TPS27081A Output Votage

Hello,

We have the same issue. We are using these values :

R1 = 5.1k

R2 = 510R

C1 = 82nF

Using R1 = 10k, R2 = 0 and C1 = open, we don't have this issue but the rise time is too fast and the input voltage drops significantly.

Regards,

Jérôme

I see no evidence in the device datasheet of an output pulldown mechanism (ie load discharge). So, it is up to the load to pull output to 0-V. A resistive load should discharge the output cap to zero while a semiconductor type load may not. Try connecting a resistor from output to GND and see what happens.

If the semiconductor type load has other voltages (such as an IC with an IO and core voltage), leakage from the higher voltage source back to the 3.3V source can keep the output of the switch biased.

Hello,

Here is a snapshot of the TPS27081 output (pink) where R1 = 10k, R2 = 0 and C1 = open. We have no 0.8V level between 0V and 3.3V. The total impedance seen by the switch is about 200 ohms when using this configuration.

I tried adding a 200 ohms resistor at the output, because the load was about 20k (when R1 = 5.1k, R2 = 510R and C1 = 82nF) and its doing the same thing.

There is a FPGA and a HDMI receiver connected to this switch.

Thanks,

Jérôme 

Hello,

On my design, there is an I/O connected to this switch.  I tried to disconnect the load, but there is no change.

Here is a snapshot of the TPS27081A output where;
C1 = 0.1uF
R1 = 10K
R2 = 1K 
Output Resistor = none

Result: We have 1.8V level between 0V and 3.3V.

 

C1 = 0.1uF
R1 = 10K
R2 = 1K 
Output Resistor = added

Result: We have 0.8V level between 0V and 3.3V.

 

C1 = open
R1 = 10K
R2 = 1K 
Output Resistor = added

Result: We don't have this issue but the input voltage drops significantly.

 

C1 = open
R1 = 10K
R2 = 1K 
Output Resistor = none

Result: We don't have this issue but the input voltage drops significantly.

 

Best Regards,
Hide

Hide,

The TPS27081A is a P+N switch. Therefore it has all the disadvantages of a P+N switch.

Before power is applied, the voltage on the "C1" capacitor (C147 is the first schematic) is zero.
When power is applied to VIN (the PMOS source) the PMOS gate (R1C1) is zero volt and PMOS drain (VOUT) is also zero volts.
Therefore the PMOS pass element is turned on by the low voltage on the R1C1 pin.

The 0V on the ON/OFF pin (NMOS gate) will keep the NMOS turned off. However the NMOS can not turn the PMOS off; it is only capable of turning the PMOS on.

Hi,

I'm working with Jerome on this.

I understand that, when VIN is rising-up, the voltage accros the capacitor C1 tends to stay at 0V for a moment (this the purpose of a capacitor). Since VOUT = 0V at T=0s, Vgate of the PMOS will tends to stay at 0V. So the PMOS will start to turn ON, because it's Vgs will be negative at this point. So you should expect a glitch at the output when VIN is rising-up.

What I don't understand: after a moment, the voltage at the gate of the PMOS will change from 0V to about VIN.... so then, the PMOS should turn OFF again (unless the NMOS is turning ON...). What I measured is a Vgs of 0V on the PMOS, but the output is about 0.825V (VIN = 3.3V in our case and the ON/OFF signal is still 0V).... This 0.825V voltage shouldn't come from another power supply, since only removing R2 correct this problem!!!?!

Can you explain me what is wrong with my conclusion? (since my theory and the practice are not in sync!)

Thanks,
Denis.

Denis,

There is a lot of data in this forum thread and my earlier explanation may not cover everything that is happening.
I will need to do some simulation and or take lab data to explain all the waveforms posted.

 

Ok,

Thanks!
Denis 

As promised I will answer all questions, but do so one at a time and in order.

This is the reply specifically for this post: Posted by HIDE on May 01 2013 02:54 AM

The uncharged capacitor from R1C1 to output will turn the output on when VIN is applied.
This is the schematic I simulated. The MOSFETs in schematic have higher VT than the TPS27081 so I increased VIN to 5V.

