TL431: vout drops much @ ~20mA Iout

Hi  Andrew, 

Thanks for your reply. 

But I'm confused on your answer.  Based on the app notes you mentioned, it indicate Cload and Ika range for stability,   but as you see from the sch, there is almost no Cload, or you mean Cgs of FET which is equal to Cload?   But Cgs is determined by MOS and cannot be changed. 

In datasheet, there is also similar reference design as below and no Cload.  the difference is that the Anode is connected to a floating GND in customer's application.    Also, the 1nf cap which seems can solve the issue  is added between C and B of  the BJT which is not Cload .

Pleaese kindly add your comments. Thanks.

BRs

Given 

Hi Given,

For AC analysis purposes, the 1nF acts as a COUT. ("Vin" and "S" are shorted). I suspect there may be some other stray capacitance that the Cathode is seeing that makes it unstable without the 1nF. Could you check with the customer if anything's causing additional capacitance?

Thanks,

Andrew

Hi Given

It's also possible that this issue has been fixed in the TL431LI. Could you post the competitor's part, so I can do a side by side comparison?

Thanks,

Andrew

Hi Andrew,

I check with customer and  there is one more 4.7uf on Vout ( between G and S).  

competitor is TL431 from NXP.  

. Thanks. 

BRs

Given 

Hi Given,

Could you link the DS? I don't think NXP has a TL431 part.

Thanks,

Andrew

Hi Andrew, 

i search the website and find there is below 

TL431 datasheet(2/24 Pages) NXP | DC/DC SWITCHING REGULATORS, LOW-VOLTAGE LDOS, MID-VOLTAGE LDOS (alldatasheet.com) 

BRs

Given 

Hi Andrew, 

customer use TL431BQDBZR(&5). Thanks. 

BRs

Given 

Hi Given,

I don't see anything too different with the devices:

Could you have them check circuit with the TL431LI?

https://www.ti.com/product/TL431LI?keyMatch=TL431LI&tisearch=search-everything&usecase=GPN

Thanks,

Andrew

Hi Andrew, 

So based on your judgement, it's stability issue, right?  for the TL431LI, I think the main difference is the stability curve is different. TL431LI can support larger stable region. But I think based on the curve, 4.7uf Cload is stable. 

Do you think common mode noise may be a suspect?  Thanks. 

BRs

Given 

Hi Given,

The effective capacitance of the Capacitor could be lowered by applying voltage. Does anything happen if they remove the capacitor or reduce it's value?

Yeah the TL431LI is an improved version that fixes some issues with the original one. The lower Iref and IIdev values enable designers to achieve higher system accuracy and lower leakage current. I think it's worth a shot to try.

Thanks,

Andrew

Hi Andrew, 

if no  1nf mentioned in before email ( in red) , Vout has oscillation. But adding 1nf will solve the issue.  But I think 1nf is much small compared with 4.7uf Cout. 

Thanks. 

BRs

Given 

Hi Andrew, 

4.7uf cannot be removed in actual application. so we try to use 2*4.7uf , but still have the issue. 

Got your point. we'll try to see if we can try TL431LI. Thanks.  

BRs

Given