The 4N25 and 4N35 share identical pinouts and the same LED-to-phototransistor isolation architecture, but their guaranteed minimum CTR can differ depending on manufacturer and device grade. In practice, the 4N35 often specifies a higher minimum CTR floor, which translates to more collector current at the same LED drive—critical when you're working with weak pull-ups or feedback loops that need predictable gain. The real question isn't which part is "better," but whether your circuit has enough margin to tolerate the guaranteed worst-case CTR of whichever device you choose.
Key takeaways:
- Pin-compatible does not mean electrically interchangeable; verify guaranteed minimum CTR, not typical values, in the specific datasheet before substituting.
- A higher guaranteed CTR provides more collector-current margin at lower LED drive, which matters in microcontroller interfaces with tight current budgets or power-supply feedback paths sensitive to loop gain.
- Both are general-purpose phototransistor devices with similar switching limitations; if your application requires sub-microsecond propagation or predictable timing, move to a high-speed optocoupler or digital isolator.
- CTR degrades with LED aging and temperature; designs relying on typical CTR often work in prototypes but fail in production or field conditions.
What the 4N25 and 4N35 Actually Are
Both devices package an infrared LED and an NPN phototransistor in a single optically coupled, electrically isolated enclosure. Drive current through the LED, and its infrared emission switches the phototransistor, transferring a signal across the isolation barrier without a galvanic connection.
The architecture is identical. The functional difference comes down to how manufacturers bin and grade the devices during production, which determines the guaranteed minimum CTR and, in turn, the collector current you can rely on in worst-case conditions.
Pinout and Internal Structure
Most 6-pin DIP versions follow this standard layout:
| Pin | Function |
|---|---|
| 1 | Anode |
| 2 | Cathode |
| 3 | No connection |
| 4 | Emitter |
| 5 | Collector |
| 6 | Base |
Pins 1 and 2 connect to the LED. Pins 4 and 5 are the phototransistor emitter and collector. Pin 6 exposes the base, which is often left floating in simple switching circuits but can be tied through a resistor to emitter to adjust sensitivity, leakage, or switching behavior.
Physical pin compatibility makes drop-in replacement tempting, but it's a trap if the replacement part's guaranteed CTR is lower than what the original design assumed.
Electrical Characteristics That Actually Affect Circuit Behavior
Current Transfer Ratio: The Only Spec That Really Separates These Parts
CTR is the ratio of phototransistor collector current to LED forward current, expressed as a percentage:
CTR = (I_C / I_F) × 100%
This is not a fixed transistor beta. CTR varies with LED current, collector-emitter voltage, temperature, individual device tolerance, and LED degradation over time.
From my experience, the most common design mistake is using typical CTR from a datasheet chart instead of the guaranteed minimum at your actual operating conditions. A part with 50% typical CTR might have a 20% guaranteed minimum, and that's the number your circuit has to survive.
The 4N35 often carries a higher guaranteed minimum CTR than the 4N25, but this is manufacturer- and grade-specific. Some 4N25 variants are binned with CTR floors comparable to certain 4N35 grades. The only reliable approach is to compare the actual datasheet sections for "guaranteed" or "minimum" CTR under conditions matching your LED current and collector voltage.
Input LED Characteristics
The LED side behaves like any infrared LED: it needs current limiting, respects a forward voltage drop (typically 1.0–1.5 V depending on current and temperature), and has maximum ratings for continuous and peak forward current.
The LED current-limiting resistor is:
R_LED = (V_IN − V_F) / I_F
Where V_IN is your drive voltage, V_F is the LED forward voltage, and I_F is your target LED current.
One detail that doesn't appear in most app notes: LED forward voltage drops slightly as the device ages, which means LED current creeps up if your resistor was calculated at the edge of the rating. I've seen circuits pass initial testing but fail after thermal cycling because the aged LED drew enough extra current to shift the phototransistor into a different part of its CTR curve.
Output-Side Ratings and Saturation Voltage
The phototransistor has maximum ratings for collector-emitter voltage, collector current, and power dissipation, but these alone don't tell you whether the device can actually pull your output LOW.
What matters is whether the collector current available from your guaranteed CTR can sink the current demanded by your pull-up resistor while keeping V_CE(sat) below your logic threshold.
