GPS versus NTP timing- splitting hairs?

If not, can someone help me better understand how accurate digitizer sampling has to be?


I have been working on making a seismometer that meets a bunch of specs and one of the specs is that the timing accuracy of the instrument was to be sub millisecond. Another specs calls for a sampling rate of 100 samples per second (or one sample every 10 ms).

So here is my question, if the time between each sample is 10 ms what is gained by timing better that 1 ms?

I have another question: If the P wave velocities can range from 300 meters per second to more that 6000 meters per second depending on what it is traveling through, what is gained with a timing accuracy of less than 1 ms if you are not sure what the P wave is traveling through?

All comments welcome.


4 comment(s) on “GPS versus NTP timing- splitting hairs?

  1. Hello Angel,If you only want to pick P and S arrivals, you will not gained much with a 1ms and less timing precision.BUT, if you want the data to be used for waveform studies, signal correlation, tomography or anything else more “scientific”, you will definitely benefit from a sub-ms timing precision.And don’t think this stuffs can only be done with broadband, several scientists in France have made volcano tomography and signal correlation (which can be used to investigate velocity changes in time inside the volcano) with analog short-period, as soon as the timing is precise enought.And finally, if your timing precision is not enough, you will not be able to deal with long period data, because at some point you will have too much time slew.I think another thing very important, perhaps even more important than the timing precision is the precision and stability of the sampling rate. And usually, to achieve such precision and stability, you use a precise clock somewhere. For some scientific studies, if you have a 100Hz sampling rate, I would say you need it to be stable or precise at 0.05Hz or better (0.1Hz can already degrade your results).I hope this will help your designs.Regards. Jean-Marie

  2. Jean-Marie! When you’re a hammer everything looks like a nail. I tend to look at seismometers and seismic networks from the point of monitoring and early warning. And for the most part most of the networks in Central America are just that, many are paid for by Civil Defense. Your observations about the need for accurate timing of every single sample when it comes to doing science and engineering are correct. More so for building engineering and structural monitoring, even though relative time between instruments is probably ok. In a structure I would use one GPS and the rest put on the PTP time protocol. Most of this comment has to do with the fact that we are building a new seismometer/accelerometer instrument we call Darien that is very inexpensive and fast and easy to deploy in the field. No GPS antenna and no possible configuration in the field. It is quite inexpensive and meant to be deployed in large numbers. Even though in our test NTP can discipline the clock to <1ms the first comment we get from potential buyers is “it does not have a GPS”. Panama now has about 58 of these Darien and all of the automatic systems work much better and we get near real time shake maps at the same time. All paid for by Civil Defense. These bunch of Dariens cost less than two bb installations. Thanks for your comments and more comments very welcome. Saludos, Angel

  3. Angel, I think the question of choosing state of the art or just what’s necessary is also driven by the origin of your funds and the objectives.It seems that in most country, you have two kind of institutions dealing with earthquake location : university/science community and civil defense/governemental organizations. And from my little experience, I can tell that the funds are differents in origin and amount. Universities often must fight for a living to fund their instrumental projects, thus they need to have plenty of justifications to obtain them. And they obtain quite often a pretty big amount of money. So they need state of the art instruments that can do not only the location, but also science and many other thing for 20 year because money will not be attributed again within this time-frame. It’s what they call the project driven funding, the bigger the project is (in terms of funding), the higher chances you have to obtain the funding.On the other side, you have governemental agencies that have a simple goal (locate as soon as possible, assess the amount of shaking), less money, but easier to obtain (you just to frighten the governement 😉 once ). For that client, the simplest, the most cost effective is the right thing and I wouldn’t bother to have NTP timing. After all, this is simply logic, you have to adapt your products to each customer.

  4. A decade ago we did a simple study to compare NTP with GPS for a high school seismic network. The schools internet connections were notorious for being slow (or down) frequently, but even so NTP did pretty well, the results are here: 

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>