C5, C6 - Feedback Capacitors
All tests were done at 14MHz. The caps that are switched in by the relays were not included in the testing. It was assumed that by getting the VFO stabilized at 14MHz and having all NPO's in the switched circuits would not effect stability.
The instruments used in the tests were the AADE Frequency counter with 10Hz resolution and a Radio Shack digital temperature readout with the sensor placed near the VFO.
The tests were done in a room with the intake to a electric central air unit three feet away from the bench. As the central air turned on/off, the temperature would change 5 to 7 degrees over a five to ten minute period.
This setup gave more realistic results rather than using a closed box. As a result, several interesting things were discovered.
The most interesting observation was that the different components in the VFO absorbed temperature changes at a different rate. No wax or coverings were used over any of the components.
The toroid core would absorb temperature changes must faster than the poly capacitors. When it was found that the RF choke was also changing the VFO frequency with temperature changes, it was discovered that the ferrite material was the quickest to absorb temperature changes.
From these observations, it was concluded that only slow changes in temperature could be adequately compensated by using combinations of positive/negative temperature compensation devices. Fast temperature changes cannot be compensated because of the different rate of temperature absorption of the components in the VFO.
It wasn't until after these tests that the RF choke was also involved in changing the frequency of the VFO. It was discovered when I was having trouble with the VFO and I placed my finger on the RF choke, in a rather cool room, and the VFO frequency changed very rapidly up and then back down after I lifted my finger off the RF choke. A 250uH molded RF choke was being used at the time.
I then used a large ferrite core and hand wound the RF choke with #24 enameled wire. The finger test was used to determined how much it would effect the frequency of the VFO. One was wound with 16 turns (1mH) and another with 7 turns (200uH). The core was an Amidon equivalent FT50A-75.
The 7 turn choke caused the least amount of change during the heating of the room with the central air blowing over the VFO. The 16 turn choke would move the frequency 500Hz, whereas the 7 turn choke would move the frequency 200Hz. The molded choke would change the frequency almost 1KHz.
Why these changes didn't show up in the tests below, I don't know. In fact, even after these changes, the characteristics of the VFO remained the same, with a very slight positive temperature coefficient. A positive temperature coefficient means that the frequency of the VFO changes the same direction of the temperature change, i.e., raising the temperature would raise the VFO frequency.
The value of the RF choke had significant effects on the VFO frequency. The 1mh and the 200uH values would affect the VFO frequency very little, in fact, changing between these values yielded nearly the same VFO frequency. However, values between these two would effect VFO frequency a lot - several kilohertz.
Since the 200uH choke had little effect on the frequency of the VFO, maybe this was why the testing below did not show the RF choke as being a factor on the temperature problems with the VFO. A 250uH molded choke was used in all the tests. But then again, why would placing my finger on the RF choke, raising the temp of the choke, have so much effect.
The above work yielded several things that should be done to enhance stability. First, the VFO coil and the RF choke should both be covered with wax. Dripping wax from a lighted candle worked fine. The covering of wax isolates the cores from the ambient air, therefore, it takes much longer for the cores to change temperature with changing air temperature.
Second, the VFO should be shielded. Unlabeled soldering holes are provided around the VFO on the PCB. Wire is looped around the holes. The wire is soldered to the ground plane underneath the board, then copper or tin shields are soldered to the looped wires above the PCB. A top is placed on the shields.
Summary of Test Results
Tests #1, #2, and #3 were just shots in the dark. I wanted to see a worst case scenario, with non-NPOs, N220s, etc. Then I went to Test #4 with all NPO caps. The improvement was noticeable.
Test #4 showed a negative temperature coefficient, so poly caps were added until I had a positive temperature coefficient. As test results show on #5, with three poly caps, there was just a slight negative coefficient. With Test #6, the VFO started to show a positive coefficient with 4 poly caps. This meant I was within one poly cap of having the VFO right on the money, but splitting up the final poly between a poly and a NPO would probably not work consistently between the kits, so it was left there for the builder to get it exactly right if one so desires.
Increasing the size of the wire on the toroid helped stability even more. Wire size #20 (#22 was used in the test) was the largest wire I could get on the core, and from the results on #7, it showed that larger wire size on the toroid helped stability.
Using NPOs for the feedback caps worked best. That part of the circuit was the most critical for frequency stability. Using non-NPO or even a poly here would wreck stability much more than at the places where the 100pf and 120pf caps were located. A poly placed in a feedback position would move the VFO in a very strong positive direction. So much so that those tests weren't even recorded.
A feedback cap of only 27pf would result in the VFO not oscillating at 10.455MHz. With a 39pf and a 47pf, the VFO would oscillate at all the frequencies. With a 39pf and a 82pf cap (total 121pf), the VFO showed the best stability. This showed that more feedback helped stability, but they must be the best NPOs that you can get!
If one desires to nail the VFO right on the money, I would suggest that tests be started without any changes to the VFO, but work with SLOW temperature changes, so that the effect of different absorption coefficients on the parts is minimal. Then take out a 100pf poly cap, and combine different values of poly and NPO till you have it as close as you can get, given the time you have to spend on the project!
|3 Degrees - 2.12 KHz|
|w/47pf feedback NPO|
|w/47pf feedback NPO|
|3.3pf tuning cap NPO|
|Freq rises during cool down from soldering|
|Heat it with lamp and goes down|
|Feedback 47pf N220|
|Found out that current through|
|the relay changes the frequency|
|Put black tube over phototransistor|
|Short term stability was horrible,|
|constantly moving tens of hertz|
|5.3 degrees - 650 Hz change|
|All NPO caps|
|Cool down from soldering,|
|Freq increased a lot|
|2.4 degrees - 20 Hz|
|#22 wire on toroid (normally #24 wire)|
|Same caps as on Test #6, #6a|
Send E-Mail || Amateur Radio Receivers || Electroluminescent Receiver