TMUX1308-Q1: Is is larger than datasheet

What are the voltages at the control inputs?

This is not the same circuit as in figure 8-2 of the datasheet.

Hi Clemens and Rami,

Customer is testing on-leakage current (as the right picture 8-3 in the following), not off-leakage current， which is 8-2.

You could see that customer test situation is ok. Customer put “D” as floating.

And because it is on-leakage, the /EN is Low.

As you could see in the waveform customer tested, when S4 from not selected to selected, there are 20mV in S4. During on and off, S4 should have no voltage change, but now here is 20mV.

Customer think this 20mV voltage jump is coming from the Is(on). And Is(on) go through the res and make 20mV.

That is the reason why customer is very very concerned about Is(on). 

By the way, customer also test TMUX1208 with the same method. And 1208 Is(on) is nA, which is the same as datasheet.

Could you give some idea about why 1308 Is(on) is so larger？

Please kindly give some advice! Here is customer SCH.

Amelia,

Does every channel show similar results or only S4 and is this persistent across multiple chips? Could they run the test given the parameters from the datasheet? These are the conditions for how we've spec'ed it (below). However, their test conditions aren't data points we currently have for you to reference. I would like to see if the results are the same given the same tests in the datasheet.



In the meantime, allow me a day or two to get lab results and see if I'm seeing the same thing. I will update you on Monday.

Thanks,
Rami

Hi Rami,

Other channel are the same.

In the figure 8-3 right side, there is no need to add Vd... Why datasheet parameters have Vd?

Looking forward for your test result.

Hey Amelie,

I got into the lab today and ran tests in the same fashion as the customer did, with the same setup the customer ran (Vref = 1V 62kohm Rs):


 
I didn't see what they were seeing. The △Vout was sub mV (~0.02mV) which would equate to a leakage in the 100's of nA range. 

Looking back at the graph of S4 being selected and not selected, I may have misunderstood what was happening there (see below)


The scope shot isn't totally clear but it looks like D is also being pulled down when S4 is selected. This wouldn't be an on-leakage test if they're looking at a voltage present from source to drain when the path is selected. Is this just checking the voltage on a selected path? It looks like there is a problem with not being able to pull the signal low on the path. 

These are fairly basic test but I ran a test with the selected path on vs off where I left the D pin with a 1V input and turned the S4 path off and S0 selected. The signal was isolated properly. 



With Dpin at 1V and S4 path on/selected, as expected the input follows the output.



If the path selected shows 20mV, as shown in your scope, this is because there is 20mV present somewhere. The schematics didn't show much that could help identify what could be happening however. Are these all just sensors being fed to an ADC? Any pullup/pulldowns on the line?

How are they testing this? Are they using an in house board and have they tried replacing the chip to see if it's a problem with just the individual chip?

Thanks,
Rami

Hi Rami,

Rami Mooti1 said:

This wouldn't be an on-leakage test if they're looking at a voltage present from source to drain when the path is selected.

If we can‘t choose Is(on)=△Vout/Rs to calculate the Is(on), how could we measure the Is(on)?

Amelie,

My comment was aimed towards the specific waveform sent. It looks like there is some biasing happening on the Drain pin (it goes high then back low). For Is(ON) the D pin is floating. If there's a path for the current to go through the switch then it isn't leakage, it's more of a characteristic of the signal path. This current would overpower any on leakage so it would be difficult to measure it properly. 
When we characterize the leakage, we typically will use a picoammeter for precision measurements vs measuring voltage drop across a resistor.

Amelie Zheng said:

In the figure 8-3 right side, there is no need to add Vd... Why datasheet parameters have Vd?

Vs = Vd so it shouldn't matter on the naming. This was just worded this way to cover Id(on) and Is(on)

Did the customer run tests with the values provided in the datasheet and see the results? Vs = either 0.66V or 2.64V

Hi Rami,

Yes, customer change Vs from 0V to 1V to test Is(on) also.

Rs=100k，Vref=1.0V，temp is 65℃，The variable is Vs(Rs voltage).

Hello Amelie,

I ran the same test on the device I have in lab with the varying Rs as above and got different results. I used a 62K resistor since it was the only through hole resistor near 100k I had available but the resistor does not affect the leakage. Aside from that, the setup was the same and the results are below

As you can see the results show show a leakage <1nA. It is to be noted here that this is at room temperature so the results will be a bit better than the 65C the customer is working with. These results align with our database results at room temperature for on leakage. 

I looked through our database and we have test results where the Vs is 1V and Vcc = 3.3V for on leakage at 85C. These results also show that the on leakage was in the 10's of nA's. 

I'm not sure exactly why they're seeing what they're seeing but I have some notes and thoughts below that I mentioned previously that may help us figure this out :

The waveform you provided where s4 was pulled to 20mV looks like they're actually selecting through the 8 channels and the error only happens on channel 4. It looks like there is a transient every time a switch is selected. It was previously stated this happens on all channels though. Could you confirm whether all channels show the same behavior?

Similarly, has the chip been replaced? This could be damage to the individual chip so looking to confirm if it's a chip problem or a layout probem.

Lastly, on the S4 waveform D has a bias on it. D should be floating when testing the Is(on) so i'm not sure what is actually trying to be shown here besides that the signal isn't being pulled down all the say. From the schematic, it looks like all there is on the line is a sensor that feeds to an ADC, though. 

Thanks,
Rami