This is the resulting waveforms. The output partially powers up as I predicted in an earlier post.

I also attached the TI-Tina (free simulator) file.

 

Here is reply to post jerome bolduc  on May 01 2013 09:03 AM

Here is the waveform with components change to 5.1k, 501 ohm and 82nF.

 

Changing "C1" capacitor to zero. Result is no significant VOUT increase.

 

Replay to post Posted by Eric Wright  on May 01 2013 09:32 AM

It was suggested the a load is needed. Here is the power up waveform with a 10 ohm load.

VOUT still increases, but the voltage bleeds off over time.

 

Hello,

Is the TPS27082L package could be an alternative to our problem? I see in the datasheet it has an excellent OFF isolation under transient.

Thanks,

Jérôme

Ron,

Thanks for the explanation.

I understand output voltage doesn't keep to stay 0V after the turning ON and it is the correct behave on this device.

 

Let me ask you one more question.  Below is the waveform where Vin=3.3V, C1=0F, R1=10K ohm, R2=1K ohm.

In this case, input voltage fall to 1V, and output voltage keeps 800mV after the turning ON.  Is this correct behave on this device?

 

Best Regards,

Hide

jerome bolduc said:

Is the TPS27082L package could be an alternative to our problem? I see in the datasheet it has an excellent OFF isolation under transient.
Thanks, Jérôme

I am looking into the TPS27082 design to see if that will be a good solution.

Jerome,

I checked the design and the TPS27082L will have much improved performance over TPS27081 for VIN rise-time immunity.
The TPS27082L will short out R1 with ON/OFF pin is low.

R2 pin will be a ground pin on the "082L" device and the R2 function will be an internal fixed 12.5k resistor.

Hello,

Cited previous post:

Does it mean that for TPS27082L, the intermediate voltage problems during OFF, will NOT happen as in TPS27081A?

 

Matt

Ron,

Got it, thanks!

 

Matt

Ron,

TPS27082L has a hysteresis symbol before Q2, is it a Schottky?

Does this hysteresis device/component length the rise/fall time?

Additionally, the TPS27082L datasheet only list lower bound

t off > (R1 × C1 sec)

for fall time. Do we have an upper time bound when VOUT falls to 10% VIN? Is falling slower or faster than rising?

 

Matt

Matt,

I don't have this information.  Is there a specific R1, C1, and VIN that you are interested?

Ron,

For

VIN = 3.3V, R1=125KOhm, C1 = 400pF

is it possible to get an accurate estimate of the falling time? Could this be done like for the TI Lisa simulation results you gave above?

 

Matt

Matt,

I have wave forms for 3.3V input, R1 = 100k, C1 = 330pF. Load is 50 ohms
For TPS27081A, R2 = 3.16k and TPS27082L, R2 = 0k (12.5k internal).

All scope plots have CH1 = TPS27081A output, CH2 = TPS27082L output, and CH4 = ON/OFF pin input.

Dear Ron,

It can be seen clear that:

1. removing C1 reduces both rise and fall time for both devices.

2. rise time of 82L with internal 12.5Kohm R2 is smaller than 81A with external 3.16Kohm R2.

3. 82L’s improvement on falling time is very obvious!

Deep thanks for measuring the real waveform for us, I appreciate the work.

 

I would like to ask two additional questions:

1. Since both devices are rated as 3A current grade, would the results change significantly if we use a smaller R_L, say 2ohm or 1ohm? Could the measurement be done again, only for TPS27082L, and without C1?

2. Can the device (TPS27082L) sustain repeated 1A-3A pulse 200μs at 200Hz? Would such periodic strong pulse change the performance measurement as comparing with individual, low current squares?

 

Matt

Matt,

I see that there are open question in your post. Is information still needed?

Hi Denis!
I have the same thought as yours. I think after a while, the output should decrease to zero rather than stay at 0.8V. That's very strange.
do you have any progress in this question?

Hi Ron!

Got it. If there is no resistive load, the 0.8V voltage may stay for a long time because there is no path to discharge the load capacitors.

Thanks for answering!

Hi Ron!
Could you please take a look at my questions about TPS27802L here? Thanks a lot !
e2e.ti.com/.../1699913