Making Variable air spaced tuning capacitors

Larry,

Very interesting on the AB amplifier, I remember in the radio constructor magazines ages ago an article for a class ab amplifier using 2 transistors. I believe the bias to the transistors varied according to the input signal and the article was by sir Douglas hall .

I also sold a kit for an FM pulse counting receiver, and maybe I will do a new post for this, as I am thinking of producing a few more of these to sell.

A long time since my last post but I have been quietly busy.

Firstly a simple test jig was needed to enable to quickly test various test caps to see the characteristics of rotation to capacitance.

Basically it consists of the tuning capacitor under test with a graduated scale marked in degrees, with a capacitance meter connected to this. By measuring the capacitance at various points it should be possible to test the rotation versus capacitance of various capacitors easily.

The more serious problem I have is to be able to construct and test variable capacitors

to give a consistent performance, without long construction, setting up and alignment times.

It is quite different to produce capacitors reasonably quickly in quantities, as opposed to making just a prototype. After a lot of work I believe I am now slowly winning so I will battle on.

I have at last managed to do the first tests on measuring the rotation to frequency/wavelength characteristics of the prototype 3 vcap.

A parallel LC circuit consisting of the prototype 3 vcap together with a MW ferrite rod coil was tested at various frequency/wavelength resonance points.

The results show that the cap has a virtually linear rotation to wavelength characteristic as shown in the photo below.

The units are the wavelength in meters. This is not what I expected as the blade profile I used which was shown in two different articles I read was supposed to give a linear rotation to frequency characteristic.

After checking online I noticed that virtually all older vcaps have a linear rotation to wavelength characteristics, whereas the newer ones have an approximation of linear rotation to frequency characteristic.

The smallest commercial vcap I have is of the newer type (made by ALP ) and tests to see how this compares would be useful.

Also the larger commercial vcap I have is the older type (from an old valve radio) and again tests to see how this compares would be useful.

Test results on prototype3 and the two commercial V.caps has now been completed and is very encouraging. I have now used a 2 pf coupling capacitor from the RF generator in order to minimize loading on the tuned circuit. Results have been double checked and are as follows:

TEST FREQUENCY 800KHZ

Prototype v.capacitor Bandwidth=6.0 khz Q=133

Smalll v capacitor Bandwidth=6. 1 khz Q=131

Standard v.capacitor Bandwidth=6.0 khz Q=133

TEST FREQUENCY 1200KHZ

Prototype v.capacitor Bandwidth=18 khz Q=66.7

Small v.capacitor Bandwidth=19 khz Q=63.2

Standard v.capacitor Bandwidth=18 khz Q=66.7

The tests indicate that all three V.Ccaps now have almost identical results and there has been a large improvement on prototype 3.

The only reason I can see for this, is that since the blades of the prototype 3 capacitor have been reduced in size (approximately half the area of prototype 2), the conductance has improved substantially.

Also the small commercial V.cap shows that the Q is slightly lower in comparison and I noticed that the blades on this V.cap were thinner than the other 2 V.caps, so again this could be due to conductance of the plates.

Although tests have not been done at higher frequencies, I believe for at least the MW band the performance should be equal to commercial V.caps.

The next tests will be to see if there is a linear rotation to frequency characteristic.

Just finished the final prototype3 as per photos. In order to make alignment of the plates easier the size of the plates has been down sized. The new cap has a width of just 4 cms and a depth of 2.5 cms with a maximum capacitance of 260 pf. For other values the depth will vary accordingly.

Also as to-day MHZ are used for most tuning dials, the shape of the rotary blades has been altered to give a linear frequency to rotation angle. Thus minimum frequency is now fully anti-clockwise and maximum full clockwise,

Further benefits of the final prototype now are:

1) More compact footprint / smaller volume

2) The plates are now more rigid

3) Alignment is substantially quicker

4) Precision miniature roller bearing

5) Values up to 500 pf possible

To finish, just need to replace the front screws with countersink screws and put solder tags for connections. Also to speed up alignment further, the space washers can be replaced with more accurate and flatter shims.

And then more tests on bandwidth, Q tests and rotation/frequency tests

Thank you for your feedback and I believe your interpretation of the results is correct.

This means however, that in order to keep the Q high the gap spacing cannot be increased, so I am going to have another go with prototype 3, to see if it is possible to change the design, in such a way to make alignment of the plates easier and thus more practical to build.

If successful I will post DXF files for the blades and the base parts, so that anyone interested in building their own V.cap or experimenting further, may find these useful.

