Tuesday, 1 October 2019

Icom IC-9700 drift solved with GPS frequency standard

I have almost finished my IC-9700 review (well a review of the operation of the radio if not the technology), but the section on the frequency issue was getting so big I have decided to split it off into a separate posting.

EDIT - the IC-9700 review is now here.

As regular readers will know, I like a bit of drama and mystery. Having been brought up on cop dramas in the "Dixon of Dock Green" style, to begin with I never really had much to work out. You could see who did it, and the writers went out of their way to assume that you were an idiot who needed all the obvious clues pasted in front of you for 30 solid boring minutes.

We will never know how bad it all was, as of 432 episodes of Dixon made between 1955 and 1976, 400 are missing. I wonder why.
Jack Warner as Dixon of Dock Green (Photo wikimedia)
Then along came Steven Bochco with Hill Street Blues and everything changed. Here the audience were assumed to be reasonably intelligent and could be trusted to work out what was happening. The plots could emerge over weeks rather than 30 minutes. Once I recall noticing a man on a horse appearing in the background of the crimes, who the cops didn't even spot. Weeks later, the link was made, but during that time I thought I was the only person who had the clues. Suddenly it was personal, and I liked it.

And after that it moved on to NYPD Blue and others took up the idea so that we have plots to resolve over weeks and series that last for years. Then there is Vera, which while every episode is complete in itself, has the peculiar distinction of having all its bodies found near GM4FVM. Mystery indeed. What is the link between all these murders and this place?

Vera has been filming at Burnmouth lately but I doubt if you can see FVM's antennas in the background unless they used a helicopter shot. 

Vera isn't really on a par with Hill Street Blues of course. The biggest mystery about Vera is how come a 73-year-old can still be a murder detective? There is hope for career development for me yet - if Brenda Blethyn can do it, why can't I? (Because she can act, Jim)

That was cruel. Jack Warner was over 80 when Dixon finally retired from the force. "Take me with you, I can see" (it worked for Donald Pleasance - well for a while it did anyway).

So where is all this going? Well, this posting is like an episode of Columbo (Bochco briefly wrote for Columbo). The thing about Columbo was that you knew who did it all along. There was no mystery in that sense, you just needed to know how he worked it out.

Plot summary - The Icom IC-7900 is suspected of drifting as the cooling fan turns on and off. Adding a GPS disciplined frequency source stops it drifting. Erm, That's it. All that remains is to work out how Columbo solved the problem.

Nothing to see here. Just stay behind the Police tape and move along please.

Well, there is one mystery to be revealed. "Just one more thing" as Columbo might say. Is drift a problem? Could it be that just one or two samples of the IC-9700 were a bit drifty and those owners went on the Internet and the rest is history?


My initial tests on 2m suggested that my normal usage never brought the fan on anyway. Maybe that is because I only run 25 watts out on 2m, and 10W on 23cms. So I did more checks.

There are two aspects to this -
1) is the frequency readout accurate to start with?
2) does the radio stay on that frequency?
So, for example, my IC-7100 routinely starts showing a 70cms GPS-locked beacon on the wrong frequency (often 100hz out), but stays rock steady on that frequency for the rest of the session.

And then again, does a few Hertz drift matter? Does being slightly off matter either? Well, yes, it does for me. I want to find weak beacons, so I need to be looking in the right place. And I use data modes, some of which are fussy about stability. So let us see what the TV audience thinks.

I used the GB3NGI GPS-locked beacon for my tests. It had been working a bit erratically recently but it seems to have settled down lately.

Taking 23cms first as that is where I expected most drift ...
GB3NGI at GM4FVM on 1296.9042. Mark and space CW ident can both be seen drifting (H scale hz, V minutes)
Monitoring a beacon 272km away on 23cms is bound to introduce some squiggles as aircraft pass and the troposphere gurgles about, but I think this shows the general picture. The rig had been running for about an hour before I started measuring. Over 20 minutes it drifted 57hz. Whether you think that is important enough to worry about depends on your viewpoint. Certainly the JT4G decoded most times.

At the start of the 20 minutes the readout was 19hz high, and it ended 38hz low (or it would have been had I used a zero beat method).

