TEA19162 PFC low input voltage limit

janhoc · ‎10-19-2020

Hi.

I am experiencing problems with my PFC design with low input voltage and high load conditions.

The PFC boost voltage drops to below 300V under these conditions.

When I try to find the cause, I notice that the problem starts when the chip starts limiting the PFC off-time to its minimum value of 1.55us during some parts of the 100Hz semi sine wave.

What could be the reason to hit this limit?

Any suggestion to cure?

Thanks

janhoc · ‎10-28-2020

Hi Thomas.

I did as you suggested. I split the SNSBOOST divider. Under these conditions, the PFC works without problem.

The Vboost voltage stays at 390V, even with VAC at 90V and full load.

So it is definitely the LLC disturbing the PFC.

With this split scenario, I do not see the SNSBOOST voltage lifting by the LLC.

So there is some instability that only occurs when both are coupled. I suspect that this is caused by 100Hz ripple on Vboost...

Any further suggestions?

janhoc · ‎10-23-2020

Yes, indeed the LLC controller is TEA19161 and it increases the SNSBOOST voltage. Would erroneous burst mode be the only potential cause? It is unclear from the datasheet and the application note which protection could be at work. In any case, OCP on TEA19161 s disabled with diodes and OPP should only restart after 1 second...

TomasVaverka · ‎10-21-2020

Hi Jan,

Two possible causes are:

OCP

PFC OCP function can be limiting the output power.

Measurement resistor value to high
Too much disturbance on SNSCUR

A quick first trial is to reduce the resistor value by 2 by adding another resistor in parallel and observe if it solves the limiting problem.

Maximum on-time

The PFC controller regulates the power by means of on-time control. The maximum on-time can be limiting the power.

The maximum on-time is as expected, but the PFC coil impedance is too high to make sufficient power (current amplitude in this time).
The maximum on-time is shorter than expected. This can be caused by a problem in the mains voltage measurement that results in a too low (maximum) on-time. The maximum on-time is depending on the mains value:

Measuring the PFCCOMP voltage helps to understand what the situation is.

Best regards,

Tomas

janhoc · ‎10-21-2020

Hi Tomas, thanks for helping out.

I have tried to eliminate OCR by shorting SNSCUR to GND. This gives some improvement in performance, but not sufficient.

Let me share two observations.

Wen the AC voltage drops, the Vboost drops also under high load conditions.

1. When Vboost drops below 260V (at 110V AC at full load) , the LLC gets into problems, resulting in protection 1 second on-off switching cycle. I would expect the Vboost voltage to remain more or less stable, so something is limiting the PFC power conversion.

2. Even before the LLC gets into problems, so at higher AC voltages or lower loads, I see inexplicable PFC behavior, namely that it stops switching during 170 to 450uS during the rising 100Hz half-sine wave. This happens when the PFC reaches the maximum frequency. I have added some pictures of the MOSFET drain voltage. This pause in PFC operation is in my opinion the reason for lack of PFC power transfer.

TomasVaverka · ‎10-22-2020

Hi Jan,

Thanks for including some scope measurements of the behavior. This behavior does not directly look familiar.

It is indeed not related to a possible issue on OCP or on-time. It also does not look like regulation instability.

It looks like a protection is interrupting the behavior for a short time.

This is the list of protections to check:

The interruption is too short for SNSMAINS brownout/in hiccup.

It is not likely that OVP SNSBOOST is triggered because it happens when the boost voltage is lower.

Good to check the SUPIC voltage. It is probably generated by an aux winding on the LLC (which is not working well at low Vboost voltage).

Just a remark: maybe it is intended, but the boost voltage seems rather low for operation. Usually (universal mains) systems use a nominal boost voltage between 380V and 400V.

Best regards,

Tomas

janhoc · ‎10-22-2020

Vboost is indeed low. This behavior happens every AC cycle when Vboost is lowering due to lowering the AC input. I know that the Vboost is supposed to be constant over the AC input voltage range, but that is just part of the problem.

I did some more measurements and I noticed a 200mV increase on SNSBoost 100us before PFC stop, see screenshots. Do not worry about the noise on the signal, that is pour earthing of the measuring probe.

Could this 200mV surge be the culprit? If yes, what could be the cause?

TomasVaverka · ‎10-23-2020

Hi Jan,

I assume that the TEA19162 PFC controller is used in combination with the TEA19161 LLC controller (as intended).

It seems that the TEA19161 controller increases the SNSBOOST voltage for Burst Mode operation (while this was not intended; something may be wrong in the LLC).

Although the levels do not really match the intended SNSBOOST voltage levels (in diagram below); the PFC stops and the SNSBOOST is pulled high.

For a more easy debugging you can temporarily split the SNSBOOST functions to see a more independent behavior of the converters. Now the LLC controller cannot influence the PFC operation anymore:

Best regards,

Tomas

janhoc · ‎10-29-2020

I did the split of the SNSBOOST divider.

With this the PFC works as expected. The VBOOST voltage remains constant over the full AC input range, even at full load.

With the split, I also no longer see the voltage increase on SNSBOOST, the signal remains clean at both ends of the split.

So it is definitely the HBC disturbing the PFC through the SNSBOOST link that is causing the problem, but it remains a mystery what cause could be, especially since there is a strong correlation with the AC waveform. The only potential cause I see is the ripple on VBOOST that causes the LLC to start some form of protection.

Any suggestions on other experiments?

TomasVaverka · ‎10-30-2020

Hi Jan,

It is good to first check, debug and optimize the LLC converter, before connecting it to the PFC again (by SNSBOOST).

In the split-up situation, the voltage on the TEA19161 SNSBOOST pin is similar to the situation when both are connected (including ripple). The behavior at several load conditions can be checked for the LLC converter.

A load sweep can provide a good overview of LLC mode transitions:

When entering BM operation, the TEA19161 will pull-up the SNSBOOST pin. Check if the power level for BM is as expected/chosen/designed and that the SNSBOOST lifting only happens in this mode.

When an (unintended) protection is triggered, the LLC controller (TEA19161) will pull-down the SNSBOOST pin.

When mode transitions are not as expected/intended or are unstable at certain power levels, the SNSCAP divider can be checked (AN11801, Chapter 9).

Best regards,

Tomas