SN74LVC1G123: pulse width variation over temperature (-40℃ - + 125℃)

Michiaki,

The pulse width is listed in the datasheet as twOUT ( pulse duration) across temp for R and C values. 

Hello Michiaki-san,

First, I would like to be clear that the SN74LVC1G123 is not a precision device. If extreme precision in timing is required, I would recommend against using it.

Michiaki Tanii said:

1. On the figure7 and figure8 in slva720.pdf.
    Could you kindly advise the reason why the pulse duration is changed by temperature ?

Pulse duration is dependent on many factors including, but not limited to: internal capacitance, comparator variation, reference variation, leakage current, internal delays. All of these have some dependence on temperature.

Michiaki Tanii said:

    In figure7, We see that the pulse duration is changed linearity at VCC=5.5V but it isn’t changed between 10℃ and 30℃ for the others.
    Why is the above seen ?

Temperature change is part dependent - the device tested shows a relatively linear change across some temperatures, but this does not mean that all devices will have the exact same variation. The data is provided to show that there is minimal change over temperature, not to give a precise characterization for precision use.

Michiaki Tanii said:

2. Multiplier K

   Is it correct that multiplier K is defined by capacitor size and supply voltage ?

Yes, K is dependent on capacitor size and supply voltage, as well as temperature.

Michiaki Tanii said:

   Could you provide the multiplier K equation ?

   Customer think that K and capacitor size would not be proportional.

K is determined experimentally - I do no have an equation to share for it.

K and timing capacitor size are not directly proportional, but they are related. For very large capacitance values, K becomes relatively constant. For small capacitance values, K changes significantly. This is related to the difference between the internal capacitance (which is non-linear) and the external capacitance.

Michiaki Tanii said:

3. Could you kindly advise the device that has smaller variation for Tw over temperature ?

All monostable multivibrators have temperature dependence and are not designed for precision delays. I would recommend using a different method of timing for precision delays, such as a comparator + RC circuit, or a crystal + microcontroller.

Michiaki Tanii said:

4.  I got the data customer measured.

     Could you kindly review the data whether it is matched with the theory or not offline.

I would be happy to review their data.

Michiaki Tanii said:

1. If Figure 7 and Figure 8 are switched, we couldn’t see the 10x difference compared Figure 9 with Figure 8.
   It is about 7.5x difference.
   Could kindly advise why the change of Cext does not correspond with Tw ?
   Tw = 14.4 us at Ta = 10℃, Cext = 100 pF and Tw = 107.5 us at Ta = 10℃, Cext = 1nF.
   Cext is 10x difference but Tw is not 10x difference.

There is also a K factor involved. The equation for pulse width is:

tw = K * R * C

For a 100pF capacitor, the K factor is significantly larger than it is for a 1000pF capacitor.  In this case, K = 1.41 @ 5.5V for the 100pF capacitor, and K = 1.04 for the 1000pF capacitor.

Michiaki Tanii said:

2. Could you share the Fig 7, 8, 9 at between -40℃ and 125℃ ?

No. All data that was taken is provided in the application report.

Michiaki Tanii said:

3. In the case of fixed Rext = 100kohm, the rate of change for Tw is smaller if Cext is bigger.
   Please advise the reason.

The K factor is primarily dependent on the internal capacitance of the device. With lower capacitor values, the internal capacitance has a larger effect on the external capacitor charge rate. With large devices, the effect is minimal and the K factor flattens out. As you can see in the datasheet, figure 6, the difference in K factor between a 0.1uF capacitor and a 0.01uF capacitor is very small, but when dropping to a 1000pF (0.001uF) capacitor, the change is more drastic.

Michiaki Tanii said:

4. If Cext is fixied and Rext is larger, the rate of chage for Tw is smaller as well as the above case ?

If Cext is fixed and Rext is increased, the K factor will remain relatively constant. I say 'relatively' because there are still factors such as leakage that can affect operation of the device. I would recommend keeping timing resistors less than 1 Mohm to prevent a significant change in operation from leakage current (up to 0.25uA at the Rext/Cext pin).

Michiaki Tanii said:

5. Customer is considering to change Rext to 470kohm to get wide Tw by changing Cext.
   if Rext * Cext is fixed, the change rate for Tw by temperature is same ? 
   For instance, in the case of the below conditions, the change rate of Tw by temperature is same ?
   - Rext = 100kohm, Cext = 100pF
   - Rext = 470kohm, Cext = 21.4pF

I would expect to see similar performance across temperature to that shown in the application report. For a safety margin, I would recommend designing with +/- 5% tolerance.

I must repeat my previous recommendation - this is _not_ a precision device, and we cannot guarantee exact operation from one batch to another. Even if your customer extensively tests this device and proves that it works perfectly in their application, it is possible that next year a minor shift in process will result in different performance. We can only guarantee what is in the datasheet.

Michiaki Tanii said:

6. If supply voltage is 5V ± 2%, how variation for Tw is ?
   if we could estimated variation, please let us know how to do.