I was measuring the warm-up drift of some of my boatanchors. This was both
the final IF signal of a crystal calibrator (checked to be sure it was stable) in the
20 meter band and the BFO, both measured by my Airspy, itself checked with
a synthesizer known good to 0.05 Hz.

None of the radios with main variable oscillators above 14 MHz were competitive,
drifting many kilohertz from turnon, in 3 hours.

But surprisingly the R390A came in second to the Racal RA17C12, both signals
drifting (in opposite directions) by 450 to 550 Hz while the Racal got 350 or so Hz for
the tunable oscillator (2nd VFO, the 1st being Wadley-ed out) and 200 for the BFO.
Tomorrow I will do the Drake 2A.

I got about the same result (400Hz warmup) when I measured my R-390A several years ago. But still, pretty impressive for back in 1950 when the R-390 first came out. And, of course, most military users kept the radio running 24/7, so the drift was probably around 10 Hz. More recently, I tried using the R-390A for frequency measuring tests and, unsurprisingly, the PTO and BFO were not stable to better than a couple of Hz over a period of a few minutes. But still pretty impressive for a 70 year old design.
I'm amused that I now do my FMT using an IC-7610 which has no oscillator to drift, and all signal processing is locked to an external Rb reference. So stability and frequency accuracy are on the order of .01 Hz. It only took 70 years to improve by a couple of Hz!

Ed W2EMN

I did some testing on the oscillators warm up frequency stability and found that the 1st crystal oscillator becomes stable (within a 2 to 4 cycle operating range) in about 12 minutes. This is with using the recommended 75 degree thermostat and it working correctly. The thermostat cycles on and off allowing a small temperature change to occur in it's oven. This causes the 4 cycle range of frequency change about every minute. Don't forget that it needs to have the correct type of crystal, a CR-27/U, as they are cut to operate best at a certain temperature. I have also found a different 1st crystal oscillator that became stable (within a 10 cycle operating range) in about 2 minutes. This difference depends on the crystal and the oven in use (they were of the correct type and specifications, but just had different characteristics).

I also did some frequency measurements on the 2nd crystal oscillator deck during power on warm up (on an R-390A I'm currently refurbishing). With the oven heater off ('oven' switch on the back), the 12 MH signal went down 98 cycles in 1.5 hours and another 48 cycles in the next 2 hours. So, that's about 150 cycles in 3.5 hours, but that's a constant change for quite a while. However, with the oven on (I disconnected the VFO oven internally), the 12 MH signal goes down 450 cycles and stabalizes in 7 minutes, in a 9 cycle range. Remember, the crystals in the 2nd oscillator deck are a different cut (CR-36/U) than the ones in the small round octal plugin oven (CR-27/U).

With the 'oven' switch (on the back) turned off, the heater power to the VFO AND the 2nd crystal oscillator deck is shut off (which is now quite normal). Thus, the 2nd crystal oscillator deck takes about 3 hours to stabalize. The reason it takes that long is that the temperature of the whole R-390A must stabalize, and it takes about that long. The same is the case for the VFO as for the 2nd crystal oscillator.

This Cosmos VFO that I just temperature compensated drifts down 39 cycles starting at 1 minute after power on, to become fairly stable after another 4 or 5 minutes. At 5 or 6 minutes after power on, it varies +/- 10 cycles for the next 3 hours. At this point it then improves stability to +/- 5 cycles per hour and stays very close to the original frequency the whole time from power on.

Due to age, the BFO can and should also be improved by adjusting the NTC values.

Regards, Larry

My R390A has only the calibrator and 1st xtals temperature regulated.
The PTO can't be regulated because when I took it apart to get the end points right
the mechanism fell apart from very very long use in its previous life. Some insulation
remains, probably slowing down its stabilization.

But how could one fix the BFO? I suppose one could take it apart and measure the
temperature coefficients of the caps externally, as there are no specs for them,
and also look at the dots. Once could then replace with NP0 ceramics and measure
the result. Then all would likely come to a halt, as the necessary temperature
compenstaion cap set could not be purchased. Surplus Sales of Nebraska has
a few old ones, but new ones seem to be an extinct species.

Is there a place to get a big selection?

The drift for an R-390A seems excessive. My Collins 51J-4 shows about 250Hz drift upwards in first minute and then settles down to about 50 Hz upwards drift after 5 minutes. Drifts an additional 50Hz upwards after 15 minutes and then stops drifting.

adding the Drake 2A:

BFO drifts down 40 Hz in two hours, is down 70 Hz at 4 hours and essentially stops drifting at
5 hours and is responding to voltage changes over about 10 Hz. At 50 kHz one expects less drift and
that's what one gets.

The VFO drifts up 200 Hz in an hour , then the slope channges and it is pretty much a straight line at
50 Hz per hour (up) over 8 more hours. Its competitive.

But when will it stop? Wild guess: when all the moisture is out of the coils.

The largest supply of NTC cap's is at Surplus Sales. They have TWO listings of them. See page 7 of my document on the R-390A disk: http://www.r-390a.net/Haney_R390_Newbie ... t_Info.pdf . I thought that there were plenty to choose from, although there were a couple values I wanted, they did not have.

I did not mean to say that you should use the heater on the VFO (that is a bad idea), I meant that the VFOs can be temp compensated by changing the NTC cap values.

And same thing for the BFO. I've done a few and it's not that bad to do.

Edit: The BFO (like the VFOs) have two temp compensating 'uncased' caps inside. They are usually 50 pf at ntc 470 and 50 pf at np0. The cover is a little hard to get off, but it will. You can try some penetrating oil around where the cover contacts the frame. The rubber seal is probably very tight and dry.



Regards, Larry

Here's some more info on stabilizing BFOs. I posted this on the R-390 Reflector forum on Aug 21, 2019.

My BFO I'm working on is drifting about 350 cycles in 3 hours. So, I opened it up and found that the 2 - 50 pf 'uncased' caps had been replaced before. One had no markings on it and the other was 50pf at N470. I replaced the 2 50's with one 100 pf at N750, and now it's quite good. At 1 minute after power on I set it at 455kc -10 cycles and in 10 minutes it's at +5 and hangs around there +/- 2 cycles for 50 minutes, then drops down 3 cycles and stays there for an hour +/- 1 cycle. That's good for one of these units. I tracked the frequency with my counter by putting an insulated tube shield on the oscillator tube after removing the normal shield. It makes a great signal pick up.

Regards, Larry

CR88 here. Once 5Y3 conducts and oscillators come on and quickly settle down the set drifts 600 cycles over the first 15 minutes. The drift is down in frequency. After another 10 minutes an additional 100 cycle drift is noted. I didn't take time to observe any longer than that but once it has been on for around half an hour I can use the BFO to copy SSB and go for at least 15 minutes before I have to re-tune ever so slightly.