SEA05 Flyback Converter Fails to Enter CC Mode; Output Voltage Rises with Load

[YAJUSH]

 Hi,

I am working on an offline Flyback converter utilizing a UCC28C50 primary side PWM controller. Originally, the converter was configured for Constant Voltage (CV) mode using a standard TL431 and optocoupler circuit, which operated correctly.

To convert this design into a Constant Current / Constant Voltage (CC/CV) supply, I replaced the TL431 with an STMicroelectronics SEA05 advanced controller on the secondary side. My implementation strictly follows the secondary-side configuration from STMicroelectronics' application note "10W offline LED driver based on VIPER27" (DN0002/AN4376).

10W offline LED driver based on VIPER27 (DN0002)

 

 

Circuit Parameters and Component Values: Target Output Specifications: V_OUT = 15 V, I_OUT = 3 A (Test Target) Current Sense Resistor (R_SENSE / R16, R22): 18 mΩ

SEA05 Pin Configuration Components: R₁₈  = 3.75 kΩ, R₂₀ = 10 Ω,  R₅ = 10 kΩ, C₈ = 100 nF, C₁₀ = 1 nF, R₁₅ = 120 kΩ, R₇ = 10 kΩ,  R₆ = 51 kΩ

The Problem: According to the SEA05's internal 50 mV current sense threshold (V_SENSE), the converter should transition into Constant Current (CC) regulation at 2.77 A (calculated as 50 mV divided by 18 mΩ). Nevertheless, the converter operates exclusively in CV mode. Rather than stabilizing or limiting the voltage when load exceeds 2.77 A, the output voltage rises progressively under elevated load conditions.

 

Output Current (I_OUT)	Output Voltage (V_OUT)	Operating Mode
0 A	15.00 V	CV
2 A	15.00 V	CV
3 A	15.23 V	CV (Should be CC)
4 A	15.30 V	CV (Should be CC)

 

Inquiry: What is preventing the current loop from assuming control of the optocoupler node, and what accounts for the substantial increase in V_OUT under high-load operating conditions?

[Peter BENSCH]

The most likely issue is that the SEA05 current loop is not actually able to dominate the optocoupler node. Please check the shunt polarity, Kelvin grounding of SEA05 GND, and whether the current loop can sink enough optocoupler LED current against the CV loop.

The rising VOUT​​ at high load usually indicates the primary-side loop is hitting a limit before true CC regulation takes over, so the supply is entering a boundary condition rather than proper current limiting.

• Kelvin GND: The datasheet emphasizes that GND should be placed very close to the converter output terminal. If the SEA05 GND is raised by the load current as well, the reference level shifts, and the current measurement or the control comparison becomes inaccurate or ineffective.
• CC vs CV: The two opamps are internally OR-gated. If the CV branch is already pulling the optocoupler so hard that the current branch cannot counteract it with its transconductance current, the converter remains in CV. This is especially critical when the primary side is designed such that the optocoupler has hardly any margin left for CC regulation.
• Voltage rise from 15V to 15.23V or 15.30V at 3A to 4A: a strong indication that the primary side is going into some kind of current-limiting state before the secondary CC loop takes over: under load, the converter can no longer deliver enough energy per switching period, the optocoupler signal is determined not by CC but by the primary regulation, the control loop loses the clean CV/CC handover, and as a result the output voltage can rise instead of clamping precisely 50 mV above Rsense.

Please check:

• What resistance does the shunt have: 2 x 0.33 ohm or 18 mOhm?
• Is Ictrl connected to the negative shunt side and Isense to the positive shunt side?
• As mentioned, the Kelvin connection at the output ground
• Whether the CC branch takes effect when the CV feedback is artificially relieved
• Whether, in the event of overcurrent, additional LED drive via OUT occurs
• The primary-side limits on the UCC28C50: current limit, max duty cycle, COMP level, start/burst behaviour

The last point is especially important, because if the primary side limits earlier, the secondary CC loop can no longer counteract it.

Regards
/Peter

[YAJUSH]

Hi,

Thank you for your previous suggestions. By properly connecting the Isense and Ictrl pins along with the Kelvin ground, I am now successfully running the IC in Constant Current / Constant Voltage (CC/CV) mode.

As you mentioned earlier, i am trying to use this IC for Short Circuit Protection

However, I am experiencing  issue when attempting to use this IC for Short-Circuit Protection (SCP) at the output side.

The Problem:

When a short-circuit condition occurs and the output current exceeds our predefined threshold, the CC loop takes control. As a result, the output voltage drops significantly. This drop causes the optocoupler LED current to decrease, which inadvertently forces the converter to lock into a low-voltage, high-current state instead of shutting down or hiccuping.

• Example Behavior: For a target 15V output supply with a maximum allowed current of 2.77A, the system stabilizes at a 5.3V output while delivering a 5A load.
• Circuit Details: The secondary-side circuit is highly similar to the STMicroelectronics DN0002 design note shared earlier, utilizing an (R_{SENSE} = 18 mOhm).
• Also i am providing Vcc of SEA05 from different stable voltage source.

Request for Assistance:

Could you please suggest a secondary-side circuit modification or an add-on circuit to implement proper SCP? Specifically, I need a solution that prevents this high-current latch-up condition—either by triggering a latch-off or forcing the primary side into a safe hiccup mode when (V_{OUT}) collapses.

Thank you for your time and continued support.

Best regards,
Yajush Sihag

[Peter BENSCH]

Well, the SEA05 is regulating CC/CV, but it is not by itself a true short-circuit shutdown controller. In a short-circuit, the loop can settle at a low-voltage, high-current operating point because the optocoupler feedback still keeps the primary controller running.

For proper SCP, you need an extra fault-detection stage on the secondary side, for example a comparator that monitors Vout and, after a short blanking time, forces the optocoupler off or triggers a primary-side hiccup/latch-off.

So the practical fix is:

• keep SEA05 for CC/CV,
• add a separate SCP threshold on Vout,
• use that fault signal to shut down the feedback path or force restart.

Practically, there are three sensible approaches:

1. Secondary-side SCP with an additional comparator and shutdown signal
   probably the best approach
2. Secondary-side latch-off circuit
   in this state, the UCC28C50 must really go into a safe state = not trivial
3. Primary-side OCP/hiccup via an additional signal
   condition secondary-side fault feedback so that it triggers a defined shutdown/restart on the primary side, which is usually cleaner than trying to misuse the CC loop itself as SCP

However, the last question no longer has anything directly to do with your original question, as it concerns a new topic (SCP ≠ CC). This new topic is also rather more general in nature, so you could discuss it in general electronics forums as well.

Regards
/Peter