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.

SN74CB3Q3305 for 1 pulse per second 5V TTL 50 ohm load switching

Hello, this is a question related to the suitability of the selected IC for this task.

The 1PPS (1Hz) pulse signal is commonly used for synchronization. We wish our equipment to include a capability to match the external cable line to 50 ohms, or disconnect it when a higher impedance is preferred, and have preselected this TI switch 3Q3305 to close / open a low impedance contact from the cable to the load.

The PPS signal is very low pulse repetition frequency but the time precision of the pulses depends on keeping a good rise flange, that we try to keep in the order of  a few ns as maximum (2V/ns). We manufacture also the reference drivers and they deliver to a 50 ohm load a 5Vpp TTL signal without any kind of over or underflow, and very fast r/f times, even when delivering 100mA pulses to the load. So we need a device that can allow the pass of this signal towards the 50 ohm.

The receiver switch IC must be supplied at 3.3V in spite of tolerate the 5V input pulses.

The duty cycle of this signal can go from a typical short dozens of ms to potentially 50% (0.5s), still this last possibility is not common, but it is "legal".

So, the switch subsystem has to cope with the 5V pulses even with long durations when supplied at 3.3V, have a low and relatively constant resistance, not to change a lot depending on the instantaneous input voltage, present small reactive impedance both on and off, fast speed to keep a fast pulse slew rate even when loaded with 50 ohms, be able to cope with a 50 ohm load during the duration of the pulses (that can last from ms's to 0.5s). This device is exactly the one that will typically deliver the signal to the switch input, that is why I describe it here. It is normally build using resistor-combined high speed TTL drivers.

As we know from the data sheet that the 3Q3305 switches (and rest of its family) rate 64mA máx. for survival, we will load each of the 3Q3305 switches with a maximum load of 100 ohm (50mA). There is another switch family with 128mA capability, but it seems to be highly distorting (and slower, so, could affect the signal flanges) compared with this relatively linear switch.

I ask TI advice to recommend this IC or not for this function, and to evaluate if a constant-like 50 mA load per switch can damage at short or long term the switch. Concerning the current, we have only the +/-64mA rating data; tests documented in the DS are only with a 50 ohm load. As to the power dissipation, as the Ron is not documented for Vi>2.4 5V and I0>30mA, I can't calculate the maximum instantaneous dissipation. I have figured out that either for Vi=0V / 0mA and 5V / 50 mA the Ron will not be very different and keep in the range below 10 ohm which means instantaneously 0.5W during the pulse duration. (In fact the Ron is another reason to split the input, so the Ron is smaller compared to the external load, and to get the two 100 ohm paths we will discount from the load the Ron average value). At 88º/W, the junction will go to 114º, which is high but allowable. Still, as I told, even if the signal duty cycle can in theory be 50%, normally is much shorter, from 20 to 100us.

a) Do you foresee any other problem that I can't for this application, of advice for a different IC for this function, or can you confirm that the idea is OK?

b) Would it be a good idea to pullup the load to +5V instead as typically to GND? The reason is just for the case that the switch dynamics and Ron could be better with 0V/-50mA than with 5V/50mA.

Thanks in advance


JM

  • Remark on the last paragraph, "b) Would it be a good idea to pullup the load to +5V instead as typically to GND? The reason is just for the case that the switch dynamics and Ron could be better with 0V/-50mA than with 5V/50mA.": I had this idea because the DS specifies Ron for Vi= 0V and positive current, and for Vi=2.4V and NEGATIVE current, and not the other way around, if I understand the IO sign (+=source from the out to the load, 0=sink from the source to the out), as in the test circuit the load is pulled down to GND and thus the maximum load current is positive and happens for VI high, not low...
  • I do believe it will work. The current rating is average constant current so spikes above the max are OK and constant 50ma should not damage the part.

    It will not matter whether you connect it to gnd or power.

  • Thank you Chris. I agree with you, that is why we chose it, but were not 100% sure. By the way there was a misprint in my first message, the DS tests are on a 500 ohm (not 50), that's why my question. I saw that older chips have a 50 ohm test.

    If it does not matter for the switch, the load will be better pulled down to GND, as it will dissipate much less (as the pulses are normally short).