="400"/>
  <r x="400 112 400 176" r="4000"/>
  <w x="208 176 304 176"/>
  <w x="304 176 400 176"/>
  <w x="304 208 304 176"/>
  <w x="304 208 400 208"/>
  <w x="304 208 208 208"/>
  <s x="304 208 304 288"/>
  <r x="208 288 208 352" r="600"/>
  <r x="304 288 304 352" r="200"/>
  <s x="400 208 400 352"/>
  <w x="112 352 208 352"/>
  <w x="208 352 304 352"/>
  <w x="304 352 400 352"/>
  <i x="112 352 112 32" f="1" cu="0.01" mv="5"/>
  <g x="112 352 112 384"/>
  <w x="400 32 400 112"/>
  <w x="208 208 208 288"/>
  <o en="20" sp="32" f="x3" p="0">
    <p v="0" sc="20"/>
    <p v="3" sc="0.05"/>
  </o>
</cir>

pfalstad · 2026-03-18T01:45:40Z

the sram thing is good

pfalstad · 2026-03-18T01:49:30Z

The optocoupler thing looks good. I would maybe just rename "CTR (%)" to "CTR Scale" because it doesn't look like the CTR is exactly 100... At least in my test case it wasn't.

pfalstad · 2026-03-18T01:57:08Z

the mac thing is good

MatNieuw · 2026-03-18T13:55:15Z

An optocoupler's CTR depends also on the current. There is usually a peak CTR between 5 and 20 mA diode drive current, and the peak varies per type. In the datasheets it is given as a percentage (like 80%), or absolute value like 0.8 . I think being able to specify a peak CTR at a certain diode current would be nice. As an example, the CTR specs from a CYN80N.

@pfalstad

Adds a configurable CTR scale factor to the optocoupler element. The underlying base CCCS expression already models a ~100% CTR baseline device; the new ctr field multiplies that response so users can model parts with different transfer ratios. Per @pfalstad feedback on pfalstad#241: the dialog field is labelled "CTR Scale" rather than "CTR (%)" because the actual transfer ratio of the base curve is not exactly 100%, so calling the input a percentage would be misleading. The value now reads as a dimensionless multiplier (e.g. 1.0 = baseline, 0.5 = half, 1.5 = 150% of baseline). Existing circuits keep working because the dump format already stored ctr directly, not as a percentage. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

@pfalstad

Adds an optional compliance voltage to CurrentElm. The redesign addresses both pieces of @pfalstad feedback on PR pfalstad#241: - No checkbox. Default maxVoltage = 0 means unlimited (ideal current source); a positive value enables compliance. The dialog field is labelled "Max Voltage (V, 0=unlimited)". - Convergence with switches open. The original implementation hard-stepped between "stamp ideal source" and "stamp open" once |vd| crossed maxVoltage. Newton-Raphson then ping-ponged between the two stamps with no fixed point, especially with high-impedance loads (open switches, large resistors). Replaced with a smooth tanh-shaped saturation: i(vd) = currentValue * 0.5 * (1 - tanh((|vd| - maxVoltage)/vt)) with vt = 5% of maxVoltage. The Norton companion model stamps the exact analytic Jacobian (di/dvd = -currentValue * 0.5 * sech^2(arg) / vt * sign(vd)) plus a small gmin so the matrix never goes singular when sech^2 is near zero. Result: the source rolls off smoothly to zero current as |vd| approaches maxVoltage rather than slamming between full-on and full-off, which is what was killing convergence in the test circuit @pfalstad shared on pfalstad#241. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

esaruoho · 2026-04-15T06:15:46Z

Closing in favor of four narrower PRs, one per change, with your feedback applied to each:

macOS Electron: fix Cmd+C/V/X/A in text fields (#138) #334 — macOS Electron Cmd+C/V/X/A fix (The CircuitJS1 app on MacOS does not accept pasting text inside the text box #138). Unchanged from your "the mac thing is good".
SRAM: restore loaded file on reset; show filename in edit dialog (#184) #335 — SRAM restore-on-reset and filename in edit dialog (Feature- Add SRAM option to make file association persistent and reload on reset #184). Unchanged from your "the sram thing is good", plus @MatNieuw's filename request.
Optocoupler: add CTR Scale parameter #336 — Optocoupler CTR scale, with the "CTR Scale" rename you asked for (no longer labelled as a percentage).
Voltage-limited current source with smooth saturation #337 — Voltage-limited current source, redesigned per your feedback: no checkbox (default maxVoltage = 0 = unlimited) and smooth tanh saturation with analytic Jacobian for Newton convergence — should fix the open-switch convergence failure you hit.

