[time-nuts] Discussion of precise time and frequency measurement

M. Simon msimon6808 at yahoo.com
Fri Dec 7 05:13:20 UTC 2012

I'm doing some things of that order with the LPC1114. The board for the project will be back from fab in a few days. If it works out well I have a frequency/period  counter designed around that chip. About $20 in parts. Not counting the TCVCXO. Add in a case, I/O and power supply plus a display and you are still well under $100. The counter is probably 3 months away due to software. 

What I lack is a front end design good to 200 MHz. You can buy counter chips guaranteed to 150 MHz for well under a buck. And there is a 74PYT74 (PYT = pick your technology - 74LVC1G74 is one) that guarantees 200 MHz (@5V - 280MHz typ @3.3V) . But I like having the enable part of the counter chip. It simplifies things. BTW the 150 MHz chip "typicals" at 200 MHz. I have done some things to speed it up - like lower the capacitive load with 10K resistors. Only Q3 of  the input counter needs direct connection to the input of the next counter. So the capacitive load of Q0, Q1, and Q2 will be well under the 15 pF (or 50pF) where the testing is done. Q3 should be lower than 15 pF.  All that will gain some MHz. Also running the chips a little hot: 3.4 V

The LPC1114 has two 32 bit counters. I add 4 or 8 bits outboard to get from about a 20 MHz max count rate to >150MHz. I'm using the second '1114 counter as a 32 bit match counter to make up the time intervals. i.e. Match on "2" to start the count. Match on "50,000,0002" to stop the count. I use the 200MHz 74xxx74 as an RS F/F to control the counter enable. I'm running the '1114 chip at 50MHz from a clock that originates from a TCVCXO - 40MHz /4 . This also allows the use of external 10MHz sources. 

And of course there is capture for period. There is a one cycle delay (20ns) for edge recognition in the counter mode. This may or may not be a problem. 

The counter will have a board available for those of you into building.  But that is down the road a ways. 

Where this could come in really handy is if you need a bunch of counters for a project. The counter will have a Serial interface that can be accessed with an RS-232 adapter or a USB adapter. Plus I2C. The poor man's IEEE488 bus. And a USB to I2C controller board for the PC side. All of the above is currently in development (boards on hand or on order) with some of the projects already on the 2nd turn so development is well along. This is not a pipe dream. But it is taking longer and costing more than I'd expected. (Doesn't It Always?)

I also bring out PCLK to a test point so measurements of the chip's PLL could be done by those interested. 



Date: Thu, 6 Dec 2012 18:55:13 -0500
From: Bob Camp <lists at rtty.us>
To: Discussion of precise time and frequency measurement
    <time-nuts at febo.com>
Subject: Re: [time-nuts] GPSDO Alternatives
Message-ID: <C851493C-C77B-41B9-A7DF-3D3F89C91B9E at rtty.us>
Content-Type: text/plain; charset=us-ascii


To be useful, you need an input capture that:

1) Runs at a fast enough clock (1 GHz would be nice)
2) Has enough bits to get to 1 pps (say 32 bits)
3) Has a built in period set, so the hardware works without a lot of silly stuff

 you find parts that will do some of the above, but not all. 16 bit 
captures running off of a "few MHz" clock are pretty common. Some (but 
not all) ARM's have 32 bit captures that run off of 10's of MHz clocks 
and have the ability to set the period.



Engineering is the art of making what you want from what you can get at a profit.

More information about the time-nuts mailing list