TINA/Spice/ATL431: SPICE model is incorrect?

Hello,

Sorry.
I got mistake with setting of the voltage generator.
I will attach the new result.

Regards,
Naoki Aoyama

Hi Naoki,

From your example it seems the ATL431 SPICE model is not working as you expected. I will ask my internal SPICE team to see what the cause of this problem could be.

Can you attach the TINA spice file?

-Marcoo

Hello Marcoo,

Thank you for your reply.

It is clear the issue that the signal start before VCC start.

It was my mistake in setting the voltage generator.

But, the oscillation issue still remain.

I attach the TINA file.

webench_design_3885918_52.tsc

Is the Rs as 4.8kohm too high?

Best Regards,

Naoki Aoyama

Hi Naoki,

I currently do not experience this issue on my TINA-TI spice. I opened the file and just ran it but did not see the signal before Vin problem. I did see the oscillation that should not exist at those current levels and capacitor values. 

My SPICE team is looking into the issue.

-Marcoo

Hi Marcoo,

The signal before Vin problem is my mistake.
I'll wait for your findings.

By the way, the result of WEBENCH designer is not reasonable.
The configuration of the customer as follows.
VIN = 20V to 24V
VOUT = 3.3V
IOUT = 1mA
Ta = 30 degree

According to the result, Rs= 205 ohm. It flows the high current at IKA.
There is no oscillation.
It can not get the merit of the ATL431 as the low IKA.

Regards,
Naoki Aoyama

Hi Naoki,

I agree with you that webench did not give the correct result. For your situation, I would calculate the value of R1 based on the worst case conditions.

We will used the input voltage of 20V because that is when the device will be the most current starved.

Lets assume R1 + R2 = 100k = R12. So the current through the resistors, I(Rs), = Vka / R12 = 3.3/100k = 33 uA

We want to use a value close to the minimum Ika = 40 uA

ILoad = 1mA If this value is not the worst case load value, use the worst case load value.

Rs    = (Input - Vka) / (Ika + I(R12) + ILoad)

= (20 - 3.3) /  (40 uA + 33 uA + 1mA)

~= 15.5 k Ohms

With this resistor value you should be able to provide a load of 1mA while the device + its biasing network consume only 73 uA which is an improvement over the TL431 where it would be another 1mA.

Looking at the stability charts, I would use figure 14 as it is more conservative. I would use values of Cload less than 10 nF or greater than 300 nF. I would be comfortable with a value of 1uF.

-Marcoo

Hello Marcoo,

Thank you for your advice.

I simulated the circuit by TINA-TI.

It remains oscillation issue.

6471.webench_design_3885918_52.tsc

This simulation does not seem reproduce correctly.

Best Regards,

Naoki Aoyama

Hi Marcoo,

Thank you for your reply.
I see. I will wait for the next release.

Thanks,
Naoki Aoyama