LSF0204: LSF0204

The description does not match the schematic. Vref_B must be higher than Vref_A, so the schematic is correct.

WIth strong pull-up resistors, all other devices must be able to sink the pull-up current.

What are the directions of these seven signals? The TXU0104/TXU0204/TXU0304 would be easier to use and more robust.

Hi Clemens,

Sorry for the Confusion

B side is connected to FT2232H IOs(3.3V) and A side is connected to DUT IOs(3.3V,2.5V,1.8V,1.5V). It operates at different voltages

We are using an LDO for 3.3V to VREF_B and since A side voltage varies for different DUT, I have connected VREF_A to 1.2V, which is the divided voltage from 3.3V(by using 1% voltage divider, pull down helps for current sink also)

TXU0104/TXU0204/TXU0304 datarate is 100Mbps for 3.6 to 1.8 translation and the drive strength is 12mA at 5V which is too low for our application

So far only with LSF0204 in one of our designs works at 30MHz (along with FT2232H) with Good Drive strength. But Pull_up voltage of B-side is fixed in that design

So kindly share your comments on the shared schematics consider the highlighted requirements

Thanks & Regards,

Rohini

The LSF is a passive switch and does not have any drive strength.

What data rate and drive strength do you actually need? (And why do you need more than 12 mA?)

Hi Clemens,

Yes, With LSF we could fix the Drive strength using external pull-up. Updated directions below

The data speed is 30MHz at 3.3<->1.5V,1.8V,2.5V,3.3V Level Translations

Signals go through a cable of 8"+ DUT signal length may vary depend on board size(Avg:5"). We may need more Drive Strength here practically

Is there any reason to not use LSF0204?

Thanks & Regards,

Rohini

The LSF0204 probably works, but the TXU0304 would work better (as a unidirectional buffer, the signals are more robust and symmetrical).

1.5 V / 120 Ω = 12.5 mA, so these pull-ups are as strong as the TXU. Please note that the pull-ups only drive the signals high; low signals are driven by the other devices (and all those devices must be able to sink 12.5 mA, or 27.5 mA at 3.3 V).

If you have doubts about the long cable, add a buffer at the other end of the cable.

Hi Clements,

We have used LSF0204 in one of our designs for JTAG itself

It has VREF_A=1.8V(constant) and VREF_B=3.3V constant. It is working seamlessly. That's the reason we are using it on this board

Instead of TXU could we use TXB0304. so that I could have a footprint compatible part with minor changes on design

Below is the change done for TXB0304/LSF0204 compatible design

Regarding addition of Buffer, our device should support 1.5V to 3.3V VCC and there are only 7 signals

After filtering for our requirement (chosen 8mA and >8mA Drive Strength) only options are 74LVC125A,74LVC126A

In datasheet, it is not very clear that it will work at 1.5V. It is mentioned as functional at 1.2V, then suggesting 1.65 to 3.6V in first row

Thanks & Regards,

Rohini

The TXB has a drive strength of 20 µA. Don't use it.

LVC devices work correctly at 1.65 V; at lower voltages down to 1.2 V, the output stays at the correct level as long as the inputs do not change.

LXC or AXC translators can be used as buffers over the entire 1.5 V … 3.3 V range.

Hi clemens,

What is the maximum signal frequency supported by LXC/AXC at 1.5V to 3.3V translation

Could not get this info in datasheet

Thanks & Regards,

Rohini

Hi Clemens,

I noticed that one of our boards use FT2232H+ SN74AXC4T774PWR and it was not working at 30MHz

Thanks & Regards,

Rohini

Hi Rohini,

Do you have waveforms capturing input/ output signals or schematic to further debug the non functioning AXC4T774? 

Regards,

Jack 

Hi Jack,

Since that board is not with me, I could not get the results now. but I will try to share it to you soon

As per Clemen's inputs we have added TXU0304 option to our design. Let me know if we could go ahead with this. U10 & U11 will be DNI. We will use it if LSF doesn't work

Thanks & Regards,

Rohini

Hi Clemens,

As per your inputs we have added TXU0304 option to our design. Let me know if we could go ahead with this. U10 & U11 will be DNI. We will use it if LSF doesn't work

Thanks & Regards,

Rohini

Looks OK.