In switching applications, saturation voltage matters more than people expect. If your logic input threshold is 0.8 V and the optocoupler's V_CE(sat) at your required collector current is 0.5 V, you have 0.3 V margin. That sounds fine until temperature or device variation pushes V_CE(sat) to 0.7 V, and suddenly your logic input floats.
Isolation Voltage vs. Working Voltage
Both devices specify an isolation test voltage, often in the range of 2500–5000 V RMS for one second. This is a dielectric withstand test, not a continuous working voltage rating.
For safety-critical designs—especially anything connected to mains—you need to verify working voltage, creepage distance, clearance, insulation class, and compliance with the relevant safety standard (UL, IEC, VDE). The optocoupler datasheet gives you isolation test voltage and sometimes a "basic insulation" or "reinforced insulation" designation, but the final safety margin depends on your PCB layout, enclosure, and system-level design.
Switching Speed: Why These Parts Are Slow
The 4N25 and 4N35 are general-purpose phototransistor optocouplers, not high-speed devices. Typical propagation delays are in the range of microseconds to tens of microseconds, depending on LED drive, collector load, and how deeply you saturate the phototransistor.
Driving the phototransistor hard into saturation stores charge in the base region, which slows turn-off. Reducing saturation improves speed but reduces noise margin. This tradeoff is manageable for control signals, status flags, and low-frequency feedback, but it's a poor fit for serial communication faster than a few kbps or any application where propagation delay variation matters.
If your application needs predictable timing or data rates above ~100 kbps, you're better off with a high-speed optocoupler (6N137, HCPL-2630) or a digital isolator (Si86xx, ISO78xx).
How CTR Differences Play Out in Real Circuits
Microcontroller GPIO Isolation
A common use case: a microcontroller GPIO drives the LED side, and the phototransistor side sinks current through a pull-up resistor to generate a logic LOW.
Assume:
- GPIO output: 5 V
- LED forward voltage: 1.2 V
- Target LED current: 10 mA
R_LED = (5 V − 1.2 V) / 10 mA = 380 Ω
Now consider two hypothetical devices, one with 20% guaranteed minimum CTR and another with 100% guaranteed minimum CTR. These are illustrative values to show the impact of CTR, not universal specs for the 4N25 or 4N35.
At 10 mA LED current:
- 20% CTR → I_C = 2 mA
- 100% CTR → I_C = 10 mA
If the output pull-up is 4.7 kΩ to 5 V, the pull-up current is approximately:
I = 5 V / 4.7 kΩ ≈ 1.06 mA
Both devices can sink that current. But if you drop the pull-up to 1 kΩ:
I = 5 V / 1 kΩ = 5 mA
Now the 20% CTR device, with only 2 mA available collector current, cannot pull the output LOW. The 100% CTR device still has margin.
This is why guaranteed CTR matters. A circuit tested with a handful of parts from the same production lot might work fine, then fail when you source a different manufacturer or grade with a lower CTR floor.
Power-Supply Feedback Loops
In isolated DC-DC converters and offline power supplies, the optocoupler sits in the feedback path, translating output-side error signals into collector current that the primary-side controller sees.
CTR variation directly affects loop gain. If your design assumes 50% CTR but production parts range from 20% to 100%, your loop gain varies by 5:1. That can destabilize the control loop, reduce phase margin, or cause the output to drift outside regulation limits.
According to my observation, power-supply designs that work on the bench but fail in production often have insufficient margin against CTR variation, LED aging, or temperature drift. The fix is to design the feedback network around the guaranteed minimum CTR and verify stability with worst-case CTR at temperature extremes.
AC Mains Detection
The optocoupler can detect the presence or state of an AC signal while providing isolation. The LED is driven through a current-limiting resistor and rectifier (or during one half-cycle), and the phototransistor output indicates AC presence.
A detail worth noting: this topology provides isolation, but it does not by itself make the circuit safe. You still need proper current limiting, component voltage ratings, fusing, spacing, and insulation to meet safety requirements. The optocoupler is one barrier in a defense-in-depth approach, not a substitute for proper mains isolation design.
When You Can Substitute 4N25 for 4N35 (and When You Can't)
Pin compatibility is necessary but not sufficient.