My interpretation of your results is that with the closer plate-spacing you were able to get the same (or greater) capacitance using smaller plates, and that with smaller plates there's less conductor loss, as the current flowing over the plates just has a shorter distance to cover. Less loss in conductors, better Q.

The different washer material could also be contributing better conductance, though that effect gets mixed in with the fact that the washers are thinner, once again shortening the distance that the current over them has to cover.

You clearly had to do some long, careful work to produce these devices and measurements. Thank you for sharing the results with the rest of us.

Good and bad news.

The brass spacer washers (0.5mm an 1mm thick) finally arrived and the original washers have now been replaced with I mm spacers. Also the vanes spacers have been installed as in the photo below.

Tests were re-done and surprisingly the Q now at both 800khz and 1200khz is almost identical to the commercial larger V.cap, I am not sure if this is due to the closer spacing between vanes, which now is 0.25 mm (the same as the larger commercial V.cap) or the improvement is due to the replacement of the steel washers with brass washers.

Re- testing, now the capacitor when fully meshed measures around 500 pf

Now the bad news, to align the vanes correctly to the 0.25 mm without any shorting between the vanes, took ages to align, and I believe, at the moment it would be very difficult to construct these on a consistent basis and in a reasonable time. So to start with I will use a larger gap of 0.4mm or 0.55mm between vanes and limit the maximum capacitance value to 250pf, so the V. cap would not be excessively large.

Once enough experience is gained the gap size can be reduced soV.caps with larger capacitance values can be constructed precisely.

Finally constructed the second smaller V.vap. The original design was used and scaled down so that the vanes are half the area of the original.

It was possible to reduced the gaps between the stationary and moving vanes from 0.55 mm down to 0.35 mm. The spacing washers used were 2 x m4 washers (one 0.8 mm + 0.4 mm = 1.2mm thick

This now increases the capacitance between vanes by 0.55/0.35 = 1.57 times but as the area of the vanes is now half, this results per vane set = 1.57/2 = 0.78 times the original approx

The new prototype uses 11 moving vanes and 10 moving vanes so there are 20 gaps between moving and stationary vanes. The total capacitance measured with the V cap fully meshed was 300 pf (i.e. conviniently 30 pf for each moving vane)

The photo below shows the difference in size between the two prototypes

The new V. capias now is far more compact and uses 5 mm thick acrylic for the front and back bases, which reduces the depth by a further 1 cm. To finish this I still need to cut the vanes spacers, so the spacing is more accurate and rigid.

Now for the test results

TEST FREQUENCY 800KHZ

Standard v.capacitor Bandwidth=6.2 khz Q=129

Prototype v.capacitor Bandwidth=6.3 khz Q=127

TEST FREQUENCY 1200KHZ

Standard v.capacitor Bandwidth=19 khz Q=63

Prototype v.capacitor Bandwidth=21 khz Q=57

I did the measurements twice to make sure and the measurements for the commercial V.cap were different from the previous results. I think this maybe because this time the ferrite rod/ coil was resting on a 5 cm polystyrene block, whereas the first test were done with the ferrite rod/coil resting on the bench.

Also tests were done with the bases spaced close to the end vanes and then I cm away, but there was no difference in the results.

The results between the second prototype and the commercial V.cap are almost identical now at 800 kHz but at the higher 1200 frequency there is still a significant 2 kHz bandwidth difference, although better than the first prototype..

As soon as the MK484 ics arrive real time tests can be done which should be interesting.

Also I am waiting for 1 mm thick brass washers to arrive and this will give an even smaller gap spacing of 0.25 mm between vanes, but I'm not sure if it will be possible to align the vanes accurately enough. Maybe with brass washers the results may improve.

After some thinking about the lower Q of the prototype I am wandering if this is due to the much larger vanes I am using to the two commercial V.caps, The thickness of the vanes on the prototype is 0.5 mm, the same as the standard commercial V.cap I have, but the size of the vanes must be around double or more, and maybe this results in greater conductance losses.

To test this, a second test prototype will be needed with the vanes around half the area of the original, but with closer spacing. Hopefully the gap between vanes can be reduced to about 0.3 mm from 0.55 mm, so this will also result in a smaller V.cap for the same value.

Time to start prototype 2 !!

I carried out more tests on measuring the Q of the V.Caps This time a small 3.9 pf replaces the 100k resistor in the original test shown below.

