Choosing the right GPS/GNSS Disciplined Oscillator
We started this investigation seeking to identify the most common Disciplined Oscillator [DO] currently in use to support Public Safety Simulcast radio transmission. We then wanted to determine the most cost effective design, that would allow us to use the same basic GPS reference receiver and control circuits for several DO grades of performance. As you may know, the higher performance grade DO is only required for the simulcast transmission of high speed data, rather than a analog FM signal, probably by a factor of 10 to 1.
The main difference between the DO performance grades is the overall stability of the temperature compensated crystal oscillator [TCXO] or oven controlled crystal oscillator [OCXO], chosen for the application, usually measured over the temperature range of 0 to 70 degrees Centigrade. The most common TCXOs provide stability on the order of 10 parts per million [ppm] up to 1 part per billion [ppb], while the OCXOs provide much better performance in the range of 100 ppb to 0.01 ppb, using small heaters to provide a uniform high temperature. This range is also expressed in scientific notation like 1x10-9 or 1E-9 for 1ppb or 0.001ppm.
For most lab or bench calibration uses, the older and simpler GPSDO will work fine providing a 10 MHz. reference output, as there is rarely a need for 1PPS on the bench. However, when using the GPSDO 10 MHz. output to synchronize transmitters simulcasting, the GPS 1 Pulse Per Second [1PPS] is also required to synchronize the information being modulated on the carrier. Only the latest GPS DO designs offer a continuous & stable 1PPS output during the occasional times the GPS is not receiving sufficient satellite signal to remain in synchronization, otherwise known as the “holdover” period. Depending upon the placement & quality of the GPS antenna, coaxial cable, and local interference issues, the typical holdover period is random, sporadic and could last for seconds to minutes daily, and for hours on an annual basis. In order to maintain at least 99.999% reliability [5 minutes/year down time], we seek the latest DO designs to provide an ultra-stable reference, on all outputs at all times.
In our search for the perfect DO, we find the GPS [or GNSS] synchronized version is the most available, since the satellite reference is cesium based & works generally anywhere on earth. However, there are great differences between the various GPS receiver designs, which makes some better suited for reliably extracting timing data. Other reference receiver systems such as eLORAN are also available, but are more costly, in short supply, and are being produced mainly as a fall-back should the existing GPS system suffer failure or interference.
Nearly all GPS DOs provide frequency and time stability of 1E-12 or 0.001ppb or 0.01 Hz at 900 MHz, when the GPS receiver is stable & locked on one or more satellites. This also applies to the 1PPS output, typically used to synchronize events. However, if your application is more critical like ours, where it must continue to operate, even during holdover, then you will require a more advanced model of DO that also continues the 1PPS output with great accuracy. To create the 1PPS, this advanced model typically divides the reference frequency down and synchronizes with the 1PPS received from the internal GPS receiver. The timing accuracy is fixed under 50 ns, and will remain stable for that brand & model of DO.
During holdover, the 1PPS stability will be most affected by temperature changes surrounding the DO. Obviously, the native compensation of the TCXO or OCXO is critical, but some manufacturers improve the native stability by monitoring the external temperature of the DO while in GPS lock, and use that data to compensate for external temperature changes, thereby improving overall stability while in holdover.
Your final consideration is whether the reference frequency [10 MHz] is phase coherent with the 1PPS output. If you are simulcasting voice or data, it is imperative that there is no phase shifts occurring during your transmission. Otherwise in signal overlap areas, phase distortion and cancellation could occur, causing the information to be garbled. During holdover, some small frequency & phase shift will occur, as it cannot be avoided, without some form of secondary backup to GPS signal. [Read our other article on Using Other References for Simulcast]
With all of these factors in mind, the old adage that “you get what you pay for” may be partially true. Below is a chart with a wide variety of GPSDO products ranging from $140 to $5,000, each accomplishing nearly the same thing. We will examine some of the product offerings to discover hidden value in some and little extra value in others.
|1PPS Holdover||5us||<50us||1.5us||20us||8 us||<25us||1us||10us|
|Power||5.5v/2.8w||3.3v/0.16w||5v / 5w||3+12v/5w||24v / 8w||120vac/40w||12v / 11w||12v / 3w|
|Temp Rating||-40 +85c||-10 +70c||-20 +75c||-20 +60c||-20 +75c||-30 +65c||-20 +60c||-10 +70c|
|Notes||External OCXO Option||Uses Kalman Filtering||Requires two voltages||Unstated Specs Typical||Public Safety Standard||Rubidium Reference||Great Value|
Before we examine some of the products listed, we should consider environmental and timing issues, and see how they will affect your choice.
