Frequency Stability & Accuracy for VHF+ WSPR
I'm starting this thread to create a place to discuss issues related to getting amateur equipment stable enough to successfully decode and generate WSPR signals as well as to discuss the value and means for having high absolute frequency accuracy as well.
Stability
Almost all of us who have put our stations on 2 m (or above) WSPR have had to deal with getting sufficient stability to allow good operation. What would pass as acceptable drift on 20 m is ten times worth and unusable by 2 m. By 1296 MHz, the problem has escalated another order of magnitude.
Generally frequency stability can be measured a number of ways, drift in Hz or ppm per second, minute, hour, day, year etc. For our purposes the critical time frame is the 2 minute WSPR window. Unless end-end signals show up with frequency change less than 3-4 Hz within two minutes, the station may not be usable for WSPR.
At 144 MHz, 1 Hz drift amounts to 1 Hz/144e6 = which is 7 parts/billion and less than a millionth of one percent. This is a much smaller value than the specifications on even the best high-stability frequency references for Icom, Yaesu or Kenwood transceivers. Normally these specifications are not very detailed and may not tell the radio owner much about the stability in an interval as short as 2 minutes but they do give an indication of the problem.
We do have stations on 2 m WSPR that are successfully operating both Icom and Yaesu hardware with the high-stability timebase but it's not guaranteed that this will work. Some radios do pretty well and some are marginal or unusable. In addition to temperature drift, particularly due to the entire RF deck warming/cooling during transmit/receive, there may be voltage dependence of the TCXO. The Icom unit has a characteristic “chirp” during the first 10-20 seconds of transmit when used in IC746, IC910 and perhaps other radios. Cutting the 14 VDC connection to the TCXO and inserting a fixed 10V regulator can eliminate this problem.
Even with the TCXO voltage dependency solved, these commercial high-stability options tend to drift around over time. It's not uncommon to see a 2 m WSPR station drift as much as 50 Hz as the hamshack warms and cools during the day/night. Although this type of long-term drift doesn't hurt WSPR operation, there are sometimes shorter term ones left, from chassis heating usually, that still limit station effectiveness.
Some stations have added Styrofoam insulation, in the form of sheets or “peanuts” around a TCXO in order to reduce these short term variations. In at least one instance, an entire TCXO was outboarded, wrapped in plastic insulation and leads brought back into the radio.
Accuracy
Frequency accuracy isn't generally as critical to VHF+ WSPR operation since, as long as an accuracy problem is detected, it can usually be manually or automatically corrected. This is the function of the A/B settings in the stock WSPR code and described in the WSPR documentation. The goal of these corrections is just to “detune” the radio to correct for its internal frequency errors. These errors may be of two kinds, offset or scale.
The A term serves to bump the frequency on every band a constant amount to correct for a radio system local oscillator which has constant error versus output frequency or band. For older radios, this could include BFOs, carrier oscillators and even 2nd or 3rd conversion oscillators.
Newer radios and multi-band/multi-function radios increasingly use a single master oscillator which, with the help of DDS, PLLs, multiplying, dividing completely determine the output frequency,and perhaps even output phase. Absolute frequency error in this oscillator therefore contributes an error in the output directly related to it's accuracy, measured in ppm, ppb or percent. A master oscillator with a 1 ppm (part per million) error will produce a hamband output signal with 1 ppm error. On 2 m this would mean 144 Hz of dial error. The B term in the WSPR frequency correction code is used to calculate the amount of error for a given band and to “detune” the radio to compensate. Radios with single master oscillators that control everything will only need a B value entered since there are no offset oscillators the A value will always be zero.
Although accuracy errors don't usually prevent 2 M WSPR operation, they can spoil a radio for gathering significant propagation information. Actually this is true of HF as well. If it is known in advance that all frequency errors, within sufficient resolution, are due to the propagation path rather than the hardware at one or both ends, we start to learn more about radio science. At HF, for example, it is possible to measure movement of the ionosphere as the bands open and close. The change in height causes a changing path length which results in measurable frequency offset on a WSPR spot. Particularly on 20 m – 10 m, one or even two Hz of offset can be directly observed.
At 2 m and up, we begin to see some even more interesting effects if we have frequency accuracy. For one, it is much easier to identify individual stations on a busy waterfall where not all stations are decoding. But we even see situations where we get multiple spots, a “main” one and a Doppler shifted one, usually due to aircraft motion changing the total path length between WSPR stations. If this change is constant enough over two minutes, and it occasionally is, then we can successfully decode a spot that came to the receiver only by aircraft scatter.
In this thread, perhaps we can begin to discuss our experiences and methods for improving both frequency stability and frequency activity with the variety of radios and transverters we are using. This could help other stations. In addition I hope to add some information about the external GPS disciplined phaselock system that some of us are working on and hope to make available before too long.
