Latest Posts
Discipline Your Gear With A GPS Reference Clock
The Leo Bodnar GPS-Disciplined Reference Clock, which has found a place in a number of hobbyists’ radio shacks.
Reviewed by: Andrew Barron
A GPS reference clock is a super-accurate, super-stable, frequency source. It is used primarily to keep a radio receiver (and/or transmitter) exactly on frequency. Cellular base stations use them to make sure that the calls they put out to mobile phones are exactly synchronised. Each cell site transmits the data at exactly the same time.
This is important when your phone call is transferred from one cell site to another as you walk or drive along.
Several SDR radios, and a few amateur radio transceivers, now have the capability of connecting an external frequency reference, in order to make their frequency readout accurate and to eliminate frequency drift in the radio.
Frequency drift is usually caused by temperature variance in the radio. Other factors, such as the DC supply rail voltage changing during transmitting periods, can also have an effect.
The IC-R6800 receiver, IC-7610 and IC-7851 transceivers, SDRplay RSP2 and RSPduo, FlexRadio Signature series, and the Expert Colibri receiver, for example, have the ability to lock to an external 10 MHz reference clock.
Frequency drift is not generally a problem at short wave frequencies, or at VHF or UHF frequencies if you are listening to FM signals. But it becomes a big problem at microwave frequencies or when monitors are receiving very narrowband digital mode signals, for instance, data signals from a geostationary satellite.
Essential Calibration
If your radio doesn’t have a reference signal input, you can use a GPS reference clock as a signal generator and calibrate your receiver against it. The Leo Bodnar unit is especially good for this, as it can generate a stable, GPS-locked signal on virtually any frequency between 450Hz and 800MHz.
However, you should be aware that the output level is much too high for most receivers. You will need to place at least 80dB of attenuation between the clock and the receiver.
Also, the GPS reference clock outputs a square wave, so the output is rich in harmonics. Particularly odd harmonics, of course.
This is not as big a problem as I suspected. In fact, it is rather useful because you can use the frequency standard at much higher frequencies by taking advantage of the harmonics.
I tested the frequency accuracy on my radio at 1296.2MHz by receiving the second harmonic of a test signal on 648.100MHz. It was only 13Hz off frequency, which is outstanding! I checked it again a couple of days later, and it had changed. The frequency error had become zero Hertz.
Note that this was not a SpectrumLab test; I was just using the panadapter, so it may be off by a few Hz.
GPS reference clocks can also be used to calibrate test equipment and to govern the internal clocks in units like frequency counters, RF signal generators, and vector network analysers.
Stability Matters
The higher in frequency a radio is capable of receiving (or transmitting on), the more having good frequency stability matters. This is because, as you multiply the oscillator (or clock) frequency up to very high frequencies, any frequency error is also multiplied.
The problem matters to conventional radios, where the oscillator is mixed with the incoming radio signal to create an I.F. (intermediate frequency). It also applies to SDRs, where high clock rates are required for wide bandwidth receivers. SDRs working at UHF and SHF frequencies (above about 500MHz) use a conventional mixing and oscillator stage, just like superheterodyne receivers, so the same frequency stability problem arises.
Frequency drift at VHF and UHF frequencies is of little consequence for FM signals, and the error is tolerable for single sideband (SSB). You might have to tune the receiver a little, but it is not a big problem.
The problems begin when you are using very narrow bandwidth transmissions or digital modes that are not frequency-tolerant. Some narrowband digital modes can’t cope with more than a few Hertz of frequency variation.
Oscillator Stability
An RF oscillator creates a sine wave signal on a wanted frequency. Sometimes the frequency is variable, in which case it is called a VFO (variable frequency oscillator).
Oscillators are used with analogue signals, such as you might find in a superheterodyne receiver. A clock creates a square wave signal. It may have a 50% duty cycle, or it might be a series of narrow pulses. Clocks are used to time data transitions in digital circuits, such as clocking the ADC in an SDR.
Oscillator stability is the amount that the oscillator moves away from the wanted frequency, due to temperature or voltage variations. It is commonly referred to as frequency drift.
Clock stability is the same thing for clocks. Short-term random changes in clock frequency are referred to as ‘jitter.’ Clock signals suffer from both drift and jitter.
PPM – PPB – PPT
Oscillator stability is commonly measured in ppm (parts per million), or sometimes in ppb (parts per billion), or even in ppt (parts per trillion). If an oscillator or clock has a stability of plus or minus 1ppm, it means that for every MHz (one million Hertz) of oscillator frequency, the oscillator should stay on its nominated frequency plus or minus one Hertz. For example, if an oscillator with a ±1 ppm stability happens to be on a frequency of 250MHz we can expect the error to be less than ±250Hz. Its frequency should always be within the range from 25,000,250Hz to 24,999,750Hz.
