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.

TPS65987DDJ: Question on TPS65987D[Dx] I2C clock frequency operation when running as a Master.

Part Number: TPS65987DDJ
Other Parts Discussed in Thread: TPS65987D

On the TPS65987D[Dx] spec. regarding I2C Master operation in Fast Mode there's this curious specification that the Typical fSCl_Master is 320 KHz.

It seems I2C master mode characteristics (such as clock frequency) are really configured by TI's FW rather than by the user as is the case in a typical microprocessor endowed with an I2C master controller - at least I haven't found a Clock Frequency setting in the Application Customization Tool. Does that means that TI programs at the factory some value for the master clock frequency which "typically" is 320KHz? Is that perhaps based on achieving minimal compliance of 1.3μs for tLow? tLow is typically a challenge to meet as it's greater than a 400KHz half-cycle of 1.25μs, generally requiring a slight reduction of below 400KHz to achieve compliance.

Furthermore, albeit such Master I2C has 3.3V Pull-Ups (powered by LDO_3V3), does this e2e.ti.com/.../tps65987d-tps65987d-i2c-input-threshold-for-ldo_1v8-and-ldo_3v3 article imply that one shall use the 1.8V VIH & VIL I2C levels to characterize I2C inputs into such TPS6597D[Dx] I2C inputs - particularly the SDa line?

Reason I ask these questions is that I for example see on my board SCl clock frequency of 335KHz (which conceivably should be able to yield a tLow of around 1.49μs) with tLow being about 1.33μs (very close to the 1.3μs I2C limit). I thus wonder if there is some tolerance (particularly across temperature) where some specimens could actually have tLow below the required 1.3μs.  Furthermore, we have N-FETs based 3.3V to 1.8V translators (along the lines of www.nxp.com/.../AN10441.pdf), where those translators tend to reduce a bit the low time so even though the TPS65987DDx may not violate tLow, a slave connected to it may be receiving a clock that is on violation.

Is it at all possible for a user to configure the clock frequency of a TPS65987D[Dx] I2C Master controller???

Thanks so very much,

Georg A. Mussenden

  • Hi Georg,

    Unfortunately the speed is programmed to the ROM at the factory so configuring the speed is not supported by the Apps. GUI.

    Ya the VIH and VIL are based on 1.8V.

    Regards,

    Peter

  • Thanks Peter for the quick reply!

    So in effect TI adjusts on each specimen fSCl_Master at the factory as to be [minimally] compliant to tLow over operating free-air temperature range?  Does that means that level translators should be avoided as they could cause tLow violations?

    Most indebted,

    Georg

  • Hi Georg,

    We have customers who have used either way and not just direct connection if you are implying, I understand in your case it's violating you can bypass the translator as it's using the 1.8V threshold.

    Regards,

    Peter

  • Ok, cool, thanks Peter.  Actually the situation is not as pressing after all as the slave devices that are on the translators are Fast+ capable so a tLow of only 0.5µs is needed, not 1.3µs.  If that weren't the case, our implementation would have been very marginal!

    TI might however want to have a small explanation (footnote) on the DataSheet on what the "Typical" 320KHz fSCl_Master means and perhaps somehow caution the user not to place translators if using Fast (not a problem with Fast+) devices.  Furthermore the user ought to be additionally warned that only Fast or Fast+ devices (no Standard) slaves can be attached, as it isn't possible to configure the I2C clock to 100KHz either for a slave that doesn't support Fast or Fast+.

    Thanks so very much and have a Great WeekEnd!

    Georg A. Mussenden