G4HUP/DL4MUP Modifications

GPS Locking
Caveat Emptor! This information is provided in good faith, and has worked successfully on my own IC706Mk2 sample. A similar approach has also worked on other 706 samples (eg MKIIg), and also on IC746 and R75 units. Whilst the mod is not difficult, it does require working in a confined space, and there is potential to damage other parts if care is not taken - if you do not feel competent to perform this modification, then seek assistance from a friend!

All variants of the IC706 (Mk1, Mk2 and Mk2g) use 30MHz as the standard master oscillator, so the principle of the mod should be equally applicable to all - however, there may be variations in physical component locations between the versions. If so, follow the principles and use the manual and service manual to guide you.

Updated!! IC706 Mk2 GPS Locking modification

The master oscillator in the IC706 series rigs is 30MHz. By replacing with a GPS derived signal, the stability and accuracy of the rig is greatly increased. The modification is simple, but requires a bit of care.

You will need:

DJ8MS, PA2V and I have experimented with using the DFS30 to injection lock the IC706. This is a much better solution than the 'one-way' modification previously documented, since the rig can be used without the DFS30 attached when necessary, with no switching or other changes to be made. All you will loose is the stability when running 'un-locked'. The modification has been tried on both MkII and MkIIg versions with good results.

Detail below is given for both versions, but note that physical location differences may be seen between them!:

IC706MkII and MkIIg

Remove the bottom cover of the 706 and locate the oscillator

Take the cover off the LO box to reveal L601, wound on an unscreened former. Using a length of enamelled copper wire (gauge is not critical, probably about 28g will work very well, as it has some degree of 'self-support') wind 5 turns tightly around L601, after soldering one end of the wire to the wall of the screening box.

Mark the lid of the box near the corner above where the coax will need to attach to the new winding and drill it 5mm dia. Deburr the hole to avoid any damage to the coax.

Prepare tails on the free end of the cable and feed it through the side of the power connector on the rear of the radio. Feed the coax across the rig, and into the LO box - I found it useful to have approx 100mm of cable free outside the rig. Slide the drilled box cover onto the cable (!) then solder the braid of the cable to the box wall, and the inner to the other end of the new winding.

That's it - job done! Put the cover back on the LO box, and replace the bottom cover of the radio. This modification is rather easier to carry out than the previously documented option. Pictures are posted below for both the 706MkII and the 706 MkIIg versions - the only difference between them is the physical layout within the rig.

IC706 MkII
G4HUP - IC 706 MkII cable routing with entry via side of power connector shown
G4HUP - close up view of cable entry into LO screened box - make sure you de-burr the hole well!
G4HUP - General internal view of the LO box for IC7606 MkII
G4HUP - View of overwound coil (5 turns) with braid of coax and return of coil grounded to box side

IC706 MkIIg
Neat solution from DJ8MS - but must have needed major stripping of the rig! Other mods also visible
PA2V - L601 in the MkIIg version - 5 turns ECW close wound
PA2V - MkIIg cable exiting cover of LO box
PA2V - Suggestion for routing coax through rig to exit beside the power connector - MkIIg


In operation we have found that a level of +2dBm or greater is required into the rig to lock the oscillator successfully and reliably. Levels slightly lower than this may cause the PLL to hunt and take time to acquire lock. The DFS30 will give approx +5dBm output, so this will lock reliably.

Caution - if you are operating without locking the rig, disconnect the DFS30 from the 706, otherwise 'suck-out' from the LO will occur!

Before modifying the the rig, I set it up to receive GB3VHF, which is a GPS locked beacon on 2m (144.430MHz). Using JT65B I observed the rigs stability and performance overnight. To receive the signal correctly, the rig was tuned to 144.42901MHz, and this gave a initial DF error on JT65 of -11Hz. Over the period of measurement, as the temperature varied, the DF error increased to -89Hz at around 0500, and started to converge towards the initial error again as temperatures climbed.

The modified rig was set up to repeat the test - this time the frequency indication of the rig was 144.428490MHz, and the JT DF error shown was 3Hz. Left overnight the error remained constant.

Sources of Error

Of course, the rig is not the only frequency sensitive item in the chain. The measurements were carried out with a PC and Joe's WSJT software. The accuracy of the PC clock, both short term and long term, since these tests were conducted overnight, will have an impact on the results. My shack PC, on which these measurements were made is time controlled by D4 internet software from Thinking Man, and is set to register every 30 seconds, so I believe that the result given here is valid, and that we really are talking about the variations in the IC706 being the only significant ones. The GB3VHF beacon is GPS controlled and my 30MHz source is also GPS locked.

Initial JT65 reception - note variation in DF
JT65 reception of GB3VHF with external GPS 30MHz - note DF is now constant
lower part of screen - GPS controlled, upper part reversion to internal oscillator
DJ8MS Spectran view drifting line is unlocked, solid vertical line is DFS30 locked
DJ8MS view 15 mins later - note how internal osc has drifted by approx 200Hz, but locked line is held steady

IC706 MkII

Fan Drive Modification - Frequency Stability
Spectran plot of IC706 before modification - scale tic marks on left of screen are at 30 sec intervals
Same signal - after modification

I had known for many years that the IC706 had a temperature related frequency stability problem in CW and SSB. This is not a major problem - the drift cycles by about +/- 30Hz over a 2 minute period as the fan is switched in Rx mode, as you can see from the 'before' plot above. In normal usage this drift is only just enough for the ear to detect - if you know what you are listening for - and would not be an issue for CW/SSB QSO's. I found this when using FFTDSP to monitor GB3MHS when I was building it in 1998.

The two comparison pictures above were captured using Spectran while monitoring the DB0AAT beacon on 23cm.

Once identified and understood,I didn't really worry about it - however, with an increasing interest in JT44/65, I decided to do something about it!

An internet search lead me to a solution given for the 706MkIIG. Instead of having the fan switched, it is run at low level permanently while on Rx - this is enough air-flow to prevent the need for it to be switched on at normal temparatures. A 220R resistor bleeds enough current to run the fan - I used a 5W type because it was available - I would not recommend going below 1W, although calculation says a 0.6W should handle the power

However, on opening the bottom cover of the rig, I discovered that the MkII has a different location of components on the PA board as compared to the MKIIG, so the pictures referenced were not a lot of help! For the MkII the correct point to pick off the +12v switched is shown in the picture below - the same circuit point as for the MkIIG, but a different physical location on the board.

One end of the resistor is connected to the +12v switched line, picked up from the top of the ferrite choke. The other end is attached to the + line to the fan - a small white two pin connector is visible in the left hand picture below - just to the side of the fan. Carefully form and solder the resistor lead to the tab corresponding to the + (red!) connection of the socket. There is not much space to bring the lead up from the connection point, so I would recommend insulating it to avoid any risk of shorting the +12v to the chassis due to vibration etc. Be careful about where you position the body of the resistor, so that any heat generated does not cause the insulation on other cables to be damaged.

View into PA compartment of IC706, showing resistor position under the wires
Another view into PA compartment of IC706, showing where the switched +12v is picked up

page created 08.02.04 page last updated: 7 Apr 16