I thought about this two ways. One, these results are on receive, whereas the reports of problematical drift are as the cooling fan turns on and off during long transmissions. And secondly, am I going to miss a beacon by looking 38hz off frequency?

In the end I resolved it this way. This is a rig valued at £1800. The frequency standard might cost anywhere from £25 to £100. Surely it is worth fixing.

Here is what happened when I turned on the frequency standard...
GB3NGI at the point of turning on the GPS from cold - note the jump in frequency

As the GPS standard was starting from cold it would take time to lock onto satellite signals (not very long actually), plus it would then need to discipline its crystal oscillator to bring it to 10MHz. It did not find the right frequency immediately.

To begin with it dropped 57hz, which was 19hz low. Then over the next 2 minutes it moved back towards the correct figure (800hz on the waterfall).

After that I ran it for 2 hours. At no stage did it vary more than 2hz from the expected point 800hz.

Rather than drifting at about 3hz a minute in one direction, it settled down to jogging 2hz around a single point, and it stayed in that frequency range for 2 hours and the frequency readout was perfectly accurate.
Under GPS synchronisation, hours of steadiness with just some aircraft scatter for variety.
For comparison, the drift at 2m was hardly measurable on receive although turning on the GPS made the IC-9700s frequency readout accurate - by moving the radio 35hz. Hardly as significant as 23cm because it seemed perfectly stable even if it was wrong. I can stand being a few hz out but I detest drift.

So there is your Columbo start. No surprise here. We know who committed the crime. Icom by sending out a rig with a fan blowing on the oscillator, without automatic GPS synch input, and then taking ages to upgrade the firmware. The GPS standard works perfectly. You know what happened. Justice has been served. But how was the crime solved?

I bought a Leo Bodnar GPS frequency standard, set it to 10MHz and lowest output level, added a 20dB attenuator to produce the right level for the 9700 (-10dBm approx), and ... well, off it went. Problem solved (not that there was a major problem, but I might have found one later if the fan ever turns itself on).

To start synchronising you need to find "REF ADJUST", by pressing Menu, Set, Function, and then open REF Adjust.

As originally supplied, the first IC-9700s could not use the 10MHz frequency standard to automatically reset the frequency - you could input the frequency but you had to make the adjustments manually. Now after the latest firmware update this works automatically.

At some stage I found out how regularly the synch was done, but either I forgot, or I couldn't be bothered to remember. Either way, this information is not currently available on this site. It synchs often enough to work, I think.

Press "Synch to REF IN". Then:-

"Start". And then for a few seconds this
... and finally ...
Success. You can find pages of bile on the internet about how to get round the reported frequency instability, but once Icom issued this firmware update this resolution became available.That made all the agony irrelevant, though I think Icom should be criticised for putting us all through the misery of waiting for this.

With regular reference to the GPS standard at my end I now find that GPS-linked beacons appear exactly where I expect them to.

It is worth remembering that GPS disciplined sources like this have a TXCO inside. In other words they are a good crystal oscillator, and then they use the GPS signal to fine tune their on-board crystal oscillator. So if you lose the GPS signal, through a fault in your GPS antenna or whatever, the oscillator will continue to function on its own at the specified frequency. Once GPS reception is restored the accuracy will return to GPS standard.

A few points arise at this point. Firstly you do not need to use a Leo Bodnar unit as I did. Jan, OZ9QV, contacted me about the RFzero unit. I looked carefully into this and it seems like a very good item and I would encourage anyone to consider it. My choice was for the Leo Bodnar was for a non-radio reason. Or at least non-technical reason. I have nothing against the RFzero, just that the Leo Bodnar was bought for a personal reason. It cost £100. They do a buffered two output one too, but that costs more. I would have liked the dual output one, but I know I have no real reason to have one.

Secondly, I could have used my Trimble Thunderbolt, which I see others have done. Apparently it works according to reports, but I wanted to keep the Thundbolt for providing PC clock synchronisation for that portable data-mode operation I will someday do outside wi-fi range. You know, that portable operation which I have never done.