Also I used an analogue RF signal generator instead of the function generator connected to a frequency counter, and again used a scope with a x10 probe to make the measurements, This setup was far more stable and the Q measured was substantially higher than the original tests. From what I checked on the web, this method of measurement only gives the loaded Q, so the unloaded Q may be substantially higher, but for comparison tests this should be OK

TEST FREQUENCY 800KHZ

Prototype v.capacitor Bandwidth=7.7 khz Q=104

Smalll v capacitor Bandwidth=6.6khz Q=121

Standard v.capacitor Bandwidth=5.8 khz Q=138

TEST FREQUENCY 1200KHZ

Prototype v.capacitor Bandwidth=24 khz Q=50

Small v.capacitor Bandwidth=22 khz Q=54.5

Standard v.capacitor Bandwidth=22 khz Q=54.5

The results show that the prototype will need more work on it to match the performance of the commercial V.caps

In my experience with medium wave, the very best coil and wire combination is Special Litz 180/46 wire from Mike radio parts. And the Type 61 Ferrite rod NiZn R40C1- 200x10mm for High Q Amateur & Crystal Radio Coils, AM SW(151884949745) from Ebay

It looks pricey, but you get two and can cut it down to 2 or three pieces

The first tests to measure the Q of the prototype and the the other two commercial capacitors have been done.. The coil was a standard MW coil wound with litz wire on a 15 cm ferrite rod and the test frequency was 800 kHz. A scope on the x10 probe was used for measurements.

Prototype V.cap Bandwidth=15 khz Q=53.3

Small V.cap Bandwidth=14 khz Q=57.2

Standard V.cap Bandwidth=14 khz Q=57.2

The difference is small, but more tests are needed at say 2 MHz as losses may be greater.

It was difficult to measure accurately as my signal generator pots for the frequency and amplitude were scratchy and intermittent. After these are cleaned or replace further tests can be done.

Also I have ordered a couple of TA7642 ICs to build a test radio, as suggested by Larry.

Finally finished this, more or less. Now the moment of truth, will it work?

Next step Testing and more testing.

I am hoping to finish the prototype this week and do the first tests to compare the Q to the two other sample V.caps I have.

From what I understand the Q of a parallel LC circuit is the test frequency (at resonance) divided by the bandwidth (3db cut-off points).

Does anybody know if the following simple test circuit below could be used to compare the Q of the V.caps to the prototype. Unfortunately I have limited equipment at hand (scope and signal generator).

Any help would be most welcome

Finally nearly finished assembly of the prototype, after some tweaking. The final capacitance with the vanes fully meshed is around 480 pf.

The vanes spacers need to be cut and added however, so that the vane spacing is accurate and to make the vanes more rigid. Once these are fitted I can do a couple of tests. Firstly to compare the Q of this capacitor in a parallel LC circuit, to other V.capacitors as in the photos below, and secondly to check if there is a linear wavelength characteristics of the rotary position.

Below is a comparison to two other V.capacitors I had. As seen in the photo the prototype is substantially larger than the other 2 commercial types. This is mainly due to the larger spacing between the vanes in the prototype. Hopefully I may be able to reduce the spacing and optimize the construction to make this more compact

Some progress today, The two bases have been cut from 10mm acrylic, as unfortunately I didn't have 8mm available. The front base has a recess to accept a 6mm bearing.

Also the rotary vanes section, consisting of 12 vanes has been assembled. I used a M4 bolt with two 4mm washers per vane to provide the spacing between vanes. The front spindle which include the bearing was made from a standard spacer which a m4 thread was made.

The vanes spacers still need to be cut.

Next step, more assembling of the stationary vanes and bases

The vanes for the prototype have now been cut from 0.5mm aluminium sheet, 12 pieces of moving vane and 13 pieces if the fixed vane. I have measured the capacitance between one fixed and one rotary vane, set to maximum with a 0.5 mm gap, which measures approximately 20pf.

With 12 moving vanes and 13 stationary vanes there would be 24 gaps, so the final capacitance of the prototype should be around 480pf, if I got it right.

With a bit of luck the 2 bases should be cut this week, so the prototype can be assembled

The rest of the parts have now been designed on Autocad. Below is the rendered 3D view from Autocad, as below:

The 2 bases, front and back will be cut from Acrylic sheet (6 or 8mm thick) and the spacer supports (brown) from 3mm acrylic. (the supports will need to be cut on a laser CNC due to the narrow slots, and the rest of the parts can be cut on a router CNC.

The overall size of the tuning capacitor will not be as compact as the smallest air spaced capacitors available, but should still be acceptable for most purposes. The air gap between vanes will be about 0.5 mm