LM3409 uses an external P-channel MOSFET, inductor, freewheel diode, and sensing network for constant-current buck LED drive. It is not an integrated power module; part numbers do not establish board power. Ordinary LM3409 and LM3409-Q1 operate at 6–42 V; HV and HV-Q1 at 6–75 V. Check external devices and transient stresses too.
Confirm buck headroom between input and LED string first
Inputs include normal/minimum/maximum/transient supply, LED-string count, forward-drop variation with temperature/current, and target average current. Buck operation requires adequate input-output headroom; increasing current settings cannot maintain brightness with insufficient input. Harness drops and connectors belong to the actual string path.
HV selection raises only the controller's applicable input range, not MOSFET, diode, inductor, sensing-component, or capacitor ratings. List maximum stress at each power node, especially switching overshoot. Q1 component qualification is not complete-board automotive-system certification.
Calculate peak control separately from average LED current
LM3409 controls peaks by sensing high-side shunt voltage. With IADJ open, the manual gives a typical maximum threshold of 248 mV; external voltage/resistance adjusts it. For continuous conduction with approximately triangular ripple, IAVG≈IPEAK−ΔIPP/2.
Example: target average 1 A and inductor peak-to-peak ripple 0.3 A imply about 1.15 A peak. Starting with the typical 248 mV threshold gives RSNS≈0.248/1.15≈0.216 Ω. Include threshold error, resistor tolerance, hot conditions, and ripple changes; this does not guarantee 1 A accuracy. Use actual RMS, not peak alone, for shunt dissipation.
| Design Object | Parameters to Verify | Output |
|---|---|---|
| Sampling Network | Threshold, ripple, tolerance, and power | Peak and average current range |
| Inductor | Inductance, saturation, DC resistance, and temperature | Ripple and thermal loss |
| PFET | Voltage rating, gate, conduction, and dynamic loss | Switch Waveform and Rise |
| Diode | Reverse stress, conduction current, and heat | Free-wheeling Capability and Loss |
Constant off-time is not fixed-frequency PWM
The controller uses off-time-related control; switching frequency changes with input, output, and operating state. Calculate ROFF, COFF, and connections through the design process rather than adjusting one part to match one frequency and assuming all inputs behave identically. Minimum on-time and low duty also constrain actual operation.
Keep switching loops compact, with short high-frequency returns among input decoupling, PFET, and diode. Route CSP/CSN independently from specified shunt endpoints, away from switching nodes. Long shared sensing/high-current paths add parasitic drop to control. Keep sensitive IADJ/COFF away from high-dv/dt regions.
Analog and PWM dimming solve different problems
| Method | Primary Variation | Items Requiring Verification |
|---|---|---|
| IADJ analog dimming | Change current threshold and LED string operating current | Low current linearity, ripple, and color change |
| EN PWM dimming | Change effective conduction time | Pulse width, establishment, and average current |
| External parallel dimming | Change LED bypass and current path | External switches, transients, and reference-topology applicability |
Deep analog dimming may enter discontinuous conduction, invalidating unchanged ripple/average relationships. Short PWM pulses may end before LED current settles, so actual average current deviates from duty. Frequency selection should consider visible flicker, cameras, driver settling, and electromagnetic behavior together.
Observe commands and LED current together during flicker
With fixed strings/supply, record dimming commands, EN or IADJ, PGATE, LED current, and input voltage. Unstable commands suggest control outputs; stable commands with input sag suggest supply; abnormal switching with normal supply suggests sensing, off-time, or power parts. Low-brightness-only distortion prioritizes settling/discontinuous conduction.
Test low, medium, and high brightness, startup, rapid changes, and sustained hot operation. Optical brightness needs appropriate measurement; stable current alone does not establish illuminance, color temperature, or mixing. After temperature changes LED forward voltage, check frequency/current against defined ranges.
Open and short circuits cannot be judged by eye alone
Analyze current/node voltages for open strings, partially shorted LEDs, and loose connections under actual topology, using approved energy-limited tests. Cycle-by-cycle limiting is not complete load-fault management; external power devices and input supplies still face transients. Record actions, recovery, and repeated-retry heating; “the lamp still lights” is insufficient.
Deliver complete suffixes, strings, supply ranges, power BOMs, dimming strategies, waveforms, and temperature rise. Recheck ripple/stress after LED-count or external-PFET changes. One original-board power label cannot cover every combination.
Constant-current accuracy must cover supply and string-temperature changes
Fix dimming commands first; record average/ripple current at minimum, nominal, and maximum inputs, then compare cold/hot strings. Temperature-dependent LED drops can change frequency/ripple under constant off-time. Hitting target average at nominal room-temperature supply does not establish identical behavior everywhere.
For LED-current measurements, state probes, bandwidth, and inclusion of switching ripple. Multimeter averages support DC checks but miss short pulses, overshoot, and insufficient low-brightness settling. Camera/visual banding calls for optical/current waveforms to distinguish dimming frequency, uneven pulses, and image-sampling effects.
Inductor saturation changes rise slope/peaks; check highest current and hot conditions, not nominal room-temperature inductance alone. Measure PFET gate drive relative to source, confirming limits at maximum input. Address ringing through layout/parasitics rather than blindly increasing capacitance or gate resistance.
For multiple strings, record each channel and simultaneous supply load, thermal coupling, and dimming synchronization. Single-channel success does not establish concurrent operation or color mixing. For wide dimming ranges, compare analog, PWM, or combined methods, retaining usable ranges and waveform evidence for each.
Bind production current settings, strings, and cooling. Different LED forward drops or power classes require new headroom, power-device, and thermal checks even with identical connectors.
Does halving the shunt resistance double brightness?
Peak settings may rise, but average current, saturation, cooling, and LED efficacy are not necessarily linear and may exceed limits. Review the entire power stage/string anew. For thermal paths and measurement points, consider PCB thermal estimation method; accept current and optical tasks separately.
Reconfirm control behavior after LED-connection changes
String wire length, connectors, and distributed capacitance affect startup/dimming transients. Establish a fixed short-wire baseline, then attach actual harnesses and compare LED current/switching nodes. Long-wire or fast-dimming-only faults call for parasitic/loop checks before current-setting changes.
After changing string count, recheck minimum-input headroom, off-time, inductor ripple, and diode stress. Retain original-string electrical/thermal states for comparable inputs. Final outputs should state actual loads and usable dimming ranges; controller voltage ratings do not summarize board output capability.
Source material and applicable scope
- Texas Instruments, “LM3409, -Q1, LM3409HV, -Q1 P-FET Buck Controller for High-Power LED Drivers”; version Rev. L, 2016-06; checked 2026-10-17; relevant sections: Peak Current Control, Controlled Off-Time, Dimming, Design Guide, Layout. Supports ordinary/HV input ranges, external power stages, and dimming methods, not whole-board power or optical guarantees.
