UCC256304: Troubleshooting Initial Turn-on Failure of New Design

Hi Tim,

Thanks for reaching out. Can you share your schematic with me? You can email it to me if you do not want to post it on this public forum: ben.lough@ti.com

Best Regards,

Ben Lough

Hi Tim,

If I could give some pointers for now, it looks like the controller is restarting every 1s. This can be overcurrent, overvoltage or overtemperature protection. Try grounding the BW pin, reduce your ISNS resistor, connect a bypass cap between ISNS and gnd (100pF or so) and see if the converter can start up.

Best Regards,
Ben Lough

Hello Ben,

Schematic and other design information sent as attachments (3 files) to your email address.

Regards,

-Tim

Thanks Ben,

I will respond with results when available.
Thank you for the suggestions.

Best Regards,

-Tim

Results of grounding BW pin:

BW PIN NOT GROUNDED

BW PIN GROUNDED:

Controller continues to cycle as though a faults occurs.

At the timescale in the traces,

VCC unchanged.

RVCC unchanged

LO gate unchanged.

HI gate moves upwards, as though bootstrap circuit now functional.

Preliminary tests show that grounding the BW pin and adding capacitance between the ISNS pin and the input ground (0.1uF) allows the converter to run, but not correctly. Testing is continuing tomorrow.

With BW grounded, and 470pF bypassing the ISNS line, I measure the following on ISNS:

Each event looks like the following:

The first event looks like the following (with zoom window):

Since the OCP1 threshold is 5V during soft start, I believe the controller faults.

Hello Ben,

I have not yet been able to get the design up and running. I have been recalculating the design values, and modifying prototype board #2 for a new set of tests. Parts procurement has taken up a bit of time. I likely will not have further results until next week.

Please keep this thread open, and look for more test results shortly.

-Tim

Hi Tim,

Ok, no problem. Let me know how the test results look and if I can be of any assistance in the meantime while you wait for components.

Best Regards,
Ben Lough

With BW connected to (input) ground, the MOSFETs do not switch continuously. 1000pF bypassing ISNS. If I increase the load, the gate signal pulse groups move closer together:

Hi Tim,

It looks like you might be in burst mode. Could you take a look at the FB pin voltage and let me know if it is oscillating? is your output voltage able to reach its regulation set point?

Best Regards,
Ben Lough

Yes, FB is oscillating. Output voltage not well regulated (load regulation is poor):

Hi Tim,

It might be a compensation issue. Try depopulating the resistor between RVCC and LL/SS for now. This will move the burst threshold to its minimum and make debug a little easier without having to worry about the controller going into burst mode. Do you see the same oscillation at the TLVH431 kathode? Is the output voltage moving above and below the regulation set point?

Best Regards,
Ben Lough

Hello Ben,

There is an assembly error on all three boards. R13 (resistor between RVCC and LL/SS) is missing. So all waveforms to date have been with this case.

The TLVH431 cathode waveform is shown following:

Sorry, I used the input ground as reference here. I will re-do the measurement of the TLVH431 cathode with the output ground as reference. Stand by...

TLVH431 cathode signal, using output ground as reference. "Busy" areas occur when low-side MOSFET is switching. BW signal grounded, resistor between RVCC and LL/SS open:

Vout and TLVH431 cathode voltages shown. Upper tip of (main part of) Vout waveform is exactly the desired output voltage:

Hi Tim,

I would try the following:
1. Measure the voltage across D6 and make sure its a stable 11V. If its oscillating try reducing R26 to ~15kΩ.
2. Connect a 1nF capacitor between TP28 and TP29. This will roll off your gain faster and should help improve the phase margin
3. Increasing the value of Ccomp will help separate the pole and zero placement for the error amplifier and should help give better phase boost. perhaps ~82nF for C24 and ~56kΩ for R29. These values are approximate so you may need to deviate some. If you are able to take a loop measurement, that would help decide the compensation values with more precision.

Best Regards,
Ben Lough

1. D6 voltage is stable.
2. 1nF across TP28 (U4 TLVH431 cathode) and TP29 (U4 TLVH431 REF) had no discernible effect.

Trying Ccomp/Rcomp changes now.

Hi Tim,

Did you get a chance to try adjustments to the compensation?

Best Regards,
Ben Lough

Ben,

After a delay, I'm continuing with troubleshooting. The present state of the design has the following changes:

• The resistor between RVCC and LL/SS (my R13) is OPEN.
• I have a 1000 pF capacitor on the ISNS signal to input ground.
• The BW divider is upper = 24.9K (R16), lower = 5.45K (R19).
• I rewound the resonant inductor (L1) so that it measured very near the desired 3.9uH value (measured 4.0uH).

I will have test results (the converter does run) Monday.

Thanks for staying with me on this.

Conditions as described in last post. Turn-on waveforms:

Oscilloscope probes have rather long ground clips used at this stage. I am inclined to ignore the HF "fuzz" on each signal once switching begins.

Does this look reasonable for turn on?

Hi Tim,

Thanks for the waveforms. This looks reasonable to me. The dip you see in RVCC when the LS gate first goes high is expected. Looks like the converter continuously switches so that's good. Are you able to get the output voltage to regulate or do you still see the oscillatory behavior at steady state?

Best Regards,
Ben Lough

Hello Ben,

For the case above, I tied input ground to output ground. I did this to allow me to have all of the scope probes use the same reference. When I did this, switching is continuous, but the input current exceeds 1.3 A (input power about 65W), and maintains that input current at very light loads, which is unexpected. But the output voltage is smooth, and there is no oscillatory behavior. The output voltage is approximately that desired (within about 2V). With the grounds shorted, the MOSFETs and associated heatsinks get quite hot.

If I do not ties the grounds together, switching is not continuous, and the output voltage varies greatly in a triangle-wave-like way, at a voltage much less than programmed:

Hi Tim,

Could you take a look at the resonant current using your scope with the grounds shorted and without? I am wondering if the transformer is saturating.

Best Regards,
Ben Lough

Hi Ben,

I do not have a calibrated current probe to use, but I can get a look at the current wave shape. I will make that measurement on Monday.

Ben,

I now also have the TI EVM (UCC25630-1EVM-291) to use. Do you have access to some of the design details associated with the EVM? It would be useful to have the transformer turns ratios for example. The Renco data sheet does not specify the ratios used - the transformer is a custom part, as one would expect. ideally, TI would release the Excel design calculator (or otherwise document the design choices made) for the EVM so others could evaluate the design.

Inductor current waveform, grounds not shorted:

Inductor current waveform, grounds shorted:

Hi Tim,

thanks for the waveforms and apologies for my delayed response. These waveforms look normal for below resonance operation. When the grounds are not shorted, could you look at the VCR pin peak to peak voltage? Do you see VCR moving between 0V and 6V?

Best Regards,
Ben Lough

Yes, When I reduce the load resistance sufficiently to cause the converter to continuously switch, the VCR voltage looks sinusoidal, 6V p-p with a 0V baseline:

State of design:

Generally functional.

1. Does not run continuously unless input power is about 50W. I would like burst mode to occur at about 20W out (730 mA).
2. The output voltage does not reach the desired value (27.5V) and varies in a random-like way

Hi Tim,

Could you share what are your current values for C13 and C16?

Best Regards,
Ben Lough

C13 is 330pF and C16 is 6800pF

Hi Tim,

Try reducing C13 to 100pF or so. This should increase the contribution of the frequency compensation ramp current.

Best Regards,
Ben Lough