Straight Line Frequency Capacitor

[pwdrive]

Al, It's the first one in Alans first photo that I was refering to, appologies for any confusion, thanks for commiting to conduct the experiment at some time.
Good luck with the house DIY.

Lawrence.

[Skywave]

The graphical results of plotting capacity vs. angle of vane enmeshment for a SLW variable capacitor with a nominal Cmax. of 520 pF.

[attachment=7679] [attachment=7680]

Remarks.

Graph 1 is a plot of angle of vane enmeshment vs. total effective capacity:
two curves: one without any series capacitor and one with a series capacitor of 150 pF.

Graph 2 is same, except that the vertical axis is the logarithm (base 10) of the total measured capacity.
There are 4 curves:
1: No series capacitor included;
2. Series capacitor = 470 pF;
3. Series capacitor = 270 pF;
4. Series capacitor = 150 pF;

At a first inspection, the curve on graph 1 with the 150 pF series capacitor looks like that that combination is what is required: a straight line (approx.), capacity vs. dial angle. But, of course, that is what we do not want, since in an L/C tuned cct. with a known and fixed value of inductance, f is proportional to 1/√C.
What is not obvious, however, is the nature of the relationship between vane enmeshment and the resultant capacity: the curve (no series capacitor) could be a variety of laws: all we can deduce is that that relationship is non-linear.
However, when we plot the log of the capacity against vane enmeshment angle (between 0° and 180°) - as per graph 2 - we can see that for dial readings from about 40° to 180° (no series capacitor), the curve is very nearly a straight line. Hence, the capacitor has a 'log-law' characteristic: this will correspond to a straight-line wavelength (SLW) variable capacitor.
The effect of introducing a series capacitor - and its effects on the law, as its value is changes - are very evident, as also is shown on graph 2.

Al.

[pwdrive]

Thanks very much for that Al, Interesting, there will obviously be a trade off if I take the padder route but I will try that first before considering other options, if I can't find a right value SLF the other possibility would be to alter the inductance, trouble is I have got the regeneration working quite well with the inductor I have over it's 500Khz tuning range and regen detectors don't like being upset.
I am not too worried about SLF in the RF tuning as the bandwidth of those tuned cicuits is quite wide, not too bothered either (within reason) if I build a VFO for it (bandset) as I plan to incorporate a 500Khz Xtal marker into the receiver so I can set the VFO (bandset) spot on.

Lawrence.

[Skywave]

I didn't have the time to complete my last post since towards the end of it I had a shout for food (Must get one's priorities right! ).

So, to resume . . . .
A closer look at your design requirement.
You have a 225 pF variable with an appropriate law (You may have an SLF or an SLW: the difference in practice is relatively small: the the SLF gives a curve with a slightly greater curvature, since it follows a cubic law; the SLW follows a square law. This is illustrated in Terman's article, to which I referred earlier.) You require 150 pF: a quick bit of arithmetic shows that a series padder of 450 pF is required.

I used a 520 pF SLW for my measurements, which is different to your available 225 pF, but my measurements & conclusions still have a validity for your design need, nevertheless: I'll explain . . .

My second graph shows how the the curves deviate from a straight line as Cp is varied; there is also a table shown of the ratio [total C / Cmax] for various values of Cp. {Cmax = the maximum value of the variable capacitor; total C = the total effective capacitance.} The closer the ratio [total C / Cmax] is to unity, the closer the law is to a straight line. [It is that ratio which is important here, so the absolute value of the variable capacitor is not that relevant - hence my earlier comment]. For your Cmax = 225 pF and total C = 150 pF, that ratio = 0.67, and looking at graph #2 and interpolating for the corresponding curve, we can see that yes, there is a deviation from the required straight line, but that it is not a dramatic deviation - so I would expect that the resultant dial / freq. characteristic will produce a degree of linearity that should be quite adequate for your need. If nothing else, it indicates that building a prototype using those component values should not be a complete waste of time - and, moreover, I would expect that it would approach quite closely to your identified need.

Having said all that, if, on test, the degree of linearity is indeed inadequate, there is (of course) always the last (and debatably dramatic) recourse of removing vanes from the 225 pF cap. to make it into a 150 pF. (I myself have had to do such a 'modification' on more than one occasion). The necessary calculation (how many vanes to remove) should be relatively straight-forward, but the 'prune-and-test' approach with the aid of a reliable and calibrated capacitance meter still has a lot to be said for it. In either case, it should not be necessary to change the value of the corresponding inductance - which, quite understandably, you are loath so to do.

All of this has reminded me of an HRO which I did a complete electrical re-design job on as a project many years ago (with apologies to HRO aficionados ). Since I chose my own mixer and local oscillator circuits, it was necessary to re-calculate the inductances, trimming and padding capacitors for all the mixer & L.O. ccts. in all the coil packs. All in all, quite a task - but it was eventually successful - and also very enlightening as a design problem: the Radio Designer's Handbook (Langford-Smith) was a useful aid.

Al. / April 10, 2013 //

[pwdrive]

Thanks again Al all your research has been noted, I'm getting the picture now, I still can't get to my project yet because of the house diy stuff which is doing my head in at the moment.
Meanwhile here is an interesting link pertaining to bandspread design etc, it covers some ground that you have already covered, some of the maths it a bit to heavy for me, an interesting site nevertheless.

http://electronbunker.ca/Bandspreading.html

Lawrence.

[Skywave]

I still can't get to my project yet because of the house d-i-y stuff which is doing my head in at the moment.
Lawrence.

You and me too! But at the end of the day, there comes a time when a bit of R & R is required: and that's when I put down the D-I-Y tools and sit down with a pencil and paper - and test ideas.

Thanks for the link to the article on bandspreading. Haven't read it in depth yet, but it looks like an interesting and informative read.
And in it, I did note that Mr. F. E. Terman gets a mention. I have three books by him: 'Radio Engineering'; its successor, 'Electronic & Radio Engineering (4th. ed.)' and 'Measurements in Radio Engineering'. Although his work 'Electronic & Radio Engineering (4th. ed.)' is substantially thicker than its predecessor, in many places its analysis is quite shallow compared to its predecessor and some classical topics are not considered at all. However, 'Measurements' is an enlightening read. Overall, I would recommend these three books to anyone who likes to 'dig in' to the theory of Radio.

Al.

[pwdrive]

Thanks Al, yes I have read a few online articles by Terman, I will try and put those books on my xmas list.

Lawrence

[Skywave]

As a follow-on to my reference to Mr. F. E. Terman, I've just opened a thread inviting readers to learn about the history of Hewlett-Packard, of which Mr. Terman played a pivotal role.

Al. / April 10, 2013 //

[Radio Fixer]

Thanks for your graphs and pondering Al and I have taken copies. I found it most interesting and it has some relevance to me as I made a Wobb with two oscillators, one fixed and the other variable. Mixed they gave me a Wobb that goes from 0 to 2mHz and is ideal for supie IFs, also TRF's and bandpass filters.

I got the padding right of the variable one empirically.

Gary

[Skywave]

Hi Lawrence -
I'm glad that you found my thoughts to be of value. Being a bit of a techie, for me, it's the way that things in electronics 'interconnect' - how the variables affect each other - that fascinates me.
As for the empirical method, yes, that is one way of determining the size of 'the variables' - indeed, I have occasionally taken that 'design' route myself. Having said that, of course, often is the case when a bit of analysis beforehand can save a lot of time by narrowing down the range of choice - and sometimes can even indicate that what is envisaged simply is not going to be achievable!

Al.