Sorry for bundling four separate things originally. Each new PR is reviewable on its own.

@MatNieuw's CTR-vs-current-curve request from this thread isn't in #336 — the polynomial would need to be reshaped rather than just scaled. Happy to do that as a follow-up if Paul wants it.

@pfalstad

Adds a configurable CTR scale factor to the optocoupler element. The underlying base CCCS expression already models a ~100% CTR baseline device; the new ctr field multiplies that response so users can model parts with different transfer ratios. Per @pfalstad feedback on pfalstad#241: the dialog field is labelled "CTR Scale" rather than "CTR (%)" because the actual transfer ratio of the base curve is not exactly 100%, so calling the input a percentage would be misleading. The value now reads as a dimensionless multiplier (e.g. 1.0 = baseline, 0.5 = half, 1.5 = 150% of baseline). Existing circuits keep working because the dump format already stored ctr directly, not as a percentage. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

@pfalstad

Adds an optional compliance voltage to CurrentElm. The redesign addresses both pieces of @pfalstad feedback on PR pfalstad#241: - No checkbox. Default maxVoltage = 0 means unlimited (ideal current source); a positive value enables compliance. The dialog field is labelled "Max Voltage (V, 0=unlimited)". - Convergence with switches open. The original implementation hard-stepped between "stamp ideal source" and "stamp open" once |vd| crossed maxVoltage. Newton-Raphson then ping-ponged between the two stamps with no fixed point, especially with high-impedance loads (open switches, large resistors). Replaced with a smooth tanh-shaped saturation: i(vd) = currentValue * 0.5 * (1 - tanh((|vd| - maxVoltage)/vt)) with vt = 5% of maxVoltage. The Norton companion model stamps the exact analytic Jacobian (di/dvd = -currentValue * 0.5 * sech^2(arg) / vt * sign(vd)) plus a small gmin so the matrix never goes singular when sech^2 is near zero. Result: the source rolls off smoothly to zero current as |vd| approaches maxVoltage rather than slamming between full-on and full-off, which is what was killing convergence in the test circuit @pfalstad shared on pfalstad#241. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Paul's pfalstad#241 test circuit with multiple open switches was still failing. Root cause: vt was 5% of maxVoltage (0.25V for a 5V limit), giving a near-discontinuous Jacobian. g jumps by ~1e7 across a few tens of mV near the compliance threshold, so Newton pings between the linear and saturated regimes. Changes: - vt = max(0.2 * maxVoltage, 0.1V) spreads the knee so sech^2 varies smoothly across ~1V instead of ~0.25V. - Gmin floor raised from 1e-9 to 1e-6 to keep the Norton resistor well below the matrix stiffness scale in open-circuit topologies. - Removed the explicit converged=false trigger; the solver's own convergence criteria are sufficient once the Jacobian is well-shaped.

esaruoho and others added 2 commits March 6, 2026 08:21

esaruoho force-pushed the v3-four-more-fixes branch from 5d6f4bb to c4d25b8 Compare March 6, 2026 06:23

This was referenced Apr 15, 2026

macOS Electron: fix Cmd+C/V/X/A in text fields (#138) #334

Closed

SRAM: restore loaded file on reset; show filename in edit dialog (#184) #335

Merged

esaruoho mentioned this pull request Apr 15, 2026

Optocoupler: add CTR Scale parameter #336

Open

5 tasks

esaruoho mentioned this pull request Apr 15, 2026

Voltage-limited current source with smooth saturation #337

Open

6 tasks

esaruoho closed this Apr 15, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Optocoupler CTR, SRAM reset restore, voltage-limited current source, macOS paste#241

Optocoupler CTR, SRAM reset restore, voltage-limited current source, macOS paste#241
esaruoho wants to merge 2 commits intopfalstad:v3-devfrom
esaruoho:v3-four-more-fixes

esaruoho commented Mar 2, 2026

Uh oh!

esaruoho commented Mar 17, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

MatNieuw commented Mar 18, 2026

Uh oh!

esaruoho commented Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

esaruoho commented Mar 2, 2026

Summary

Changes from master version

Test plan

Uh oh!

esaruoho commented Mar 17, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

pfalstad commented Mar 18, 2026

Uh oh!

MatNieuw commented Mar 18, 2026

Uh oh!

esaruoho commented Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants