TPS23756: TSPS23756 Output unstable and low efficiency

Vincent Li00

Intellectual 2035 points

Part Number: TPS23756
Other Parts Discussed in Thread: , UCC2897A, TLV431A

Hi Guys

Our customer use TPS23756 for PoE application. Please find the attachment which is the schematic. So far we have some issue now.

1.Load regulation test: After 2.3A, the output voltage is pulled down to about 1~2V. We are currently suspicious of CS circuit problems, I would like to ask about the selection calculation method of this cs peripheral.

2. When the load is about 1-2A. 12V input and 5V output. The efficiency is only 70%. This is a relative big gap with datasheet.

Regardless of DC12V, DC24V or PSE power supply, the phenomenon is consistent. The two figures are DC12V power supply, the output voltage is normal, two gate signal waveforms and DC12V power supply, and the output voltage is not normal, the two gate signal waveforms.

POE-POWER_V1.0-20180615.pdf

this is normal waveform:

this is abnormal waveform:

This case is very urgent. Your feedback are appreciated.

Thanks

-Vincent

over 7 years ago

0 Thomas Amlee over 7 years ago

TI__Genius 9230 points

Hello Vincent,

The customer is using a flyback transformer in an active clamp forward design.

The TPS23756EVM User's Guide recommends Coilcraft JA4249-CL for T1/T2 which can be ordered from the Coilcraft website.

Thank,
Tom A.

0 Vincent Li00 over 7 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Sorry. I don't really understand the flyback you mentioned. I think they do use the forward transformer in this application. Right? There just a incorrect label which should be forward transformer.

0 Vincent Li00 over 7 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Do you think this issue is related to the CS circuit design? So how to design this section circuit? resistor or cap value?

0 Vincent Li00 over 7 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Tom
I have double check with customer. They do use a forward transformer in an active clamp forward design not flyback transformer. So could you provide some idea: what waveform will I have to check? I can co-work with customer to capture these waveforms.

0 Thomas Amlee over 7 years ago in reply to Vincent Li00

TI__Genius 9230 points

Vincent,

For probing, you can:
1. Measure the switch node (drain of Q4)
2. Add a current loop before T1 (on pin 1/2) & measure current during abnormal behavior
3. Measure Vcs when the abnormal behavior occurs

Additionally, could you provide the datasheet on the transformer? The current sense circuit is dependent on the specs of the transformer.

I caution against straying away from the EVM design since the transformer is appropriate spec & the current sense transformer dissipates much less heat than a current sense resistor.

Thanks,
Tom A.

0 Thomas Amlee over 7 years ago in reply to Thomas Amlee

TI__Genius 9230 points

Thank you for the datasheet. The transformer specs are the same as the recommended Coilcraft.

Using a 200mohm current sense resistor, the customer is hitting Vcs(max) (0.55V) at 2.75A. This will terminate the duty cycle prematurely, so the converter is unable regulate past 2-3A load. Additionally you are dissipating 1.5W of power across this resistor at this peak current (2.75A).

I suggest that you stay with the current sense transformer for your wide Vin application keep power consumption down. Otherwise, I suggest reducing your Rcs resistor considerably.

Best,
Tom

0 Thomas Amlee over 7 years ago in reply to Thomas Amlee

TI__Genius 9230 points

Regarding the calculation method for the CS resistor, see page 31 of www.ti.com/.../slua535.pdf

Although, make sure to subsititute the Vcs(max) value of the TPS23756 instead of the Vcs applicable to the UCC2897A.

0 Vincent Li00 over 7 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Hi Thomas

Thanks for your feedback. I will double check the two points:

1.check the Rcs value and choose a proper value;

2.Check if we use TPS23756 not UCC2897A.

Thanks

-Vincent

0 Vincent Li00 over 7 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Hi Thomas

We have changed the sense resistor and test the load regulation again. But we found it still failed when the load go up over 2.4A.

This is our new resistor value: Change R10 to 0.067m ohm. when the output load is 3.5A. The input current will 3.5A*0.8=2.8A and plus 0.5* current ripple. I think it will cover the 0.55V voltage threshod. But the test still failed.

we test the CS pin:

1. 0A load CS pin:

2. 2A load CS pin :

3. 3A load CS pin:

So how to optimize our circuit to realize the normal operation?

Your feedback are appreciated.

-Vincent

0 Thomas Amlee over 7 years ago in reply to Vincent Li00

TI__Genius 9230 points

Hi Vincent,

1. The voltage reference on the secondary side (D27) should be TLV431A (the low voltage variant)***
2. How come you change the R9 resistor from 1k to 100ohm?

Regards,
Tom A.

0 Vincent Li00 over 6 years ago in reply to Thomas Amlee

TI__Intellectual 2035 points

Hi Thomas
We checked the test way again with customer. We found the current limit has gone up to 5A when we changed the CS resistor value now. R9 we keep it 1k ohm. Rcs we changed to 50m ohm. It is just incorrect test way before.
Thanks Thomas
-Vincent

0 Thomas Amlee over 6 years ago in reply to Vincent Li00

TI__Genius 9230 points

Hi Vincent,

My pleasure. If the customer is satisfied & they have found resolution to their issue, could you let me know by clicking the green button?

Thank you,
Tom A.

Power management

Power management forum

TPS23756: TSPS23756 Output unstable and low efficiency