• State TI Thinks Resolved
  • Replies 7 replies
  • Answers 4 answers
  • Subscribers 34 subscribers
  • Views 136 views
  • Users 0 members are here
Support feedback
Options
Options
Related

TXB0604: TXB0604 Alternative for 1.8-V eSPI With Only ±2-mA Source Drive Strength

HongXi Chiu
Prodigy 30 points

Part Number: TXB0604
Other Parts Discussed in Thread: TXB0104

Hello TI Support Team,

We are looking for an auto-bidirectional line redriver for an eSPI interface between an AMD CPU/PCH and a Super I/O device.

Application conditions:

  • eSPI_IO[3:0], bidirectional push-pull
  • VCCA = 1.8 V
  • VCCB = 1.8 V
  • eSPI clock = 33 MHz
  • No DIR control signal is available
  • CPU/PCH guaranteed output drive strength is approximately ±2 mA

The total PCB routing length exceeds the AMD recommended eSPI routing-length guideline. Therefore, the device is required mainly as a line redriver or signal booster to divide the PCB channel and improve signal integrity. Voltage-level conversion is not required because both sides operate at 1.8 V.

The TXB0604 appears to be functionally suitable because it supports eSPI, push-pull signals, automatic direction sensing, and long-transmission signal boosting.

However, the TI TXB-type translator comparison table specifies an input driver requirement of ±3 mA for the TXB060x family, while our AMD CPU/PCH is guaranteed for only approximately ±2 mA.

Could you please confirm:

  1. Is ±3 mA a mandatory minimum requirement for reliable TXB0604 direction sensing?

  2. Should the TXB0604 be considered unsuitable when the source device is guaranteed for only ±2 mA?

  3. Does TI have another 4-bit auto-bidirectional device supporting:

  • 1.8-V to 1.8-V operation
  • Push-pull eSPI_IO[3:0]
  • No DIR control
  • 33-MHz eSPI
  • Input driver requirement ≤ ±2 mA
  • Line-redriver or signal-boosting capability

  1. Could the TXB0104 be considered because its input driver requirement is listed as ±2 mA?

  2. If TXB0104 is a candidate, could TI confirm its suitability for eSPI turnaround and distributed PCB trace loading at VCCA = VCCB = 1.8 V?

If no TI auto-direction device meets these conditions, please advise whether a DIR-controlled transceiver with external direction-generation logic is the recommended architecture.

Thank you.

  • 0
    TI__Guru 61061 points

    Hi Hongxi,

    Can you verify in device datasheet for IOH/ IOL specs? Typically, CMOS structure devices provide multiple IOH/ IOL parameters given VOH/ VOL conditions. 

    Please note that the input driver requirement is there to ensure that the device interfacing with the TXB-type translators can change the logic state/ direction of the signals during communication, due to the auto bidirectional capability of this device.

    Consider if an input is driving a low state into the TXB0604 before transitioning to a high state. To force the output side to logic high (lets assume 3.6V), we can consider VCC/2, or 1.8V in this case as the threshold. If we calculate the current required to drop 1.8V across the 1kohm resistor, it would mean, 1.8mA is needed to force a state change, meaning then can the output switch states/ direction. As margin, an additional 1.2mA of drive current is added to ensure robust signaling, hence the 3mA requirement. 

    Regards,

    Jack

  • 0 in reply to Jack Guan
    Prodigy 30 points

    Hi TI team,

    We are evaluating TXB0604 for a 1.8 V eSPI application.

    We have checked the IOH/IOL capability of our source device, and it appears that the source device cannot meet the approximately 3 mA drive requirement of TXB0604.

    Could you please recommend an alternative TI device that is suitable for a 4-bit bidirectional eSPI interface and requires lower drive strength from the source device?

    Thank you.

  • 0 in reply to HongXi Chiu
    TI__Guru 61061 points

    Hi HongXi,

    The alternative would be the TXB0104 requiring 2mA as the input drive requirement.


    Regards,

    Jack

  • 0 in reply to Jack Guan
    Prodigy 30 points

    Hi Jack,

    Thank you for recommending the TXB0104.

    I understand that its ±2 mA input driver requirement is more suitable for our source device.

    I would like to confirm one more point regarding our intended application.

    As mentioned in my original post, our main purpose is to use the device as a buffer/repeater-like device to divide a relatively long eSPI PCB channel and improve signal integrity, rather than for voltage-level translation.

    However, I noticed that Section 7.3.3 of the TXB0104 datasheet recommends short PCB trace lengths due to the output loading and one-shot considerations.

    Would TI recommend the TXB0104 for this type of application, where the device is placed between the CPU/PCH and SIO to divide a long eSPI channel into two shorter segments?

    Thank you.

  • 0 in reply to HongXi Chiu
    TI__Guru 61061 points

    Hi HongXi,

    Unfortunately the TXB0104 should not be considered as a buffer solution for your application. It would only be able to level shift the eSPI signals here. I suggested it previously  to satisfy the input driving requirement needs. 

    Regards,
    Jack

  • 0 in reply to Jack Guan
    Prodigy 30 points

    Hi Jack,

    Thank you for the clarification.

    I understand that the TXB0104 should not be considered as a buffer solution for this application.

    Since our main requirement is to divide a relatively long eSPI PCB channel and improve signal integrity, could you please recommend whether TI has another device or solution suitable for this application?

    Ideally, we are looking for a solution supporting:

    • 4-bit bidirectional push-pull eSPI I/O
    • 1.8-V operation
    • 33-MHz eSPI
    • No external DIR control signal
    • Buffer/repeater-like functionality for improving signal integrity on a long PCB channel

    If TI does not have an auto-bidirectional device suitable for this application, could you please advise what architecture TI would recommend instead?

    Thank you

  • 0 in reply to HongXi Chiu
    TI__Guru 61061 points

    Hi HongXi,

    Unfortunately, there isn't a solution that we can provide today that satisfies the input driver requirement ≤ ±2 mA with the above requirements you are lookin for. 

    Regards,

    Jack