Glenn n6gn cm88ok
7 October 2012
------------------
The crystal is likely mounted on thin springy wires which are spot-welded inside to the lead frame. However, the quartz is probably not in an evacuated environment so there are two sources for eating, the leads (conduction) and the air inside the can (convection). The question is, which one is biggest and wins the battle.
If the board, chassis etc of the radio ran warmer than ambient but didn't change and if the ambient air temperature was constant then the crystal would eventually settle at some temperature between them. If the air inside the crystal is by far the dominant (lowest thermal resistance) path then blowing constant temperature air on it will result in stable crystal temperature. OTOH, if the leads provide the lowest thermal resistance, minimizing the changes of the radio may result in lower delta-T at the crystal.
My first guess is that coupling through the leads and case is the dominant source of heat flow, but it may not be. This would indicate maximum air through the PA. If it isn't, maximum constant-temperature air on the crystal case may work better.
Both KP4MD and K6PZB saw significant improvement in stability by cooling the transmitter better and with the addition of cotton to reduce coupling to ambient air. You can try it both ways to see what you find works better.
Glenn n6gnThe intent is to keep the crystal at room temperature.
The room temperature is relatively constant at about 72 degrees and rarely changes quickly.
A +- 4 Hz drift during a 110 second transmission can prevent a decode.
Otherwise, +- 100 Hz is OK.
I hope to set this up in a few days and will report results.On 06/27/2012 03:26 PM, Carol F. Milazzo, KP4MD/W6 wrote:Chuck, N0SSC stuffed his oscillator compartment with cotton to provide thermal stability. As Glenn suggests, the external fan will help stabilize the overall temperature of the radio during transmit-receive cycles. Blowing external air directly onto the oscillator circuit without insulation may render it more sensitive to short term changes in room temperature. What do you think, Glenn?
--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To post to this group, send email to 2-mete...@googlegroups.com.
To unsubscribe from this group, send email to 2-meter-wspr...@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.
To post to this group, send email to 2-mete...@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
The preliminary data reveal a frequency variation of 50 Hz that correlates with the 75-85° F XV144 case temperature range measured over each 24 hour period. The frequency drift persists at a mean -3 Hz over each 2 minute WSPR transmission at 5 watts. This may be the best frequency stability that the Elecraft XV144 transverter can achieve with its internal 116 MHz local oscillator. The measuring station N6GN has a GPS disciplined frequency reference.
While watching the WSPR waterfall tonight I noticed many KI6STW traces that I was not decoding. I realized that most successful KI6STW decodes occur immediately after one of my own transmissions. Curiously, both KI6STW and I have a mean -3 frequency drift during our WSPR transmissions, but my decodes of KI6STW show a drift that varies between -1 and +1 Hz, apparently my rebound frequency rise after each transmission period the KI6STW's frequency drop.
--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.
To post to this group, send email to 2-mete...@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
> --
> You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.
> To post to this group, send email to 2-mete...@googlegroups.com.
> Visit this group at http://groups.google.com/group/2-meter-wspr?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
In service.
That means that it found the difference between the 1 PPS GPS signal and the counted OCXO frequency greater than an internally set threshold. The usual reason for this is loss of satellite fix. Unplugging the active antenna or shielding the antenna might do it. WW6D saw this occasionally until he moved the antenna to a better location.
Another reason, less likely, is that the OCXO itself took a large jump. This does happen occasionally but with reduced likelihood the longer the OCXO has been continuously running.
Glenn n6gn
On 06/21/2013 08:12 AM, Steven Hess wrote:
My GPS10V just went briefly into the a red flashing OCXO LED mode and returned to Green. This is the first time I have noted this so I thought I should mention it.
Is this expected behavior?
Steven
--
____________
Apply appropriate technology. Use what works without prejudice.
Steven L Hess ARS KC6KGE DM05gd22
Google Voice 661 769 6201 +SMS
openSUSE Linux 12.2 KDE 4.8.5
Known as FlameBait and The Sock Puppet of Doom.
--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr+unsubscribe@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr.
--
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr+unsubscribe@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr.
23 July 2013 - The Flex 1500 exciter at KP4MD was connected to the GPS disciplined frequency reference at 0330 UTC on 19 July 2013. The cumulative spot reports of GPSDO controlled 144 MHz WSPR study group stations since that time show resolution of the diurnal frequency variation that was previously observed while the Flex 1500 used its internal TCXO frequency reference. (N6KOG station frequency is not GPS disciplined).
Frequency Stability & Accuracy for VHF+ WSPR
I'm starting this thread to create a place to discuss issues related to getting amateur equipment stable enough to successfully decode and generate WSPR signals as well as to discuss the value and means for having high absolute frequency accuracy as well.