It all gets a bit confusing when we start talking about super-stable oscillators and clocks. Generally, after about 0.1ppm, people switch over to scientific notation. A clock or oscillator with a stability of 1ppm is said to have a stability of 1 x 10-6. It means that the error is one-millionth of the oscillator frequency. If the oscillator is extremely stable, for example, 1ppb (0.001 ppm) it will have a stability of 1 x 10-9. Our 250 MHz signal will now be within ±0.25 Hz. Some GPS referenced clocks or oscillators can attain 1ppt or 1 x 10-12 stability.
That is an accuracy of ±0.001 Hz at an oscillator or receiver frequency of 1GHz.
The oscillators in our receivers are getting better and better. Valve (vacuum tube) based receivers used to get very hot, and they tended to drift a lot until the temperature stabilised. In the 70s and 80s, radios with PLL (phase-locked loop) frequency synthesisers would typically have oscillators with quite poor stability, sometimes 20ppm or worse.
Many FM mobiles and handhelds still have oscillators like that.
FM receivers are very tolerant of frequency drift. Most radio manuals in those days didn’t even specify the oscillator stability. My Yaesu FT-847 from 1998 has an oscillator with a stability of 2ppm. A good modern radio will have an oscillator with a stability of ±0.5ppm. To get better than ±0.1ppm, you will need to use some kind of super- accurate signal source – such as a GPS-referenced or atomic-referenced clock.
With oscillator stability being measured in parts per million, 1ppm is thus a frequency variation of 1Hz per MHz. If the main oscillator (or clock) is off-frequency, all of the higher frequencies that are generated from that master clock or oscillator will be off by the same ratio.
GPS reference clocks and GPSDOs (GPS-disciplined oscillators) use very accurate time pulses received from GPS satellites to control the drift of a TXCO (temperature-controlled crystal oscillator). A GPS reference clock can easily achieve frequency stability better than 1ppm (1x10-9), which is about 500 times better than a 0.5 ppm TXCO. It could hold a receiver or transceiver to less than ±1Hz of error, even on the 1240MHz band.
The stability of some GPS reference clocks can approach 1ppt (1x10-12) which is about 500,000 times better than an uncorrected 0.5 ppm TXCO.
Radios receiving the short wave bands don’t need super-stable oscillators. A ±0.5ppm oscillator will hold the radio to plus or minus 15Hz at a receive frequency of 30MHz. Such a small amount of drift is unlikely to cause any problems. However, at a receive frequency of 450MHz, the possible error becomes ±225Hz. At 1GHz, the error becomes ±500Hz.
This isn’t a fault of the radio or poor design. An oscillator with a 0.5 ppm is very stable indeed. The problem arises when we need it to be even better.
Impressive Stability at a Good Price.
Super-stable clocks and oscillators used to be expensive. The choices were commercial GPS clocks or ‘atomic’ clocks such as Rubidium clocks or oscillators. If money was no object, you could buy a Caesium-based atomic clock.
The GPS navigation system relies on being able to receive timing signals from three or more GPS satellites and triangulating your position by comparing the delay between the times that each signal is received. For the navigation system to work, the time signal from each satellite must be extremely accurate. GPS satellites have both Caesium and
Rubidium clocks onboard. They are so accurate that they have to be corrected for time dilation due to the satellite’s orbital speed. Nowadays, manufacturers can take advantage of these extremely accurate time signals from space and use them to control the frequency drift of much cheaper TXCOs, at a much lower cost.
The Leo Bodnar GPS Reference Clock
I had seen advertisements for this neat little device (Fig. 1) before. However, I did not buy one as yet, because I decided that I did not need my HF radios to be synchronised to a super-accurate frequency reference. I knew it was a pretty interesting piece of kit, but the main oscillators in my HF radios are easily stable enough for my needs.
Pictures - Andrew Barron
Fig.1: The Leo Bodnar GPS Reference Clock unit.
But then I purchased an Icom IC-9700 transceiver which works up to 1300MHz.
The radio has a 10MHz reference signal input on the rear panel, and Icom has just released a firmware update, to allow the main oscillator in the radio to be governed by a device just like the Leo Bodnar GPS reference clock.
While the 0.5ppm oscillator in the IC-9700 performs well within the specification – which is perfectly adequate for most users – there is the now an option to govern the internal oscillator and vastly improve the frequency stability.
This level of control reduces the frequency drift to around ±1Hz on the 23cm band, rather than the ±680Hz that can be expected from an un-governed TXCO.
You need very good frequency stability if you are using very narrow band transmissions or you want to use the radio as an I.F. stage for a microwave band receiver or transmitter.
I bought one mostly because I wanted to play with it. Just because it is pretty cool.