First, consider the published temperature specification in holdover has little meaning in the real world unless you intend installing your equipment in an outdoor cabinet. In most cases, your equipment will be installed in an air conditioned building, with only maybe a 5 degree temperature swing. Therefore, this makes this specification almost useless. By careful placement of the DO in a enclosure without significant air flow, will dramatically improve the holdover stability in both the reference frequency and the 1PPS. Keep in mind, other components in the DO like the DAC that controls the heating element in the OCXO will affect the temperature stability also.
Second, the time usually specified relating to holdover is 24 hours. This extended period of time is unrealistic except for critical applications such as Public Safety communication. A more realistic approach would be to test the holdover accuracy over a scale of 5 minutes or 15 minutes, while controlling the temperature withing the expected range of operation. We encourage you to test prospective products in these real world conditions before committing any real money. We have not taken the opportunity to do this currently, but may in the future and publish the results.
Now with all this in mind, let us examine some of the products listed.
Spectracom SecureSync at the top of the price range is used heavily by Public Safety communications systems that simulcast transmissions for police and fire. This is an expensive unit, but with a distribution panel, it could feed 20 or more transmitters at the same location, thereby lowering the cost per transmitter. It is complete and ready to use. This unit is enclosed and is rack mountable. This unit is our standard for comparison.
Spectratime SRO-100 is the best product regarding overall stability. What makes it so good is the Rubidium oscillator instead of a quartz crystal. This is a really good choice if your are using a weatherproof cabinet outdoors. It will really remain stable in holdover because of the oscillator, but is not the most economical for an indoor installation. The product in enclosed in a metal case with connectors, designed for direct mounting in vendors equipment. Excellent unit.
Spectratime GXClok-500 in the next in line regarding overall stability. It is a board level product with a OCXO designed to be installed inside the vendors equipment. The price is reasonable and the power consumption is only 3 watts. This is my second most favorite product.
Trimble Thunderbolt E is the next in line and is the latest offering by Trimble in this line of DOs. However, I found the specifications are not clearly presented or vague in nature on the web site. The older Thunderbolt model is widely available used and users have mixed reviews as to the stability. The unit is self contained in a metal case and known to be very durable. The price is reasonable.
Brandywine MGPSDO is a nice little module but requires two voltages to operate. Nothing really stands out beyond the excellent OCXO.
Furuno GF-8704 is a board level assembly but is not a stand out product. It was difficult to get information from the manufacturer and it appears not to be a stock item in the USA.
IQD IQCM-110 is one of a series of 4 modules made by this oscillator manufacturer. The price is somewhat high and the holdover stability on the 1PPS is above average. The manufacturer has great DO technical information on their web site.
Jackson LTE-Lite is my favorite product. This is the most versatile DO found. While in lock, it is as stable as the best DO available. It only draws 0.16 watts, which is a fraction of most other DOs draw. It is a small board level product that can be mounted on a larger board and enclosed, to further stabilize the unit. It is designed to easily add a higher specification OCXO or TCXO to meet your specific holdover requirements, up to 2E-11 or 0.02ppb. A compatible high performance voltage controlled OCXO or TCXO can be purchased for under $100 in small quantities. This one product can satisfy a wide range of applications, by simply adding a better OCXO when needed. We have selected this unit as the core of our Open-Source Disciplined Oscillator project, and it is only $140 in single units.
As a final note, we also looked at a more basic design starting with the u-blox LEA‐M8F GNSS module, and adding the necessary voltage controlled OCXO or TCXO, micro-controller and Digital to Analog Converter to achieve something equal to the Jackson LTE-Lite. The cost for the LEA-M8F was $57 in lots of 250, or you could buy an evaluation board from csgshop.com for $190 for one. The EVK-M8F evaluation kit from u-blox is $349 each. In the end, we simply felt the LTE-Lite was a far better solution at a reasonable price.
We hope to have a kit available by the end of 3th Qtr 2018 at OpenLMR.com which includes a extruded aluminum case, carrier board with voltage regulator, buffer amplifiers, and Si51218 clock chip providing at least 3 additional outputs.
Edited by Fred W Daniel, May 24, 2018