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.

SN74LVC1G14: nanosecond pulses and signal transmission concerns

Part Number: SN74LVC1G14
Other Parts Discussed in Thread: SN74AC14

Hello all,

I'm on my way to design a nanosecond pulse generator. I've reached the 10 ns time scale with 10 ns resolution. I believe that this is where transmission effects are to be considered.

I'd like to describe my configuration and raise some questions about it:

The output driver is a SN74LVC1G14 (Vcc = 5 V) with an added 25 Ohm series resistor as described in the FAQ .

This is a 100 ns pulse when connected to the 1 MOhm input of an oscilloscope. One can observe a fast edge with no ringing:

However, the graph shows an "overshoot" or "hump" that ranges from 0 to about 27 ns.

I believe this is due to reflections, i.e. reflections from the oscilloscope (open circuit, reflection with same amplitude) back to the driver and from the driver (short circuit, reflection with negative amplitude) to the oscilloscope. So after about 27 ns the reflected signal interferes with the original signal. As the reflected signal has a negative amplitude, it decreases the overall amplitude at the oscilloscope input. That would also explain the negative overshoot or "dip" from 100 ns to about 125 ns.

The cable length between driver and oscilloscope is 2.5 m of coaxial cable. A reflection back and forth will travel a distance of 5 m. I calculated that a signal will travel  about 25 ns in a coaxial cable of 5 m. That fits well with the observation, I think. Further, changing the cable length changes the width of the "hump".

Does that make sense?

To eliminate reflections, I added a 50 Ohm load resistor at the oscilloscope input. Then the signal looks great:

The reflections are gone, however, the amplitude is only 3 V and not 5 V any more.

How do I achieve full 5 V at the scope input?

Further, a SN74LVC1G14 that drives 75 Ohm with Vcc = 5V probably exceeds its absolute maximum rating (least from a DC point of view).

Will this driver be fine or are other drivers recommended?

Thanks a lot.

Dan

  • Hey Dan,

    When the input, transmission line, and termination are all perfectly matched, the transmission line seems invisible to the circuit, so you end up with this final circuit:

    In other words, you should be losing exactly half of your signal at the load if the match is perfect. You probably have an imbalance, but not enough to cause severe reflections.

    Just using the scope shot you've provided, I'd estimate your source only has about 33 ohms of resistance - which makes sense for an LVC device and an added 25 ohm series resistor (LVC typically has about 10 ohms of output resistance).  If you increase that 25 ohm to ~40 ohms, you will probably see a perfect match.

    As for 'how do you get a perfect signal at 5V at the receiver' -- you start with 10V. I know that isn't particularly helpful, especially considering that I don't have a 10V logic device to offer you that can do this. In RF circuits, we usually expect a certain amount of loss across transmission lines, and use gain at the other end to make up for it.

    Typically, in logic circuits we don't worry about this -- the goal is to get a digital signal from one end of the cable to the other, not to make the cleanest looking signal. Adding a series resistor can clean up the signal enough to make it work, and that's generally what we do. There are some additional tricks that can be done to clean up the signal further without imposing the DC shift -- for example, adding a series capacitor with your 50 ohm termination.

    If you're trying to do really high speed signalling, we generally work in (1) lower voltages and (2) with differential signals. Neither of which are really my area of expertise.

    -

    Your concern for output current is valid -- we can't guarantee operation of an LVC device beyond 50mA of current, so having a 75 ohm load could result in damage to the device. You can parallel two channels of a dual-channel device to spread out the current and get past that (up to 100mA limit instead of 50mA), but it's generally not something I would do.

    -

    It sounds like the purpose of your circuit is to just create a 1ns pulse as cleanly as possible.  Is there a reason that it has to be at 5V? What are you trying to achieve with this 1ns pulse?

  • Hi Emrys,

    I appreciate your response. It's very helpful and I like the way you explain things.

    Regarding my application, it's about laser pulse shaping. I need to provide the electrical signal to control the RF driver of an acousto-optic modulator. The RF driver is controlled using an input signal specified as "TTL". The RF driver output has rise/fall time of max. 3 ns. I want to generate a pulse in the range of 5 to 20 ns with 1 ns resolution to control the RF driver. So the purpose of my circuit is not to generate a perfect signal, just the digital signal to trigger the TTL input stage of the RF driver. 

    The TTL input is specified as having a 50 Ohm input impedance. I've yet to validate this. If it's 50 Ohm, I expect that the RF driver sees 3 V at the input (as shown on the 2nd scope shot above) and that should be sufficient to trigger the driver (as TTL defines 2 V as the min. level of a HIGH input signal). Provided, the LVC device operates properly in this setup. Would you suggest to use a different device?

    Thank you.

    Best regards,
    Dan

  • I've run into this same issue on a few occasions -- usually engineers are trying to get down to ~1ns pulse widths for time of flight laser systems. I could never guarantee that our logic devices can handle this reliably, but I know multiple companies out there are using logic devices as a solution.

    For a small, single channel solution, you probably can't beat LVC for drive strength. You could also try a dual channel (LVC2G14) version with the two channels in parallel to distribute the load and avoid too much current on one channel.

    You could also try the SN74AC14 as an option if the larger size isn't an issue. It has 6 channels and very strong output drivers.

    This would allow you to parallel channels to get stronger output drive strength and distribute the load current across multiple outputs to prevent reliability issues.

    There's a great video on youtube here that goes into details on creating really fast edges this way.