The Mechanics of the Site

Site Chronology.

December 1999. An Internet Museum idea was forming. The technology of the 19th century, Steam Power, was large and impressive and readily attracted visitors to the remaining examples of Victorian Engineering. By contrast the defining technology of the 20th century, Electronics, was lacking in large visible artefacts. An exhibition of valves fails to capture the imagination, even of the converted. But - take a detailed photograph and the valves becomes a work of art sealed in a vacuum to preserve it over time. Not only a time capsule, but also a working demonstration of the black art of electronics.

January 2000. The software and database systems were designed and tested. From the outset of the project it was planned to offer a CD version of the museum so that the valves could be seen in detail in high resolution photographs. At the end of the software generation process a careless deletion erased all of the source code. A moment too traumatic to get angry. However, within 24 hours the programs were re-written and improvements made.

February 2000. The site was first launched on February 5th by the uploading of the equivalents system. This had taken three weeks of solid work in early January. The first edition had 3700 references.

February 20 2000. The first 46 full exhibits were uploaded together with the metal valve dissection article.

March 5th 2000. The exhibit list reached 127, with 1326 index terms presented.

April 2000. The museum photographs went digital with a Sony SLR camera.

July 2000. The then current edition of the UK Amateur Radio magazine RadCom, presented a review of the site, this boosted visitor figures to a rate of 10,000 per annum.

Summer 2000. The exhibits grew steadily in number to over 700. By October the number was in excess of 800 thanks to several generous donations.

June 2001. The site by now had 1176 valves, 3046 valve names and over 4100 equivalents. In total the site covered 2,892 files and 46.1 Mb. Hosting moved to a professional server company. The site statistics by then were showing a visitor rate of 40,000 per annum and a peak page hit rate of over 15,000 pages per day.

June 2001. As well as The RadCom article we were also featured in Radio Bygones.

July 2001. As curator of an Internet museum with an appetite for cash, it was necessarary to also have full time paid employment.

Summer 2002. An Olympus E20 digital camera was purchased to take the photography to a new level.

December 2003. By the end of 2003 the museum was well loved by its users but the site had only grown slowly from the initial burst of activity. The records show some 16,000 page hits per week, more importantly we were regularly attracting 1,800 different users each week. With the aid of my artist daughter the site typography was transformed. Cascaded style sheets were introduced. The redesign also involved an overhaul to the code generating programs that transform the database into pages of HTML for uploading and use.

2004 was a year of activity at work and slow growth of the museum.

2005. By September it was possible to increase the museum to 1,700 exhibits and expand the equivalents system. At over 6,000 equivalents the computer processing load had increased to the point where the merging of the exhibits database with the equivalents Excel file now took place in four stages.

January 2006. Photoshop Elements was added to the photographic system. The move away from Photoshop 4.0 enabled the use of the 10Mb RAW file images from the E20 camera. This immediately raised the image quality and simplified the picture handling.

March 2006. The exhibits database had grown to 1,800 and the equivalents was at over 7,300 entries. The high resolution images exceed 680 Mb and the museum had outgrown the CD format for distribution. The first DVD edition was launched with 835 Mb of data and 14,141 files. The web stats showed over 5,000 separate visitors a week for the first time.

Summer 2007. With the kind permission of Frank Philipse, many of his PDF datasheets were incorporated into the museum. This has the twin benefits of enriching the museum and reducing the costs of his web hosting by reducing traffic. Many thanks.

November - December 2007. The equivalents database has been overhauled by comparing with the eighth edition of the Wireless World Radio Valve Data book. Wherever possible primary data sources are used ie manufacturers data, where this is not possible I have used contemporary secondary sources such as the Wirelass World publications and the AVO valve tester data books.

January 2008. 2113 exhibits. The equivalents database has been updated by comparison with the CV register including updates to 1969. This exercise took about 45 hours spread over the month. The database now has more entries and fewer blank spaces. The software has been revamped to include the datasheet references in the equivalents listings as well as the exhibits. The cross referencing files required made the processing less easy to understand and so the batch processing file has been altered. A free text file of program steps and file names plus uses is now converted to the operational batch control file. Additionally the programs write directly to destination directories as a new utility enables the file names to be changed to all lower case in a single operation. The web server is unix based and the file names are case sensitive.

