Comparison: LED, Standard-Neon / Cold Cathode

and HOTFIL-tubes

back to main page

Author: Marcus Thielen

This comparative report has been published first in 2003; the present version has been edited and updated in 2006 (see attachment).

A performance comparison of the three lightsources

Lightsources used for electric signage today mostly employ Neon (= cold cathode) tubing with metal shell electrodes, further referred to as Standard Neon. Since a few years LED's ("Light Emitting Diode") are used more and more for the illumination of channel letters.

Much less used for the same purpose are up to today hot cathode tubes. Their handcrafted manufacture requires skill and very expensive vacuum processing ovens, but have been proven reliable for long as industrially mass produced straight units called "fluorescents". Hot cathode tubes with hot filament electrodes for architectural and sign applications are available now as handcrafted items according to a new, patented process which does not require a vacuum oven and sold under the brand "HOTFIL Luminous Tube System".

Luminous efficiency: comparing apples and peas?

Some manufacturers and distributors of LED's have spread uncertainty in the sign business -at least in the beginning- by irritating slogans and vague comparisons. For example: "Our LED system consumes less than a quarter of the energy than a comparable neon system".

In these comparisons often only one parameter (i.e. the energy consumption) is compared, while other important parameters like light output, efficiency etc. will not be mentioned.

The main variable in the comparison of lightsources is the luminous efficiency, i.e. the luminous flux emitted per watt of electrical consumption, stated in "lumen per watt" as a measure of efficiency of the light generation.

International standards on light source measurements have been established first in the years 1931 and 1948. Even today many numbers are published which have been obtained not according to these standards (and therefore they are not comparable!). In addition, only lightsources with similar spectral distribution can be compared correctly, as i.e. the sensitivity of the human eye is different for a dark red than for a yellowish green (see for example DIN 5031 part 3).

At the end of the year 2001 first data on the luminous flux of LED, determined according to modern lighting standards has been published by the manufacturers. Then it was possible for the first time to seriously compare LED's with other lightsources (see Table 1).

Only lightsource, without losses in resistor or power supply; data state Aug. 2001. Up to today, June 2003, laboratory samples from different LED manufacturers have been presented with improved values, but data for mass production components have remained nearly unchanged (see: USENET:

Top of page      back to main page


Sample calculation 

From the physicist's point of view three illumination systems shall be compared on hand of a sample sign. The comparison presented is restricted on the lightsources itself, as the number of light technical parameters from the lamp's surrounding are so numerous (like channel letter, character geometry, inner surface and cover material), that even with the use of complex mathematical models the results never would be universal to other installations. Further the development on the semiconductor market is so fast that recent technical data from the manufacturers and the resulting comparison can be invalid in short time.1)  

A calculation onhand the example of an illuminated channel letter sign shall be carried out parallel for Red (best case for LED) and for Green (best case for Neon). The results for other colors will be inbetween these two values. We use for comparison the lettering "MOTEL" with a character height of approx. 60 cm; in total 7.6 m neon tubing in 6 tubes, 22 mm tube diameter, operating current 100 mA (or 140 mA; see footnote b)  in table 2). The light technical data on standard Neon stems from measurements of the G. Ferrara Institute, Milano, published in the catalog of TecnoLux, Desio, Italy; the data on LED stem from data sheets of Teledyne, Palo Alto, U.S.A.

1)  (Note: See Appendix)


Table 2: sample calculation

a) Mostly three LED's (or groups of three LED each) are mounted on a board together with a resistor for current stabilization to make a module. The modules are operated on 12 Volts DC and have a current draw of 30 mA/module, so each module has a power requirement of 0.36 Watts.  

The 12 Volts DC supply (assumed to be of the switching type) has an average efficiency of 0.97, therefore the power drawn from the mains supply line for each module will be 0.371 Watts. Due to the availability of Neon in almost any spectral distribution an optimum match to the transmission of the cover material can be achieved - as long as not simply "daylight white" is put behind any colored acrylic face. LED's are available only in a few colors with a narrow spectral range, which will lead in case of narrow band transmission of the acrylic materials to considerable, additional losses - further reducing the efficiency. In addition, the pointsource characteristics of the LED makes a stronger scattering action of the face material necessary to achieve a uniformly lit surface. All these additional losses will reduce the total efficiency, but considering them here would exceed this article by far.

b) As filament electrodes with neon discharges do not show a high life expectancy, it is recommended to use an Argon/Mercury discharge with the Phosphor "Rot/2" (Ciba special chemicals, Heidelberg/Germany) or "special red 99" (Tecnolux Italia) for illumination of red faced signs; in this case the operating current needs to be set to 140mA to achieve the same luminous flux as with a clear glass neon discharge operating at 100mA. The luminous efficiency is taken from data published by the phosphor manufacturer.

c) The luminous efficiency of a module is reduced by the losses in the resistor when compared to a single LED. To achieve the same luminous intensity (i.e. luminous flux) in the installation as achieved with neon, the number of required modules is calculated by dividing the total luminous flux of the neon sign by the flux of a single LED module.

d) Determination of the effective operating voltage according to Wickman's gradient formula for 7.6 meters of tubing, 22mm diameter in 6 tubes. Instead of cathode fall of 250 Volts / electrode pair for Standard-Neon for HOTFIL-hot filament electrodes only 15 Volts/pair are necessary and have been taken into account.