Stability
Almost all of us who have put our stations on 2 m (or above) WSPR have had to deal with getting sufficient stability to allow good operation. What would pass as acceptable drift on 20 m is ten times worth and unusable by 2 m. By 1296 MHz, the problem has escalated another order of magnitude.
Generally frequency stability can be measured a number of ways, drift in Hz or ppm per second, minute, hour, day, year etc. For our purposes the critical time frame is the 2 minute WSPR window. Unless end-end signals show up with frequency change less than 3-4 Hz within two minutes, the station may not be usable for WSPR.
At 144 MHz, 1 Hz drift amounts to 1 Hz/144e6 = which is 7 parts/billion and less than a millionth of one percent. This is a much smaller value than the specifications on even the best high-stability frequency references for Icom, Yaesu or Kenwood transceivers. Normally these specifications are not very detailed and may not tell the radio owner much about the stability in an interval as short as 2 minutes but they do give an indication of the problem.
We do have stations on 2 m WSPR that are successfully operating both Icom and Yaesu hardware with the high-stability timebase but it's not guaranteed that this will work. Some radios do pretty well and some are marginal or unusable. In addition to temperature drift, particularly due to the entire RF deck warming/cooling during transmit/receive, there may be voltage dependence of the TCXO. The Icom unit has a characteristic “chirp” during the first 10-20 seconds of transmit when used in IC746, IC910 and perhaps other radios. Cutting the 14 VDC connection to the TCXO and inserting a fixed 10V regulator can eliminate this problem.
Even with the TCXO voltage dependency solved, these commercial high-stability options tend to drift around over time. It's not uncommon to see a 2 m WSPR station drift as much as 50 Hz as the hamshack warms and cools during the day/night. Although this type of long-term drift doesn't hurt WSPR operation, there are sometimes shorter term ones left, from chassis heating usually, that still limit station effectiveness.
Some stations have added Styrofoam insulation, in the form of sheets or “peanuts” around a TCXO in order to reduce these short term variations. In at least one instance, an entire TCXO was outboarded, wrapped in plastic insulation and leads brought back into the radio.
Accuracy
Frequency accuracy isn't generally as critical to VHF+ WSPR operation since, as long as an accuracy problem is detected, it can usually be manually or automatically corrected. This is the function of the A/B settings in the stock WSPR code and described in the WSPR documentation. The goal of these corrections is just to “detune” the radio to correct for its internal frequency errors. These errors may be of two kinds, offset or scale.
The A term serves to bump the frequency on every band a constant amount to correct for a radio system local oscillator which has constant error versus output frequency or band. For older radios, this could include BFOs, carrier oscillators and even 2nd or 3rd conversion oscillators.
Newer radios and multi-band/multi-function radios increasingly use a single master oscillator which, with the help of DDS, PLLs, multiplying, dividing completely determine the output frequency,and perhaps even output phase. Absolute frequency error in this oscillator therefore contributes an error in the output directly related to it's accuracy, measured in ppm, ppb or percent. A master oscillator with a 1 ppm (part per million) error will produce a hamband output signal with 1 ppm error. On 2 m this would mean 144 Hz of dial error. The B term in the WSPR frequency correction code is used to calculate the amount of error for a given band and to “detune” the radio to compensate. Radios with single master oscillators that control everything will only need a B value entered since there are no offset oscillators the A value will always be zero.
Although accuracy errors don't usually prevent 2 M WSPR operation, they can spoil a radio for gathering significant propagation information. Actually this is true of HF as well. If it is known in advance that all frequency errors, within sufficient resolution, are due to the propagation path rather than the hardware at one or both ends, we start to learn more about radio science. At HF, for example, it is possible to measure movement of the ionosphere as the bands open and close. The change in height causes a changing path length which results in measurable frequency offset on a WSPR spot. Particularly on 20 m – 10 m, one or even two Hz of offset can be directly observed.
At 2 m and up, we begin to see some even more interesting effects if we have frequency accuracy. For one, it is much easier to identify individual stations on a busy waterfall where not all stations are decoding. But we even see situations where we get multiple spots, a “main” one and a Doppler shifted one, usually due to aircraft motion changing the total path length between WSPR stations. If this change is constant enough over two minutes, and it occasionally is, then we can successfully decode a spot that came to the receiver only by aircraft scatter.
In this thread, perhaps we can begin to discuss our experiences and methods for improving both frequency stability and frequency activity with the variety of radios and transverters we are using. This could help other stations. In addition I hope to add some information about the external GPS disciplined phaselock system that some of us are working on and hope to make available before too long.
Glenn n6gn cm88ok
7 October 2012