April 2008. 2118 exhibits. Photoshop elements 6 is now in use and small changes to the way the program works have made significant improvements to the image preparation for the museum. The main camera remains the Olympus E20 but for macro photography a 60 mm Micro-Nikkor has been added to the Nikon D80 system. I had wanted a search facility from the start but how to achieve it remained a problem. Following a user's comment at the end of March I spent a day researching the problem again. Zoom Search provided the answer with tools for both on-line (uses the web server) and off-line for the DVD (uses Javascript). The graphics for the search logo were again designed by ToriArts.

Site Photography.

The photographs were always planned as central to the museum. Initially conventional photography and scanning of photographs was employed.

The camera used was a Canon T70 supported on a Slik 88 Tripod. The lens stack comprised:- 25mm extension tube, FD x2-B extender and FD 135mm 1:2.8 Lens. The lens was set to manual and stopped down to f16. Kodak colour print film ASA 400 processed and printed to 6 x 4 was chosen as negative stock. The images were scanned at 300dpi on a HP scanjet 5p flatbed scanner. A reasonable line-up for early 2000 but the results, even after six weeks of experimentation were less than inspired.

In April 2000 it was considered that a digital camera would improve image quality. Tests showed that the improvement would be dramatic. Mainly by cutting out the many transfer stages. A Sony DSC D770, 1.3 megapixel SLR camera was purchased. The sensor is complemented by large good quality optics. The image quality was excellent for close-in work.

The 'studio' was modified to take advantage of the close focus of the camera. An integrating cube was built from five expanded polystyrene ceiling tiles fixed into the inside of a tri-wall corrugated cardboard box about 20 inches on each side. It was found that with the artificial light setting on the camera, a single Tungsten Halogen 500 W lamp provided sufficient illumination for the lens to be stopped down to f9.5 for good depth of field. An exposure of 1/90 th of a second was used to minimise movement. The camera images were 0.75 Mb JPEGs and these were transferred to Photoshop 4.0.

The pictures are optimised to see through the glass to the inner structure. Unlike conventional photography where the labs try to average the scene to an 18% grey this process enabled the background to be burnt-out to white most of the time. The images were saved as JPEGs at best quality for the hi-res images for the CD. Only the excess white around the valve is cropped out. For the online version the images are reduced in size to either 800 pixels wide or 600 pixels high. The resultant image is saved as a low quality, small file size JPEG. These actions took the original 750kB image to approximately 250kB for the CD and 25kB for the online version.

In the Summer of 2002 an Olympus E20 Digital SLR was purchased for the museum. The 5 Megapixel sensor was supported by supurb integrated optics and built-in macro focus ability. The 1kW video light heated the camera too much and in early 2003 a used Bowens 400 J monobloc studio flash was sourced from ebay.

The potential of the system took many months to be realised. A new 'studio' was constructed out of white Contiboard and a sheet of photographic opal was attached so as to form parabolic background. As much of the direct light as possible was blocked so that the white box could provide a sensibly even illumination but still generate catchlights to bring the exhibits to life.

With the addition of Photoshop Elements and a remote cable release, at the end of 2005, the system reached a comfortable level of maturity.

Whilst on a short holiday to Cornwall at the beginning of January 2007 my wife and I made plans for some decorating. This involved turning the radio shack into a dressing room, reworking the valve store as the radio shack and converting the walk-in roof space that had housed our wardrobes into the valve store and studio. The shack took four whole weekends to empty. A clear sign of too much stuff. The valve room had 72 banana boxes full of valves as well as many smaller containers. The whole project took over six months.

November 2007 and time to commission the new studio. The light source is a pair of Bowens Prolite 60 studio flash monoblocs. The integrating cube is a small purpose made fold down light tent bought from an ebay supplier. The light inside the tent is a flat 26 watt compact flourescent lamp. This provides sufficient light to aid focussing. The camera on the stand is the Olympus E20. The monoblocs are fired by a wireless trigger. The scene was photographed with the new Nikon D80 camera.

The whole studio. On the left is a rack for holding trays ready for photography. The laptop connects to the cameras and transfers the data wirelessly to the main computer in the study. Its other function is to play music. As the studio was a wardrobe the doors close off the space when not in use. The large valve crate to the right acts as a stand for the working tray of valves. The obligatory mug of tea completes the scene.

