TMUX1511: RMII 50MHz clock isolation using TMUX1511

Hi Ian,

Placing the control logic pins at ground logic low will create a high impedance path between S and D. This doesn't automatically mean that D will go to ground unless you have pull down resistors on the COM side of the circuit.

Can you post a Scope shot off the 50kHz signal on S and D?

Thank you and best regards,

Ambroise

Yellow = Input on S1

Pink = Output on D1

SEL1 = 0V

Hi Ian,

does the same happen when you hook up a logic low voltage to SEL? (between 0V and 0.45V) 

And how is the off isolation on the other pins? 

Best regards,

Ambroise

Hi Ambroise,

Yes there is no difference when SEL is between 0V and 0.45V.

The isolation between other pins is very good.

Adding a 3k3 pull down resistor on the output D reduced the output swing to around 120mV whilst using a sig gen as the input.

I tired this in-circuit with the actual 50MHz clock from the 10/100 Ethernet PHY going to the OMAP and the pull down resistor had no affect.  More-over in this real setup the output voltage swing is ~450mV

Scope Trace: TMUX1511in off state in circuit inbetween PHY and OMAP

Pink = input into S

Yellow = output D

SEL = 0V

*Note different V/Div

Scope Trace: TMUX1511 in off state in circuit inbetween PHY and OMAP

Pink = input into S

Yellow = output D (with 1k pull down)

SEL = 0V

*Note different V/Div

Thanks

Ian

Hi Ian,

Just to be sure, did you try the 50MHz clock on other channels? 

And could you share a partial schematic?

Thank you,

Ambroise

Hi Ambroise,

Yes, I have tried on all channels and get the same result.

We are at the stage of approving the IC before it makes it into our design, so I dont have a final schematic.  I have however quickly drawn a schematic of how we would be using the IC.  I have purposly left off siganls / components that are not relevant to this issue.

Thanks

Ian

Hi Ian,

Is the Impedance on S & D known? the off isolation can be calculated as 10log10((Vout²/Rout) / (Vin²/Rin)) if the impedances are not the same.

Can you check with your set-up?

Thank you,

Ambroise

Hi Ambroise,

The input and output impedance should be matched between ~50-68 ohms.

As mentioned the schematic provided is roughly how we would like to use the switch.  I am not trying to debug our circuit,  I have been trying to prove on the bench that the switch behaves as per the datasheet before we include it in a further design, and I can't see how it does.

Are you able to supply a working test setup or better still provide scope traces of the switch isolating a 50MHz digital clock?

Thanks

Ian

Hi Ian,

thank you for the details, I'll have to look into this in the lab, as we have not seen this behavior yet.

Thank you for your patience,

Ambroise

Hi Ian, we are expecting the results from the lab mid next week. I will post the results around Wednesday.

Best regards,

Ambroise

Hi Ian,

we just got back the results from the lab.

We found that the Off-Isolation measurement is very dependent on the layout and set-up.

In Image 1 you can see our replication of your issue. In the 2nd Image you can see the Cross talk between S1 and D1 without any chip on the board.

If you subtract this from the original measurement without IC you should achieve roughly -30dB Off-Isolation.

For the layout of the actual board, you'll have to keep in mind not to run the wiring in parallel, if possible.

Best regards, 

Ambroise

Hi Ambroise,

Thanks for the results.

Repeating your tests using an the EVAL-16TSSOPEBZ from Analog Devices I got the following:

TMUX1511 in off mode:

No TMUX1511:

0.446-0.159 = 0.287V

20log(10) 0.287/3.3 = -21.2dB

This is close to the isolation I require in my design.  I will just make sure we take extra care routing the clock to minimise crosstalk and  have options to condition the signal both sides of the switch to make sure we don't exceed the maximum input voltage into OMAP whilst its unpowered.  I am doubtful however that I will achieve the datasheet values.

This is the most suitable switch I could find from TI website - but do you know of another better / more suitable device? It has to be of small size and ideally 4 channel.

Thanks

Ian