0
\$\begingroup\$

I have two USB2512BI-AEZG(datasheet) ICs and would like to drive them using only one external crystal for well known reasons.

What is misleading me is ultra low input low voltage specified in the datasheet enter image description here

Driving second USB_HUB_IC XTAL_IN input pin from XTAL_OUT pin of first USB hub IC is not an option due to power management (there will be situations where only one USB hub will be working).

I simply cannot find any clock buffer with such low output level(this is not even LVCMOS) - even with 1.8V supply Output Voltage High of clock buffer will be 0.35V max.

Is this a glitch in the datasheet? Does clock buffer which I need even exist?

\$\endgroup\$
2
  • 1
    \$\begingroup\$ This means anything below 0.3V is guaranteed to be interpreted as low and anything above 0.9V is guaranteed to be interpreted as high, and the inputs only draw 10uA. I doubt you've seen any devices which don't meet this spec. \$\endgroup\$
    – τεκ
    Commented Jan 29, 2018 at 15:08
  • \$\begingroup\$ @τεκ - I think the OP's problem is that he can't find any clock buffers which guarantee a low output will be no more than 0.3V which his USB IC demands as an input. \$\endgroup\$
    – brhans
    Commented Jan 29, 2018 at 15:24

2 Answers 2

2
\$\begingroup\$

Actually, there is a glitch in the datasheet. But not in the area of min-max VIN on XTALIN pin. Here is the problem.

The USB2512 chip has XTAL driver designed to accommodate small-formfactor crystals in tiny 2 x 2.5 mm (and smaller) packages, which are getting increasingly popular in portable electronics, and cheap. These tiny crystals have very low power dissipation limit, 10 uW - 100 uW. To be not overdriven, these tiny crystals require low driving voltage. In USB2512 this requirement is met by having a fairly weak XTAL driver inverter running from 1.2 V rail. This is a simple CMOS inverter, and it will switch around 1/2 of its voltage rail, or at about 0.6V. So the specification essentially says that it needs 0.6V +- 0.3V to operate. The value of 0.3 V is taken from thin air by technical marketing with a very good fudge factor. In fact, anything that oscillates around 0.6 V (+-0.15V) will reliably toggle the internal buffers.

The problem is that any input signal that exceeds the LVCMOS12 levels (0V-1.2V) will stress the input I/O pad cell buffer leading to accelerated aging. So it is strongly recommended to feed the XTALIN with no more than 1.2 -1.5V signals.

This requirement is very easy to meet by using a simple resistor divider at the pin input, something like 150 Ohms/100 Ohms. More accurate resistor network can be selected by playing with your driver model in any SPICE simulator, even if you have long trace and have to account for transmission line effects.

NOTE: the effect of input waveform can be easily tested on XTALOUT pin, and the resistor network values can be corrected to get a square wave output with 50% duty cycle (although it isn't really necessary).

The glitch in this datasheet is that the absolute max value is listed as 4 V. This is too much, and the IC will likely die in a couple of month of operation due to pad fatigue.

ADDITION: There are several available datasheets, from different periods of product development. I found this note in one of datasheets,

7.2 External Clock

50% duty cycle ±10%, 24 MHz ±350 ppm, jitter <100 ps rms. The external clock is recommended to conform to the signaling level designated in the JESD76-2 Specification[5] on 1.2 V CMOS Logic. XTALOUT should be treated as a weak (<1mA) buffer output

\$\endgroup\$
3
  • \$\begingroup\$ So, I should go with the following - 3V3 PLL circuit, external clock buffer and at the end resistor divider. \$\endgroup\$
    – Bip
    Commented Jan 30, 2018 at 7:56
  • 1
    \$\begingroup\$ @Bip, you don't need any special "clock buffer", any Tiny Logic (say, SN74AUC1G00) will do the job. Or SN74AVC1T45. \$\endgroup\$ Commented Jan 30, 2018 at 8:34
  • \$\begingroup\$ I realize this question and answer are old, but I think it's useful to read a bit below where the 4.0V maximum was specified. In note 6.1, that datasheet says "...functional operation of the device at any condition above those indicated in the operation sections of this specification are not implied". In the operational section, it says the max voltage on ther XTALIN line is VDD33 (min is -0.3V). That does surprise me a bit -- I'd still consider a divider -- but at least it's not 4V. \$\endgroup\$ Commented Feb 22, 2023 at 1:02
1
\$\begingroup\$

https://www.idt.com/document/dst/49fct3805-datasheet

enter image description here

Since the inputs draw 10uA, meeting this specification is trivial.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.