Golborne Vintage Radio

Introduction

This is a project I have been undertaking, on and off, for the past several months. As is often the case, I let other projects and restorations get in the way from time-to-time, which has served to delay its overall progress. However, recently I have been giving it my undivided attention once more. So if I may, I would like to share some of the work already achieved.

I first became aware of the existence of the rare Ekco A22T whilst carrying out research into the different types of A22 dial a couple of years ago. More recently, I fortuitously acquired a semi-complete A22 chassis and my thoughts again returned to the export model with the idea of making a replica version.

In common with all export sets, when compared to their domestic equivalent, the model dispenses with the long wave band in favour of additional short wave ranges. Specifically, the wavebands offered on the A22T are; MW (550-1500 kHz) and 2 x SW bands (3.0-8.5 MHz and 8.5-23.0 MHz).

Consequently, not only would I have to reproduce the correct dial used by the A22T but calculate and wind the appropriate short wave aerial/oscillator coils required, then carry out a complete rebuild of the coil-pack/band-switching section in order to offer the correct wavebands.

Other than those two major differences, the A22T is virtually identical to the standard A22 model. The only further minor difference, is the additional voltage tap selections found on the mains transformer.

After extensive searches of the internet, I eventually found several images of the A22T suitable to work from. The following complete image of an original A22T and the others accompanying it, showing closer segments of the dial, proved most useful;

[attachment=4182]

I was able to combine the dial segments into a collage, of almost the complete dial face, good enough to obtain the detail I required to begin attempting a reproduction artwork.

I also came across the following website;

http://www.radioantigo.com.br/ekco.html

which has a series of images of this rare model, unfortunately too small in size and resolution to be really helpful with anything except overall appearance.

Creating the dial

The dials are such a prominent and attractive feature of the A22 models that without being able to produce a satisfactory reproduction, proceeding with the rest of the project was somewhat pointless. So with reference to my 'Reproduction Ekco Dials' article (BVWS Bulletin, Vol.36 No.4, Winter 2011 issue), I set about producing a reproduction A22T dial using exactly the same techniques and methods outlined there. However, I will be expanding on that information, showing the various steps involved in greater detail that are required to create such a dial from just a series of images rather than from a scan of an original.

Despite only having a limited number of A22T dial images to work with I could still place the dial legends and details fairly accurately. This was achieved by loading the dial face collage into my graphics software and noting the X,Y co-ordinates of all the major detail; frequency markers, station name positions, and so on. From this co-ordinate information, and by already knowing the co-ordinates of the exact centre of the dial, I could calculate using trigonometry the angular position of all the dial details to transfer to my reproduction.

At this point I assign an arbitrary colour to each waveband legend, in this case black for MW, red for S2 and green for S1. The exact shades are not important at this stage as when the artwork is complete the legend colours and the background shade can be varied independently of each other to achieve an acceptable visual colour match that approximates the colours of the original dial.

The following set of images show in steps how the dial artwork is created, with each image building on the last.

I begin with a base image and just the three semi-circular waveband legends;

[attachment=4183]

Then I modify the black MW legend to include the required station and frequency markers;

[attachment=4185]

And to complete the MW legend I add the station names and frequency points;

[attachment=4186]

Next the red S2 legend frequency markers and bands;

[attachment=4187]

And the addition of the S2 legend's station names and frequency points;

[attachment=4188]

Now the addition of the green S1 legend's frequency markers and bands;

[attachment=4189]

And the S1 legend's station names and frquency points are added;

[attachment=4190]

Finally the legends around the band switching, volume and tuning controls plus the Ekco logo;

[attachment=4191]

Each of these main steps are broken down further into several sub-steps but essentially adding layers of detail in this way, a little at a time, is how the dial artwork is eventually produced.

Colour matching the dial

The artwork is now complete but only in terms of detail, below is the same artwork with the legend colours and background shade visually approximated to those used in an original dial;

[attachment=4192]

Note: The colour shades displayed on your particular monitor will in all probability be quite different to those displayed by mine. Similarly, zooming into the images will show jagged legends and other detail simply because the images have been reduced in resolution for the forum. The original artworks are 6786 x 6786 pixels in size at 600dpi resolution and display no such 'blockiness' when printed.

Just as the colours displayed vary from monitor to monitor, so do the actual colours that are printed by a particular printer. I only use inkjet's, so my comments specifically refer to those types. Other types of printer are probably the same but I have no experience to confirm whether that is the case.

