TPS3307: reset cause by noise

Hi User,

yes, connecting /MR directly to Vdd is ok.

What do you mean by "the filter"? You have a decoupling directly from Vdd to GND, haven't you?

Kai

Hi Kai, 

Yes, we have decoupling cap for at Vdd to GND, now the question is it might have some noise effect MR pin, whcih causing the error reset. 

Just want to confirm if connect VDD to GND is ok, and do we nee to put extra cap between MR pin and GND.

Thank you.

Hi Abhinav,

Thanks for your reply, i am confused is because there is no configuration for MR pin in typical application.
just want to confirm if extra decoupling cap is needed?

Thank you.

Hi User,

can you show us a schematic of your TPS3307 circuit? This schematic should also show, what you are exactly doing with the /MR pin.

Kai

Hi Kai, 

Please refer to below figure for schematic. 

originally it's open, but there are some noise problem causing error reset. 

So i am asking if it can be connect to VDD and if it needs extra decoupling cap.

Thank you.

Hi Abhi,

Because i am not sure if there is having noise that may effect the reset signal, so I wonder if it will be better to connect to VDD?
Thanks for your help.

Hi User,

connect the /MR pin directly to VDD. This will be the best solution.

Are you sure that the noise is not entering the TPS3307 via the sense inputs? The resistors at the "sense 3" input are rather high ohmic. Here noise could enter the TPS3307 due to stray capacitance coupling. Or could the noise directly come from the +12V line or CN1?

Why do you have R36?

Kai

User,

You can connect the /MR pin to VDD but that shouldn't be the source of the noise because it is internally pulled up.

Thanks,
Abhinav.

Hi Kai, 

Thanks for letting us know. 

not sure about this yet, do you mean that we should change the resistor divider? These may also be one possibility. I will check on that.

After confirm, the resistor is used to prevent transient from reset pin to protect CPU input pin. 

Hi Kai

 The above scope we record is when the reset happen.

CH1:V_Sense1，CH2:V_Sense2，CH3:V_Sense3，CH4:V_Reset

As we can see in the picture,V_Sense1 and V_Sense2 is ringning,V_Sense3 is drop under 1.25V for 0.285us,when reset  happen.But these conditions cannot make reset happen according to the discrimination shows in the datasheet, right?

So we check the power VDD and MR and take another picture as below.

CH1:V_MR，CH2:V_Reset，CH3:V_VDD，CH4:V_Sense2

VDD is drop when reset happen.Is the drop of VDD cause the reset happen?Howerver, we also record some pictures that VDD is drop but reset  did not happen.

Hi Abhi,

Noted and thanks for your reply.

Hi User,

if you look very carefully to the yellow line of first scope plot you will notice that the 3.3V supply voltage is superimposed by very narrow and heavy noise spikes, which seem to go below 3.0V, exactly at the moment when the /Reset output goes low. So, I think that the TPS3307 suffers from noise on the 3.3V supply line! You should improve the decoupling measures at this pin.

You can also see, that the pink curve (sense 3) is free from these narrow noise spikes. This is definitely the result of low pass filtering due to C29.

It seems to me, that your PCB does not use a solid ground plane. Am I right?

Kai

Hi Kai, 

one thing i want to confirm with you.

the parameter in following table is the pulse time causing IC to reset, is this correct? 

Or the device reset once VDD drop below 0.7VDD? 

thank you.

Hi Gary,

only the sense inputs and the /MR input are creating a reset. VDD dropping below 0.7VDD does not create a reset. You must connect one free sense input to VDD, if you want a dropping VDD produce a reset. You have done that in your schematic, right?

Gary, your application suffers from a huge amount of noise. This is your problem I think. You should intensify the decoupling measures. Please read again my last post.

Kai

Gary,

The RESET goes active when SENSE or /MR drops low. When the device has less than 0.4V at VDD, the RESET output is undefined. I also believe the issue to be caused by excessive noise. We have seen noise propagate through the VDD pin when not enough input capacitance is at VDD. Please increase your decoupling capacitors and let me know if the issue continues.

-Michael

Hi Kai,

Got it. Thanks for your reply, now RD is willing to change the schematic to tight MR pin to VDD and increase the decoupling cap on VDD pins. 

We will try to test after this is done and see if this can improve the problem.

Thanks for your help.

Hi Michael,

Thanks for your explanation. We will work on trying to improve the noise and decoupling cap on VDD pin firstly.
Thank you.

Hi Gary,

and using a solid ground plane can be extremely helpful... :-)

Kai

Hi Kai

The PCB layout shows as above.The big red plane is the GND copper pour and the white part is the GND layer.Is that the solid ground plane you mentioned before?

Hi User,

yes, the red is the ground fill and the white is the ground plane.

Where is the decoupling cap of TPS3307? It should sit very very close at the VDD pin of TPS3307 and should connect to the GND pin of TPS3307 via the solid ground plane.

The aura arround your vias is a bit large. This destroys the solidity of ground plane. We use in out projects a drill diameter of 0,4mm, a copper diameter of 0.8mm and an aura of 0.3mm arround the via.

Place the vias allways in such way, that there is a piece of ground plane between them. Only by this the vias are shielded against each other.

Kai

Morning Kai

The decoupling cap is at another side.Show as below.Actually ,in this PCB ,the drill diameter is 0.4mm,and the copper diameter is 0.8mm ,and the aura  arround the via is 0.4mm(company constraint).

The /MR pin is internally connected to the VDD pin so if /MR is causing false triggers, increasing the decoupling on VDD should help. You can also tie /MR directly to VDD externally if not being used. Under normal conditions, simply leaving the /MR pin floating should be OK. How did you determine that /MR was causing the reset but not any of the sense pins?

-Michael

Hi Michael

Until now ,I still can not determine which pin cause the reset .If you check my  before post,you can find that before the reset comes,narrow and heavy noise spikes apears at sense pins last less than 1us,which do not meet the condition for causing the reset according to the datesheet, am I right?

Hi User,

when you have a look at your scope plots again you will find, that all signals are falling down exactly at the same time. It cannot be seen the propagation delay times which usually should be visible. So, I think the circuitry you have connected to the /RESET output is forcing down the /RESET line. And because the /RESET output is a push pull stage the whole chip is short circuited, the VDD pin and all the sense inputs!

Take care, there are microcontrollers with a /RESET input which behave sometimes as output! The microcontroller can toggle this input down, as if there is an output and not an input! In such a case you would need a reset chip with open drain output and not push pull stage.

Kai