Using the Frequency Reference
The unit is available directly from this website:
However, I bought mine from SDR-Kits:
Their version came complete with a USB cable, a GPS antenna, and a handy carry box. There are two models, one has a single output to an SMA connector, the other can output the same frequency on two BNC ports (Fig. 2) or on two diverse frequencies. I purchased the more expensive two-port model, but I’m not sure that I needed to.
Fig. 2: The two-ports on the Leo Bodnar GPS Reference Clock.
Both models have an identical frequency range and performance. You can power the unit from a 5-14V DC supply or from the USB port. Once you have programmed a frequency, or a pair of frequencies, using the free software, you can disconnect the USB cable. The clock will remember the frequency setting, even if power is temporarily disconnected.
After the frequency has been loaded, the unit can be used completely stand-alone. If the GPS antenna is disconnected, or no GPS satellites are detected, the clock will continue to run with no interruption, although it will no longer have the extreme stability. When GPS lock is recovered, the clock will return to high-stability operation, with no interruption or frequency glitch. It will just pull itself back onto frequency over a second or two.
After a GPS lock has been made, the frequency stability is excellent.
In tests performed and published by W.J. Riley of Hamilton Technical Services, the frequency error of the tested unit never exceeded plus or minus 2.5 x 10-9 over a 15 hour period, and it achieved a long term average stability better than 0.001 ppb (1x10-11). This equates to a maximum error of less than 0.008Hz at an operating frequency of 800MHz.
I didn’t realise that the choice of the second frequency is very restricted. Output 2 is derived from the same multiplier and divider chain as output one. The software has a dropdown box that allows you to select some of the frequencies, but it doesn’t always present you with all of the options. You can manually enter an even integer into the NC2_LS text box to input your own multiplier.
Some of the options end up being rather odd frequencies and they are not necessarily multiples of one Hertz.
For example, with output 1 set for exactly 10 MHz, one of the output 2 options is 11428571.4285714 Hz. The two-output version has BNC connectors, rather than SMA, which makes connecting test cords easier.
I suggest that if you plan to use the unit to semi-permanently GPS-discipline a test instrument, receiver or transceiver, it is better to buy the single output model. If you are looking to use the clock on an occasional basis as a stable test signal, the dual-port model could be useful.
Using the Software
The free-to-download software (Fig. 3) works well but is a little quirky. It could definitely use some improvement, especially when it comes to the selection of the second output frequency. Entering the first frequency is straightforward. You just enter the frequency that you want, and press Find. After the program has worked out the settings, you can press Update and the PLL will lock to the wanted frequency.
Fig. 3: Software screenshot. Output one is at 10MHz; output two is at 15MHz.
Selecting the second frequency is not intuitive. You use the dropdown box to select one of the possible frequency options and then press the Update button. However, the dropdown list does not include all of the possible frequencies.
You can manually enter the GPS reference frequency, multiplier, and divider settings, but there is no easy way to work out the required multiplier and divider settings for a wanted pair of frequencies.
If you already have the multiplier and divider settings, it is fairly easy to work out the resulting output frequencies. The output 1 frequency can be determined from GPS ref / N31 * N2_HS * N2_LS / N1_HS / NC1_LS.
The output 2 frequency is GPS ref / N31 * N2_HS * N2_LS / N1_HS / NC2_LS.
Note that NC2_LS and LC1_HS must be even numbers.
The clock outputs have very low phase noise, better than -100dBc at 100Hz offset and -143dBc at 1kHz.
There are four output levels named as output drive strength and stated in mA. The output voltage is always at CMOS 3.3 Volt level.
According to the website, the 8 mA setting has a level of about +7.7 dBm. The 10 MHz reference input on the Icom IC-9700 radio is expecting a level of around -10 dBm.
Therefore, you should connect a 15 dB attenuator between the GPS unit and the Icom radio. The 16 mA step outputs at about +11.4 dBm, which would be OK with a 20dB attenuator inline.
Using the GPS Reference Clock as a Signal Generator
You can apply the square wave output of the GPS reference clock – via 80dB of attenuation (cf above) – to the input of a receiver and use the device as a frequency-accurate signal generator. At HF frequencies, where the second and third harmonic are within the range of the receiver, it is a good idea to remove the harmonics and turn the output into a sine wave by passing the signal through a bandpass filter.
Mini-Kits have some nice 7th-order Chebychev bandpass filter kits for the HF amateur radio bands.
Each receiver in the Icom IC-9700 has an individual front end bandpass filter so an external filter is unnecessary for that radio.
Leo Bodnar (Electronics) LTD
Units 7 - 8 New Rookery Farm
Silverstone, Northamptonshire
NN12 8UP
Tel: 01327 850666
E-mail: [email protected]
References:
https://www.minikits.com.au/electronic-kits/filter-kits/hf-bandpass-filters
W.J. Riley Review: https://tinyurl.com/y4fgs8oz
This article was featured in the August 2019 issue of Radio User.