Before substituting, compare:
- Guaranteed minimum CTR at your LED current and V_CE
- Maximum collector current rating
- V_CE(sat) at your required collector current
- Isolation voltage and safety approvals
- Package and pinout (some manufacturers offer different lead spacings or surface-mount variants)
- Temperature range and derating curves
If the replacement part's guaranteed minimum CTR is equal to or higher than the original, and all other ratings meet or exceed the circuit's requirements, substitution is generally safe. If the replacement has lower guaranteed CTR, you need to verify that the circuit still has sufficient margin for the reduced collector current.
From my project experience, the safest approach is to design the circuit to tolerate the lowest guaranteed CTR you might encounter across all acceptable part numbers and manufacturers. That way, you're not locked into a single source or forced into a redesign when a part goes obsolete.
Common Problems and What They Actually Mean
Output won't pull LOW: Insufficient collector current. Check LED current, guaranteed CTR, and pull-up resistance. A weaker pull-up or higher LED current may restore margin, but verify that you're not exceeding LED or phototransistor dissipation limits.
Slow switching: Phototransistor is saturating too hard, or collector load capacitance is too high. Reducing LED drive or lowering the collector resistor can help, but if you need faster switching, choose a different device architecture.
Works with some parts, fails with others: Classic CTR variation issue. The circuit has insufficient margin and only works with higher-CTR samples. Redesign around guaranteed minimum CTR.
Temperature sensitivity: CTR, V_CE(sat), and LED forward voltage all drift with temperature. If your circuit works at 25°C but fails at −40°C or +85°C, you need more margin or tighter component selection.
When to Choose Something Else
The 4N25 and 4N35 are solid choices for isolated control signals, status indication, and low-speed feedback, but they're not universal solutions.
If your application requires:
- Data rates above ~100 kbps: Use a high-speed optocoupler (6N137, HCPL-2630) or digital isolator.
- Predictable propagation delay: Digital isolators offer tighter timing specs and lower jitter.
- High common-mode transient immunity: Modern digital isolators often specify >50 kV/μs CMTI; phototransistor optocouplers are slower to respond to fast transients.
- Low power consumption: Digital isolators typically draw less supply current than LED-based optocouplers.
The 4N25 and 4N35 remain relevant for applications where cost, simplicity, and availability matter more than speed or timing precision.
Final Thoughts
The 4N25 and 4N35 are functionally similar, pin-compatible, and often interchangeable—but only if you verify the guaranteed electrical characteristics in the specific datasheets and confirm that your circuit has margin to tolerate the worst-case specs.
The most useful distinction is guaranteed minimum CTR, which determines how much collector current you can rely on at your chosen LED drive level. Everything else—output LOW voltage, switching speed, temperature drift—flows from that.
Design for the guaranteed minimum, not the typical. Test across the full temperature range. Budget for LED aging. And when you need faster or more predictable isolation, don't try to squeeze performance out of a general-purpose phototransistor—pick the right tool for the job.
Frequently Asked Questions
Can I directly replace a 4N25 with a 4N35 in an existing circuit?
Pin compatibility makes it physically possible, but electrical interchangeability depends on whether the 4N35's guaranteed CTR, voltage, current, and isolation specs meet or exceed the original circuit requirements. Compare datasheets and verify operation across your full operating range before committing to production.
Why does my circuit work on the bench but fail in production?
Most likely, the prototype used parts with higher-than-guaranteed CTR, and production units are within spec but at the lower end of the CTR range. This is why you design around guaranteed minimums, not typical values or individual samples.
How much LED current do I actually need?
Enough to produce the required collector current through the guaranteed minimum CTR, without exceeding LED or phototransistor dissipation limits. If you're running the LED at 1 mA and getting unreliable switching, increasing to 5–10 mA often restores margin. But if you need more than 20 mA continuously, consider whether a different isolation topology would be more efficient.
What's the real isolation voltage I can count on?
The isolation test voltage is a one-second dielectric withstand rating, not a continuous working voltage. For safety-critical designs, consult the relevant standard (UL, IEC, VDE) and verify that your PCB layout provides adequate creepage and clearance for your working voltage and pollution degree.
When should I use a digital isolator instead?
When you need faster switching, tighter timing, lower power, or higher common-mode transient immunity than a phototransistor optocoupler can provide. Digital isolators also eliminate CTR as a design variable, which simplifies production qualification and long-term reliability.
