Post Go back to editing

Stand-alone regenerative latched comparators

Hello all,

My question is about regenerative latched comparators.

I'm not sure if I posted this question in the correct forum, but since I know that these circuit topologies are frequently used inside modern A/D converters, I thought there might be some experts here who can shine a light on this.

My question is not  related to a specific part number but I was wondering why regenerative latched comparators are not offered as discrete parts by any manufacturer.

As much as they are indispensable in modern ADC architectures , they would make for a killer component in many mixed signal applications where high speed analog comparing and low power are key requirements. The speed to power ratio could be vastly improved for any time-discrete design, where there is now no other alternative than to use a power hungry linear comparator. Such a component would be a perfect supplement to modern ultra low power logic families and with only 6 pins needed could be housed in similarly small packages.

Is there a technical reason why such components don't exist or is there just no market for them?

Just wondering......


Paul Beijer

  • Hi Harry,

    Thanks for your reply, I appreciate.

    You maybe right about market demand. That's for the marketeers to judge. Maybe availability could fuel demand.

    I only know that I ran into the problem several times and ended up using a micro-power linear comparator followed by a D-flipflop to realize the same function. That burdened me with an awkward trade-off between quiescent current and speed. A regenerative comparator would have a far better figure of merit in that respect, since it uses no static power, only dynamic. I figured, I may not be the only one confronted with this dilemma. 

    I've used home-made regenerative comparators in my designs that I built using matched mosfet arrays and that's ok for prototyping and proofs of principle. Compared to the comparator/flipflop solution the trade-off improved by at least an order of magnitude. But for actual production it's far too costly and bulky and package parasitics make it difficult to get the input offset voltage within an acceptable range without factory adjustments.

    As for your argument of system integration I don't quite agree. That reasoning holds for many small scale integrated parts like simple op-amps, comparators, logic etc. and ultimately even for discretes. They all can be integrated into more complex IC's. Still there is a substantial market for them because many designs require tweaks of the standard applications. Or for new applications or circuit topologies where no standard IC solution fits the bill, and one has to invent the wheel themselves.



  • I agree that at the extreme high-speed side of the application spectrum, it will probably be difficult to design a usefull standalone component because of parasitics that are much more dominant then what would be the case on-chip.

    But that's not necessarily the field of application for such a component.

    A typical ultra low power linear comparator that draws 1uA of quiescent current from its supply, has a ballpark propagation delay in the tens of microseconds. Very nice for low power battery operated or energy harvesting type of applications, but hardly usefull if you need to sample the input at speeds higher than a few tens of kHz.

    A typical 100uA Iq comparator  with a Tp of say 40ns would do the job, allowing for sample speeds well into the MHz range, but can become a significant battery drain.

    In such applications a regenerative comparator could give you the best of both worlds, and in this domain of relatively low speed they wouldn't need be so "bleeding edge" technology.



  • These are nice comparators but the speed/power ratio is not fundamentally better than many competitor products.