UCD9248 Vin off / Vin on hysteresis does not seem to work.

"When I turn on the input power supply, at this 6.5V the UCD9248 does not start up!"..... What I meant is that the UCD9248 does not start the power conversion, the UCD9248 is of course still running.

Hi Ben,

I cannot reproduce this problem on EVM, probably because I can only manually adjust Vin.

You can try to use Vin Under Voltage fault response to Shut Down Immediately and Restart Continuously, see if it solves the problem.

Thanks,

Zhiyuan   

Hi Zhiyuan,

Thank you for your reply.

In the figure you can see that when the Vout is shutdown due to VinOFF setting of 5.5V, the input supply is jumping up about ~0.7V. This is because the load is switched off and the cable losses are not there anymore.

You can simulate the cable losses by adding a low resistance resistor (~0.25 ohm, depending upon EVM load) between the lab-power-supply and the EVM. When the EVM then switches of the load, the supply will also jump a little bit.

When I don't have the 'jump' in the Vin supply ( no cable losses), then it works also fine here. But you always have some cable losses in your end application.

The VIN_UNDER_VOLTAGE_FAULT condition has the same issue.

For now I'm doing the assumption that:

The UCD9248 seems to 'forget' that there was a 'Low Vin' condition, due to the probably fast jump in the Vin, when switching off the load.

Can you confirm this and have a possible solution? Or do you see the same problem with the above EVM setup, with the added low resistance resistor in the power supply?

Thanks in advance,

Ben

 

Hi Ben,

The state machine takes two cycles to disable and then re-enable the output. If Vin jumps back too fast it may be an issue. In the case of cable resistance, you can try increasing input capacitance to slow down the Vin jump. Let me know if it works.

Thanks,

Zhiyuan

Hi Zhiyuan,

1) "The state machine takes two cycles to disable and then re-enable the output".

I don't think this is a problem, because the voltage jumps to 6.5V and the VinON is 8V. So the output does not have to go off and on fast.

2) "If Vin jumps back too fast it may be an issue. In the case of cable resistance, you can try increasing input capacitance to slow down the Vin jump."

As seen in the schematic (below) of the UCD9248, it has already an RC filter to the input pin of the UCD9248. I did increase the 100nF to 3uF but the problem still persists....

You could quite easy reproduce the problem with the lab-power-supply with an added low-resistance resistor (~0.25mOhm) in the power cable to the EVM.

Of course the EVM should have some load to switch off. Did you see the same effect?

I still think the UCD9248 seems to 'forget' that there was a LowVin and if then the VinON (8V) is reached, as seen in the graph earlier, the output does not switch on. But there is no reason not to do this.

Kind regards,

Ben

Ben,

Vin UV fault shut down with Restart Continuously should work. I guess you didn't set Restart Continuously last time. Also, set Vin_Off threshold lower than Vin UV threshold, such that Vin UV fault will trigger first.

I was referring to an internal state machine, which cannot properly address the scenario that Vin kicks back immediately after Vin_Off shut down.  

Thanks,

Zhiyuan

 

Hi Zhiyuan,

I did the settings as you suggested.

VIN_UNDER_VOLTAGE_FAULT = 6.5V

VinOFF = 5.5V

VinON = 10V

In the graph below you can see that this does not work.

I checked the setting for the "restart continuously", as seen below:

Are there any other possibilities?

Thanks in advance,

Kind regards,

Ben

Hi Ben,

I put 10 ohm on my EVM Vin port and cannot replicate the problem. Both Vin_Off and Vin_UV work fine. 

Vin_Off:

Vin_UV:

Is there any turn-on dependency that may prevent the rail from turning on? I can take a look at your project file.

Thanks,

Zhiyuan 

Hi Zhiyuan,

Ok, that is nice and as it should!

Attached my project file, I hope you can find something.

There are two UCD's used. I use UCD with address 100 as a master. If that one has started up, it enables the second one with address 101, through GPIO.

Kind regards,

Ben

Hi Zhiyuan,

Thanks for the feedback!

I don't have a PMBUS host if there is no power. So I'll have to go for the UCD9248E version option.

I have some questions about this UCD9248E version:

1)

In the datasheet there is a difference in the UCD9248 and UCD9248E.

UCD9248E pinning table:

UCD9248 pinning table:

Is this an datasheet error? I have an datasheet of the UCD9248E through a local representative, so it could be an old version. Could you post the datasheet of the UCD9248E?

2)

In the UCD9248E datasheet there is mentioned in the pinning table:

The current UCD9248 does not have a powerPad and this is not mentioned. Is this a datasheet error in the UCD9248E?

3)

Do you know if the UCD9248E version will become available for mass market and thus on the website?

Thanks for the support!

Kind regards,

Ben