• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 20 subscribers
  • Views 1808 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LSF0102: LSF0102 and LSF0204

Bruce Dove
Prodigy 60 points
Part Number: LSF0102
Other Parts Discussed in Thread: LSF0204,

Hi,

We have an MDIO hot swap application that also requires level translation. Would like to use either the LSF0102 or LSF0204 parts. In our application, the backplane interface is 3.3v, this is the hot swap interface. Internally, the other side is either 2.5v or 1.8v.

For either LSF parts, will they handle the hot swap situation, done on the 3.3v side? Datasheet for LSF0102 mentions this in a bullet, but it does not discuss. The I/O pins would see voltage initially, at 3.3v, however both vref pins and the EN would be unpowered. After powerup, the VrefB side would be powered on, with EN pulled high also. VrefA would remain unpowered until firmware decides to turn that interface on. This could be quite a while.

Are there any issues when either vrefA or vrefB is unpowered, but  the other side is powered, for either LSF0102 or LSF0204 parts?

Is there any incorrect behavior if the LSF0102 EN pins are not powered, but the VrefB power is applied? They are always shown connected, but we may wish to have another control point. LSF0204 seems to separate this entirely. Would that be a better choice if this is necessary?

  • 0
    TI__Guru* 97046 points
    Hi Bruce,
    Typically, a backplane means high parasitic capacitance (> 70pF), and the LSF family doesn't do well with large load capacitance (to be fair, there aren't any auto-bidirectional translators that _do_ like large capacitive loads).

    I would recommend using a transceiver (ie direction controlled translation) if at all possible.

    The EN pin of the LSF01xx translators must be tied directly to the VrefB pin, which is to be connected to a 200k resistor and then to the larger supply (3.3V in this case). If you want to put the device into a high impedance state, connect an open-drain buffer to the EN/VrefB voltage node and drive the line low when the high impedance is desired.

    The LSF0204 works differently, allowing the EN pin to be driven from a normal buffer (push-pull output) operating at any voltage higher than VrefA and lower than 7V.

    We just finished creating a video series explaining these parts in detail, but it hasn't been released yet. I will make a note to post the link to this thread as soon as it is posted to the TI training site.
  • 0 in reply to Emrys Maier
    Prodigy 60 points

    Hi Emrys,

    Unfortunately the other parts we need to use here do not provide any direction pins on the MDIO interfaces, so we cannot use a direction controlled part.

    We have looked at some of the other alternatives, TXB, TXS, etc, and given that we have no control over direction, and that this is open drain, and also hotswap, it seemed to us that going the LSF route would be the lowest risk approach for us. We know there will be capacitance on the backplane side.

    If there are any other or newer parts to look at besides the LSF parts, please suggest.

    We can limit the MDIO link speed; we also have drive limitations with the parts on either side of the link, with the lowest performing part being 4mA typical.

    However we want to be sure that, if the answer is the LSF part, that we can use either the LSF0102 or LSF0204 in a hotswap isolation application, and if there are any special requirements for that.

    Thanks,

    Bruce

  • 0 in reply to Bruce Dove
    TI__Guru* 97046 points

    The LSF family is the best solution we have for what you want to do. There's always a balancing act on pull-up resistor selection (on the high side) for passive translation.

    The pull-up resistors need to be large enough to limit the current for the weakest driver on the line: \

    3.3V/4mA --> RPU ≥ 825 Ω

    The pull-up resistors need to be small enough to allow for the required data rate:

    Data rate (in bps) = 1/(6*Cp*Rpu)

    Assuming 250pF parasitic capacitance and 1 kΩ pull-up resistor, maximum data rate is ~667 kbps

    The LSF device _can_ be put into a high impedance state (as described in my earlier post), however it does not provide precharge or other live insertion protection.  You might like to read these application reports that discuss specific hot insertion issues:

    Live Insertion

    Logic in Live-Insertion Applications With a Focus on GTLP

    Live Insertion With Differential Interface Products

  • 0 in reply to Emrys Maier
    TI__Guru* 97046 points

    I'm happy to announce that we have officially released a series of 8 videos that explain how to use the LSF family of translators in detail. These videos are part of a new training series called "The Logic Minute" and are available here:

    The Logic Minute

    Please let me know if I can be of further assistance.

  • 0 in reply to Emrys Maier
    Prodigy 60 points

    Hi Emrys,

    That link does not work. Access is denied, seems to be at the TI end.

    Thanks,

    Bruce

  • 0 in reply to Bruce Dove
    TI__Guru* 97046 points
    Thanks Bruce,
    I'm talking to my digital marketing team to see what's the issue. Hopefully we can get that up and running today!
  • 0 in reply to Bruce Dove
    TI__Guru* 97046 points
    I updated the link above. It turns out that the link I originally posted was for internal use only - sorry about that!