To illustrate what I mean, below is the same artwork as the last with the colours and background shades altered so that my printer actually prints the dial with the correct colours on to photo-paper;

[attachment=4193]

You can see that the colours displayed on the monitor are quite different to those actually printed out by the printer. In order to achieve the correct colour values required I first printed several colour swatches, such as the one shown below;

[attachment=4194]

These swatches are freely available on the internet, but in any case it is easy enough to create your own. Each square of colour has its RGB values printed on it, and when printed by your printer will give you a permanent reference chart so you can determine exactly how your own printer will produce a specific colour combination. When doing this be sure to print on exactly the same sort of paper and with the same printer settings as those you intend to print the artworks with or the resulting colour chart will be invalid.

The problem of colour matching is not only confined to the differences between what a particular monitor displays and what a particular printer actually produces.

It is a fact that the human eye perceives colour hue and saturation differently in response to contrast. Put in terms of colour matching a reproduction dial, if a dial artwork is viewed against a light backdrop then the shade used for the background of the dial will appear darker than if the same is viewed against a dark backdrop.

I had not realised just how striking this effect can be until I began experimenting with printing out my artworks. Often the background shade of the artwork that came out of the printer looked too dark, against the white of the photo paper I was using, until I trimmed and test fitted it into an actual bakelite case when the same dial background then looked too light.

To overcome this problem I created a graphic version of a bakelite A22 case, accurate in all respects, that I could test fit the artworks 'virtually' before physically printing them out. This method is not perfect, but what it does do very well is give a close 'ball-park' figure for the RGB colour values of the required background shade for the artwork. It is then a case of 'tweaking' this value until a final shade is found that is visually pleasing and acceptable.

Below is the completed A22T dial, printed and laminated ready for trimming;

[attachment=4195]

And the virtual brown/bronze case I created with the completed A22T dial artwork test fitted;

[attachment=4196]

Hopefully, when the project is fully complete, the replica A22T will look something along the lines of the above image.

Next instalment; - Calculating the SW aerial/oscillator coils.

Robert,
an insightful treatise.
To prevent loss in in the mists of time duly stuck!
Brilliant my good man.

cheers Mark

Excellent dial work & description Robert, I did the graphics for the logging dial on my Marconi CR300 a while back, not half as complex as yours, it took me ages as not to good with computor programs, I used Inkscape and managed to find the tool that replicates angular degree rotation. I was going to put it to transfers but it's on hold at the moment due to house diy. It sounds as if your project is quite challenging and comprehensive.
Good luck.

Lawrence.

Robert, yet one more of your excellent threads, at this rate, you will soon need your
own section.

The effort that goes into those dials just takes my breath away, and the detailed explanation outlined above shows the true extent of just what effort and expertise is involved in creating a dial 'from a blank sheet of paper', but even to get to that stage, there's a great deal of research to be undertaken. The size has to be acurate, the stations exactly where they belong on the dial, the text needs to be in the right font and the right colour. Dials are such an important aspect of a sets appearance - none more so than the iconic Ekco A22 - in my view, by far the best circular Ekco.

Quite a task too, to calculate the winding of the short wave coils and to re-build the coil pack. Not a task for the faint-hearted!

Thanks for the write-up Robert - it's done nothing for my fragile flaky ego to see such fine work!

Thanks Mark, Lawrence, Paul and David for those very kind remarks, much appreciated.

Calculating the SW aerial/oscillator coils

The equation used for calculating resonant frequency from a specified inductor and capacitor is the familiar;

[attachment=5686]

For convenience, the equation has been modified from the standard to accept values in kHz, pF and uH.

The limits of the range of frequencies covered in a practical LC tank circuit are determined by the capacitance ratio of the particular variable capacitor being used, this often results in the tuning limits extending well outside of the required range. Ideally what is needed is only the desired tuning range to be covered and for it to be spread out over the full range of the variable capacitor. Not surprisingly then, this is termed; 'bandspreading'.

The tuned circuits commonly used in superhet designs make use of combined parallel and series bandspreading, by including a parallel 'tracker' (usually an adjustable trimmer) and a series 'padder' (usually a fixed capacitor) to modify the main variable capacitor's capacitance range.

The following is a general diagram of a LC tank circuit;

[attachment=5687]

CV is the variable capacitor, CT the tracker (trimmer), and CP the padder. CS is any stray circuit capacitance present, typically 10pF or less.