Thirdly, while the Leo Bodnar has a USB plug which provides power and a data connection, with my set-up this cannot come via my PC. The way my system here works, everything is isolated from the mains power using "the big switch" method. I turn the big switch on which starts the rigs. The computer is powered up but it does not start until I press the button on the front. If the computer was then powering the GPS, the 9700 would have lost synchronisation as it had started before the frequency standard. You do get a warning about this
I know I am old-fashioned, but after sitting through all those ergonomics or "man-machine interface" lectures, as we called them in those days, I hate seeing warnings regularly. Warnings are to tell you something is not right. You should take action on them.

To eliminate this warning I simply powered the GPS reference via a separate power supply which comes on at the same time as the radio. My PC, which is started later, does not power it. The drawback to that is that I would need to unplug the GPS and plug it into the PC if I needed to change the reference output frequency - something I have no plans to do.

I know as I write this that somebody will criticise me for leaving the radios on and letting the big switch do the on/off role. Sure, there is potential for a fault in the power supplies as they come on and off which might damage the radios. However, this has not happened yet. Well, a power supply did go bang once, but no damage was done (to the radios ...). My angle on this is that keeping the big switch off is more important, but I might be wrong about that. Having too complex a shut-down routine will probably mean I will just forget to do it.

Turning briefly to the Leo Bodnar GPS standard...
Leo Bodnar GPS frequency standard with a Danish Krone for scale
Maybe I might have chosen a different scale object. OK, let us just say that you can judge how small the thing is from the SMA socket on the left hand side. That socket is for the supplied GPS antenna which is equally tiny. I took the GPS antenna, thinking it would need to be placed in the loft or even outside, but temporarily stuck it to a metal shelf inside the shack (it is magnetic). I had GPS lock within seconds, even with all the RF floating around in here.

You set the reference output frequency using very simple software. I imagined the set-up would be difficult, after setting up the Trimble Thunderbolt years ago. The Trimble only has a fixed 10MHz output (plus loads of harmonics of course). The Leo Bodnar standard was simple to set up by comparison. With a huge range of available frequencies the Leo Bodnar must have many uses around the shack.

1) the drift on the IC-9700 was there on 23cms but not a big issue on receive.
2) firmware updates allow GPS synchronisation so why not do it?
3) as well as stopping the drift, the frequency readout becomes dead accurate - turn it on each morning and it comes on at the same frequency every time
4) the GPS frequency standard unit is great and it will find lots more uses

Was the drift bad enough to worry about? Probably not. But it certainly is not worrying me now.

At the end of Dixon of Dock Green, the man himself stood on the steps of Dock Green nick and gave a little homily to the audience at home about bad people getting their come-uppance and good folk having nothing to fear. This always stuck me as odd, because even as a child I knew he couldn't address me personally as he was a fictional character acted out in front of a TV camera by a dozy old actor 500 miles away.

I should give a similar homily now but, looking out at the world in 2019, I don't think you would believe me that we good folk have nothing to fear.

"Evening all."



  1. Hi Jim, can this he applied to a IC 7300 or IC 746 Pro?

    Thanks Dave WA3HSC

  2. I ordered a Leo Bodnar unit and it arrived the other day. However, the 20dB attenuator that I also ordered, from a different supplier, has not arrived yet, so I did some tests of the frequency drift using GB3MHZ, as I can't hear either GB3NGI or GB3CSB at the moment.

    After listening for over an hour the drift was +60Hz which an online calculator tells me is +/- 0.04627ppm.

  3. This comment has been removed by the author.

  4. Update - The 25dB attenuator came today, after swapping information with other IC-9700 users, I had put the unattenuated 10MHz into the GPSDO socket with no apparent problems.

    When I added the attenuator and reconnected the GPSDO to the 10MHz input, I noticed that the indicated audio frequency out put from GB3MHZ had moved about 25kHz. So, I stopped the synchronisation, waited a few seconds, and then restarted the synchronisation process.

    Now the output, using CW mode, from GB3MHZ is back to exactly 600Hz audio offset.

  5. Dave. I have tried to reply several times but Blogger will not accept it. Not sure why. Anyway I think that the attenuator is worth doing as long term damage may result from overdriving the 9700 via the 10MHz input socket. You should be well set up now.
    73 Jim GM4FVM