tidu789c Having startup issues with UCC2877 flyback

Hello Scott,

There are several issues to address here, and I’d like to start with the “gradual increase” of input voltage versus “hot plug” into 240Vac.

This TIDA-00355 design is fully capable of being hot-plugged into a 240Vac line without damage, provided that it is working in the first place. TI generally does not provide reference-design boards directly, but provides the schematic and pcb design files for customers to be able to recreate them. The user guide (TIDU789C) recommends gradual increase of input voltage because a customer-constructed board may have inadvertent bugs in it that a slow increase in voltage may reveal a problem before damage is done that a sudden hot-plug into high voltage could not avoid. This is generally a good practice for any design that has not already been tested and working.

You have received a board from a client and it is not clear what issues your client had with it, or why the start-up resistor values are lower. (Perhaps the values were cut roughly in half in anticipation of normal running at 120V instead of 240Vac.) In your posting you made two statements: 1. that pin 2 of UCC28722 was “shorted” (it is unclear to what it is shorted), and 2. that on a second board, VCC pin 2 was clamped to 13V. It is possible that both boards were already damaged when you obtained them.

The voltage at VCC (pin 2) must rise to 21V before the UCC28722 will start switching. If pin 2 is shorted to GND, the IC or the VCC cap is damaged and must be replaced with good parts. On the 2nd board, being clamped to 13V indicates that the D21 diode is not blocking and the start current is running though Zener-clamp D19. This 15-V Zener is likely operating around 13V with the micro-amp level of current available. So, on the second board, I recommend to check D21 as well as U8 and C52 for damage, and replace as necessary.

I recommend changing R49, R53 and R57 back to 3.3Meg each as the original design calls for. Before startup, the UCC28722 only draws ~1uA at VCC, so even 31Vdc on B+ can provide that much current through ~10Meg by the time VCC got up to 21V. Because of tolerances, higher voltage is advised, but the point is that 240Vac or even 120Vac is not necessary to start the Aux supply controller. Section 5.3 of the User Guide indicates that the Aux supply should regulate with a minimum of 100V on the bulk rail. This corresponds to the peak of ~70Vac. It will attempt to start up with input voltages below 70Vac, but may not stay running or may not regulate properly. All this pre-supposes, of course, that there are no problems with the board(s) in the first place.

With 3x1.69Meg at R49, R53 and R57, the current to VCC when the system is running comes from a PFC output at ~400V. Istart = (400V-13V)/5.07M = 76uA. This current is much less than the 2mA that IC alone draws (even neglecting Q8 base current) when the system is charging a battery, but in stand-by condition, the Aux load is turned off, and the UCC28722 operates most of the time in a Wait state. In this state, Iwait at VCC is typically 95uA. If the IC requires less than typical wait-state current, or more likely, if there is a leakage path around the resistors (due to pcb surface contamination with flux or moisture) the current into VCC could exceed what the IC requires and VCC will rise above the absolute maximum rating of the device and damage it. This is not the situation where a Zener clamp on VCC is recommended in the datasheet, but it is another situation where such a clamp might be a good idea, although higher resistor values and a clean surface would avoid this.

So, I recommend that the Auxiliary power stage of both boards be thoroughly inspected and tested at low input voltage so that you are satisfied that it works correctly. I recommend to add a 68-ohm 5-W resistive load across C48 (6V5 output) to do this, to avoid having to deal with any of the other system outputs. Once the Aux supply works, the rest of the board can be powered up, hopefully without requiring additional debug. Conversely, you can remove R49 and Q8, and provide external sources for 6V5 and 13V to power up and test the rest of the system (temporarily) without a working Aux stage.

It may not be clear, but this TIDA design requires a battery or battery-emulator load to get the PFC and LLC stages to power up. Wen plugged in, the Aux stage operates all the time. It will be in stand-by mode when no battery or battery-load is applied, and bias power to the PFC and LLC stages is shut off to keep total stand-by power low. In order for the PFC and LLC to get bias power (VCC1 and VCC2), the signal LS must be high enough to turn on Q9, which then drives switches for the other bias voltages.

The signal LS is derived from the sense-resistor string on the main battery charging output. The User Guide indicates that the charger is applicable for 20V-42V battery voltages, so an external voltage of at least 20V must be applied to J3 output in order to activate the PFC and LLC stages. Do not use a bench supply for this purpose because most power supplies are not capable of sinking current (absorbing power). A battery-emulator or a real battery with suitable safeguards should be used.

Finally, you indicate that the eventual goal is to operate at 120Vac. If you intend to be able to operate at 120Vac in addition to 240Vac, then the PFC stage will need to be redesigned to handle the higher current at the lower voltage, but the LLC stage can stay the same. If you intend to be able to operate at 120Vac instead of 240Vac, then the PFC stage will need to be redesigned to handle the higher current at the lower voltage but stay at 400V output and LLC need not change, OR change the PFC output down to 200V and also change LLC stage for 200V bulk voltage. This can improve the low-line efficiency. In either case, I recommend to follow the design procedures in the respective datasheets rather than that in the TIDA-0355 User Guide, because the guide design descriptions may not have included every possible consideration that the datasheets may mention.

Good luck with your debug. Please indicate whether this answers your questions and concerns.

Regards,
Ulrich

Ulrich,

Thank you for the detailed response.  I shall inspect the board as per your recommendations.  I do appreciate the amount of effort that you put into this response as I was an applicaitons engineer and understand how demanding it can be time wise.  Hoping for a happy ending so I can write a piece on this circuit at my Planet Analog blog.

Scott Deuty