The equation combines all these factors to give the overall 'net' capacitance presented to tune the inductance to resonance.

If your interested in delving deeper into the subject of bandspreading, paying a visit to Robert Weaver's excellent website; Bob's Electron Bunker is well worth your time. I referred extensively to his in-depth explanation of the mathematics involved and made use of the on-line resonance calculators available.

Normally, you design a LC tank circuit to cover a desired tuning range then mark out a tuning dial afterwards to tally with specific frequency points. Unfortunately, my problem was the exact opposite. I already had the dial with specific frequency points marked on it, for which I needed to design LC tank circuits to track those points with a high degree of accuracy. A much more difficult proposition.

To begin to solve the problem, I temporarily fitted a test print of my A22T reproduction dial to the chassis and rigged up a suitable power supply for the illuminated tuning arm. Whilst accurately placing the tuning cursor on the major frequency markers on the dial, I noted down the corresponding capacitance reading from a digital LC meter I had connected across the main tuning capacitor. Repeating this procedure for both short wave ranges, I obtained the specific capacitance value that produced a known frequency at several points across the dial. I now needed to try, by trial and error, various values for inductance, tracker and padder that best matched all of the required frequency points as accurately as possible using the above equations.

It will be appreciated that many combinations of values needed to be searched through. To do so by hand with only a pocket calculator would have taken a very long time, instead I wrote a small computer program to do the 'number-crunching' for me. Even so, it still took the computer several hours of continuous calculations for each short wave range to arrive at the best match having the lowest error difference between required frequency and calculated frequency.

Surprisingly, the final values found produced a maximum error of a fraction of one percent across all required frequency points for both aerial and oscillator tank circuits on both short wave ranges. More than accurate enough for my purposes, even allowing for the inevitable errors involved in positioning the tuning arm accurately on a dial frequency point and in the accuracy of measurement of capacitance by the LC meter.

Here are the results presented in tabular form;

[attachment=5690][attachment=5691]

Note: In both tables the calculated frequency values under the 'Oscillator' heading are higher than the specified 'Dial' frequency points or the calculated frequency values under the 'Aerial' heading due to the addition of the 465 kHz intermediate frequency offset.

Some of those frequency errors may seem quite large, for example; the largest error on the S1 'Aerial' table is at the 23000 kHz point which actually comes out at 23147 kHz in the calculations. In reality, the 147 kHz difference amounts to less than the thickness of the illuminated tuning cursor. So more than acceptable in practice.

Next instalment; Winding the SW coils and rebuilding the coil-pack.

Robert, this is a brilliant insight, even though it contains equations "shudder" but it is easy to follow, even for an no brainier non mathematician like me!

Would it be possible for you to do a couple of things?

1 provide the program you used

2 do this as a pdf download for the main site as Tony has done for some other radio repair things?

either in the the archive or the home pages?

If for the download section could you let Mark H have it to upload.

This is brilliant and I am in awe of the lengths taken to effect a restoration. It harks of the original designers!

cheers Mark

Appreciate your very kind comments Mark, thank you!

I would be more than happy to provide the computer program I wrote. However, it wouldn't be of much use to anyone I'm afraid.

This is because it is specific to an Ekco A22T, using the frequencies from the dial and the corresponding capacitance readings. In fact, its actually four separate programs each one tailored for each of the four coils I needed to make.

Unfortunately, the program is of no use in a general way for calculating other coils needed in different makes/models of radio. Remember, I needed to do the calculations in 'reverse', as it were, to match the existing dial.

Hope that explanation makes things a bit clearer.

Regards

Really great stuff Robert... your perserverance is inspirational.

I found the use of colour swatches most interesting as I too have created dials. See BVWS Bulletin 2009 Autumn. Some were only black or pretty basic colours that were easy to match by eye. Of course how much have the colours changed with time but you can only match to what you have.

For a silk screened dial for an HMV 650 / 655 I was in the hands of the screener and it was up to him to do the best he could to an original but damaged dial. Actually the result he achieved is amazingly close to my eyes, so well done to him.

Now a plug I still have a few dials left from the small quantity I had made (CNC glass + screening) so if anyone wants one please get in touch. When they have gone there will be no more.

I will try to add a pic of the dial... (I normally go the PhotoBucket route)

Wow! that was so easy... great stuff, well done Admin

cheers Gary