e) The ferromagnetic neon transformer has a typical efficiency of 0.90, the power drawn from the mains line is accordingly higher. Here it is assumed that the neon installation will be operated with an existing high voltage transformer and the efficiency of an electronic power supply is 0.97.

f) The luminous efficiency h is calculated from the total luminous flux  fS and the total power drawn  WS to:

Top of page      back to main page



For correctly manufactured high voltage neon tubes an average lifetime of approximately 20 000 to 30 000 hours has been proven, with more than 80 000 hours being no exceptions. During this time, in tubes without fluorescent powder the luminous flux remains almost constant, whereas in coated tubes the intensity can be reduced due to aging of the fluorescent material. Classical sign green ages fastest, after 10 000 hours the intensity is reduced to almost 25% of the initial value; modern green rare-earth phosphors show more than 70% intensity after more than 15 000 hours. Therefore here the useful life of the standard green powder (Zn2SiO4:Mn) tubes is assumed to 10 000 hours, the life for clear red neon tubes with 30 000 hours. For hot cathode tubes, the proven lifetime depends on the power supply and operating mode (EPS, choke, number of starts) and can be taken 10 000...40 000 hours. For the HOTFIL-System here the worst case is assumed. 

Some chain stores in the U.S.A. which had been using LED's for channel letter lighting, have changed back to Neon after about 1 to 1.5 years as the light emission was reduced to an unacceptable level. 

As investigations have shown g) even short time temperature rises to ambient temperatures of more than 65 °C can damage the LED permanently. The light emission is reduced thereby within a few weeks by up to 70%. This effect is based on diffusion within the semiconductor crystal of the LED and depends on parameters partly unknown today. After repeated experiments have been carried out it can be taken for sure that the often advertised lifetime of 100 000 hours for LED can not be reached in outdoor installations. Realistic seems a value - corrected in the meantime by some manufacturers- of 4000 hours (white), 5 000 hours (blue) up to 10 000 hours (red). 

g)  Tim Brosnahan/Steve Kieffer, Signs of the Times, Cincinnati, U.S.A.; August 2001, S. 85ff; Experiments commissioned by
   the International Sign Association ISA, U.S.A.


Economical estimate

A daily operating time of the sign of 12 hours per day will give a total operating time of 4380 hours per year. Here, the investment and operating cost for 5 years of usage (equivalent to 21 900 operating hours) shall be calculated for LED and Neon onhand the sample calculation. This does not cover the additional cost for channel letters, metal work, transformers/power supplies and installation.  


Table 3: Economical estimate

In this comparison the highest economy for illuminating channel letters is achieved using custom made hot cathode (hot filament) lamps. This counts especially for signs with high illumination levels, as for luminous tube applications in architectural lighting, as it is seen on modern buildings more and more.

Top of page      back to main page


Appendix to comparison above

Author: Marcus Thielen;  Duisburg, October 2006

Recently I have been asked if the development of new types of LED's would have improved the situation pictured in 2003 (see above). As in every field of semiconductor manufacture, also LED's are constantly improved. So I searched different sources for new products and new data sheets. The results have been compiled as addition to the existing comparison.

In general, it can be stated (especially due to my work as independent expert judge) that...

1) ...the new "high power LED modules" are less efficient compared to standard LED's; 

2) ...especially in "high Power LED's" the current through the component determines a reduction of lifetime; thus the lifetime figures given by the manufacturers are normally only achieveable with slightly reduced operating current (and thus reduced light output!). The proposed cause is electromigration, i.e. material flow along the semiconductor heterostructure of the P-N junction; investigations are currently undertaken, results not yet available.

3) ...the internal operating temperature of the semiconductor crystal is specifying the light output decrease over time, and in most cases causes the LED to fail prematurely. The encasing plastic material is a bad thermal conductor, so at ambient temperatures inside a sign cabinet, the (even small amount of) heat generated by the LED can not be dissipated and the semiconductor inside overheats.

4) ...very different qualities of similar LED's have been recently put on the market, it is almost impossible to compare. Especially with low-priced white LED's, intensity and color of LED's vary so far that it is impossible to obtain a uniformly lit sign surface. If the manufacturer does not offer this, it seems to be required to individually select every single LED to match brightness and color. 

5) ... the fluorescent materials used for Neon and hot cathode neon also have undergone some improvements in efficiency, but they are as small as 6-7% of their average value (based on the efficiency in the year 2003) and so will be neglected in favor to the LED as the coating and lamp manufacturing process can have a similar effect.


The efficiency of newer LED modules (standard power level) from different manufacturers - where serious information could be obtained- are as follows:

Table 4: Efficiency comparison 2006

1) Data sources: (N) Data sheet of Nichia Chemical Industries (Japan); (T) catalog of Toyoda Gosei, Japan 

The given data is for the bare lamp only and does not include any current stabilizing or power supply losses; here the same calculations as in the 2003 statement apply.

Still, even from large manufacturers, hardly comparable total luminous flux data can be obtained, and many efficiency values stated stem from laboratory experiments. Here are only guaranteed values taken from product data sheets. 

Summarizing, it can be stated that for any color except red, Neon and especially the HOTFIL-Neon is much more efficient than the LED. 

This does not take into account the thermal problems, decay and discoloring of the (white) LED's during lifetime. LED's definetely have it's application in the sign industry where the spacings do not permit the use of Neon; but for backlighting coves, signboxes and channel letters, Neon still is the most efficient way. Future will show if the fast semiconductor development can reach or surpass the technology of gas discharge lighting proven and developed in almost a century now.


Top of page      back to main page