LSF0108: Output lines not being pulled to VccA voltage

Irfan Shaikh

Part Number: LSF0108

Tool/software:

Dear Team,

I am using LSF0108QWRKSRQ1 and i have also attached a snapshot.
In this VccB i have used as 3.3V and VccA as 1.8V mentioned as Vadj FMC.
The HPD from the TX connector is given as 3.3V on VccB side when i measure, but after the level translator it is giving 2.7V instead 1.8V.

Regards,
Irfan

6 months ago

0 Tyler Townsend 6 months ago

TI__Guru 59128 points

Hi Irfan,

The LSF0108 is incorrectly connected for level translation mode. See Figure 8-1 in the datasheet.

VREF_B and EN should be directly shorted. 200k then pulls-up to 3.3 V supply.

You currently have EN --> 200k --> VCCB.

Also see figure 9-1 as another example:

There should be pull-up resistors on both sides of the device for each Ax/Bx channel. Can we confirm this is the case?

Regards,

TYler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

There are no pull-ups for all but i have given only for the SCL and SDA.
Also, what if I remove 200K and put a 0E resistor and not put a 200k pull up for VccB.

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

Pull-up resistors should exist on SCL and SDA. For downstream channels, a PU resistor is required.

If you remove 200k resistor and replace with 0 ohm resistor, there is potential for low-impedance path to be forms from VCCB and VCCA and if EN voltage is large enough:

Current will flow from 3.3V VCCB through the 0 ohm resistor into Vref_B through Vref_A into the 1.8 V supply without current limit other than the resistance of the channel FET.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

There are pull ups on both upstream and downstream for SCL and SDA.
Is there any other solution by which we can achieve the proper functionality.

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

I don't think I am quite understanding your point. Is the fact that LSF0108 needs PU resistors on both sides problematic?

If so, please look into TXS0108, it is an 8-channel bi-directional level shifter with one-shot accelerators and internal PU resistors on each side of the device. No need to implement pull-up resistors externally like in LSF.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

No, that's not the problem.
I am facing the voltage mis-match as mentioned before on HPD signal.
As you mentioned about the 200k and EN, is there any other option to do?

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

I will respond shortly.

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

Are there any level translator of same footprint available so that I can check on it.

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Tyler Townsend

TI__Guru 59128 points

Hi Irfan,

The main problem is the connection of the VCCB and EN pin. Fixing this should fix the voltage mis match you are seeing.

Because EN = 3.3V, the channel FET is always ON. This means the voltage on "A" side (1.8V) sees a voltage divider from 3.3V side --> PU resistor --> Bx --> channel FET --> Ax --> PU resistor --> 1.8 V supply.

The voltage at Ax will measure greater than 1.8 V since the channel FET has such high gate voltage.

In normal voltage translation mode, EN = 1.8 V + 0.6 V = 2.4 V. Connecting to 3.3V directly through 200k is 0.9 V higher. 1.8V + 0.9V = ~2.7 V.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

If i remove that 200k and add 0E in place of it , will it work?
And 200k pull up is not required?
Or the 200k is must?

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

The way that the device is connect sets the gate voltage always to 3.3V.

Whether you short with 0 ohm or use 200k ohm, it should make no difference.

ALso, the FET should be experiecing a large amount of current. There is a current going from 3.3V into the 1.8V supply, limited only by the RDS_ON of the channel FET which is only 10's of ohms. I expect reliability issues for this device in the near term.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

Are there any alternate part with footprint compatible to this?

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

I don't think we have a footprint compatible device. The EN and VREF2 connection would still need to change.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Hi Tyler,

What if i short the EN pin to VrefB pin and mask the pad of the IC and pull up with 200k externally to 3.3V?
Will this work properly or i should change the design in next revision by connecting the EN to Vrefb andf pulling it through 200k.

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

To get this to work as is, you could cut the trace and blue wire VCCB to EN.

This should get the board working properly. Next board revision, I would follow similarly to figure 9-4:

This is a pretty good example of the correct connection scheme.

Regards,

Tyler

0 Irfan Shaikh 6 months ago in reply to Tyler Townsend

Intellectual 490 points

Dear Tyler,

I did the rework as mentioned, i masked the pad of VrefB and wired the EN pin to VrefB, but i am not getting any output form the A5 .
I probed the B5 and got 3.3V and i probed the A5 side which is output of B5 but i am not getting any output. its 0V.
Can you please tell me what might be the issue?

Regards,
Irfan

0 Tyler Townsend 6 months ago in reply to Irfan Shaikh

TI__Guru 59128 points

Hi Irfan,

What is the EN+VREFB pin voltage reading?

Regards,

Tyler

Interface

Interface forum

LSF0108: Output lines not being pulled to VccA voltage