Exhibits System.

Every exhibit is stored in a special purpose database system. This software was written as a free text database for commercial publishing applications in the mid 1990s and has been very successful. For the museum, specific data processing modules were built to generate the indexes and exhibits. Random access to each record is fast and easy through a control program. Each record is held as a text file and the editing is with Notespad. The latter being a very useful replacement for the Microsoft Notepad.

Almost all database management software sold is of the relational type. Free text databases have largely been consigned to history. Their advantages come from the freedom offered to the designer and the fact that any field can be pages of text in length. Our records use a tagged structure similiar to HTML and employ a mark-up language as well as a field structure. Fields can be added as required to specific records, if so desired. When PDF references were added in 2005 it was a matter of defining the tag and processing software and then adding the tag to just those records with PDF additions.

To enable classified indexing, a set of fields are recorded that use keywords that are chosen from a predetermined list. The list can easily be expanded by the database administrator. When the exhibit text file is closed the record is batch processed to check for structural integrity and that the 'look-up' fields contain valid information. Any departure from the keyword list is flashed as an error and the operator has to correct the record before the exhibit can be stored on disk. Manufacturers, base types, usage, age of design, heater parameters etc. are all held this way.

The database contains descriptive text in paragraphs, as many paragraphs as are required. A whole essay is possible in a single field. The mark-up language enables picture references, article references and cross references to other exhibits, to be added in shorthand form.

The electrode connections are stored as a set of sub-fields in a single text field. The maximum operating conditions are also held as sub-fields in a text field. In this case a field of sub-fields is recorded for each electrode system in a multi-part valve. No restrictions are placed on this part of the database apart from structure. In this way we can deal with four pin valves and 14 pin cathode ray tubes with the same tools.

When the web pages are made by the HTML page generator software, each set of sub-fields is automatically converted into a table for ease of reading. In the case of the pin information the table automatically includes the base type and a hot link to a help file.

One of the longest parts of the computer processing to produce a new update for the site is the double cross referencing that takes place. This adds hyperlinks within an exhibit to any equivalents that we have on the site and also populates the equivalents pages with hyperlinks if a valve listed has an exhibit. The equivalents are held in four spreadsheets and each segnent is processed by two programs. This is eight operations just for this phase of the work.

Equivalents System.

Each valve is entered into the spreadsheet that records the valve type, the electrode system, e.g. triode and any equivalents given in the source documentation.

For this project bespoke software has been written. The master equivalents file is manually compiled and edited to show valves that are equivalent or near substitutes. As an example the EL37 can be used as a replacement for a 6L6G but the 6L6G is not a replacement for the EL37, if it is used to its maximum capabilities. These differences are taken into account in the system.

For an update to the museum, the databases are processed to generate new HTML files. The cross references to valve equivalents are made dynamically in this process and, therefore, any existing entry in the museum can be updated in any revision to the site.

Part of the equivalents system is the addition of hypertext links to the exhibits database. When preparing a new update to the web site the exhibits database is interrogated for new and updated entries. This list is processed by a page generator, specially written for the project, that not only gives the exhibit pages but a list of cross references for the equivalents process.

During part of the equivalents regeneration the exhibit cross references are merged into the data stream. These are also multiplied so that every equivalent type is hot linked to the museum exhibit for that type of valve.

The National Valve Museum on DVD.

The DVD data is updated every time an amendment is made, and each DVD sold is made from the latest files.

The DVD takes the Internet site and adds a click through picture to a hi-res image. This image will normally fill the screen and shows many valves much greater than life size. The images are available for importing to other applications for printing, if required.

The DVD makes possible thumbnail images within the main idexes. This makes browsing much more enjoyable. At first the collections indexes were deemed too large to equip with thumbnail images as it took too long to load on a reasonable power computer. In 2006 the thumbnails were transferred to the on-line museum as well.

The R-type valve on the title page of the Web Site is an example of the DVD system in use. To use the DVD: insert into your DVD drive, use My Computer to bring-up the DVD drive contents and double click the Index file or icon. Internet Explorer (or any other browser installed) will open and display the opening page. From this point navigation is identical to the Web Site, but the pictures are clickable to the hi-res image.

For additional information: please contact your webmaster allan{@}virtual-museums.org