Noise Levels due to fast running:

Track underlay and ballasting options for the new loft layout.

Note that the smaller pictures can be enlarged by clicking on the images.


The original loft layout had the undesirable feature that fast trains created high noise levels.

The original layout used 1/8th inch cork track underlay, which was chamfered using a sharp craft knife and then ballasted with fine/medium sized grey stone OO gauge material, bonded using diluted PVA adhesive. The baseboard is made from half inch chipboard, supported on a softwood frame.

Research on the internet revealed two possible methods to reduce noise levels:

1) Use a low noise track underlay such as the Woodland Scenics "Track Bed". This is a long life foam strip with pre-chamfered edges. The HO gauge version is 5mm deep and 45-50mm wide at the base.

2) Instead of using traditional PVA adhesive to bond the ballasting material, use a latex based glue such as Copydex. (PVA glue dries to produce a rigid stone "skin" bonding the rail to the baseboard. The stone skin acts both as a soundboard in its own right and as a coupling medium to transmit rail vibration into the baseboard, which can also act as a sound radiator. I came across several model rail forum entries recommending the use of an alternative latex based adhesive, which surrounds each ballast grain in a flexible rubber skin. This was reputed to provide quieter operation. 

This webpage documents the process followed to evaluate the above techniques, before starting on the new loft layout.


Test track sections:

Three small test tracks were used to evaluate noise levels and to explore assembly techniques.

1) The first test track was the prototype Code 75 rail test piece, built to compare Code 100 and Code 75 track. This used 1/8 inch thick cork underlay combined with traditional PVA bonded stone ballast. (This is identical to the original loft layout build standard, so it acted as the reference piece.)

2) The next test track used code 75 rail glued to a strip of Woodland Scenics foam track bed. Approximately 60% of this test track had stone ballast applied, using 1:1 diluted Copydex adhesive.

3) The final test track used code 75 rail glued to 2.5mm cork underlay. Again approximately 60% of this test track had stone ballast applied, using 1:1 diluted Copydex adhesive.

The test tracks use identical softwood bases and use identical ballast material (medium size Woodland Scenics grey coloured stone ballast material).

The test tracks can be connected together so that an item of rolling stock can be propelled by hand directly from one track to the next.


Test track 1


Test track 2


Test track 3

Showing 2 test tracks coupled together for comparison testing

Rather jerky view of the test process



It is not possible to provide a quantified set of results without sound analysis equipment, however there were clear subjective differences between the three test tracks:

The quietest section of track was the Woodland Scenic foam with no ballast cover. However this is not a practical solution for cosmetic reasons. Next quietest was the  Woodland Scenic foam with Copydex glued ballast. This was not noise free, but the sound characteristic was at a lower level than the other ballasted options and the noise spectrum appeared to cover a lower band of audio frequencies.

The cork underlay and PVA ballast glue track was the noisiest of the three track systems. The noise spectrum appeared to occupy a higher band of frequencies than the other track options. The higher frequency content added to the detrimental impact of the noise on the realism of the scene.

The cork underlay with Copydex ballast glue came roughly midway between the other two ballasted options in terms of noise level. The band of frequencies generated was similarly about mid-way between the other two options.

A graphical summary of the findings is attempted below:


General Conclusions

The use of a non-rigid-when-dry latex based glue such as Copydex, diluted to secure fine grain stone ballast, gives a worthwhile improvement in noise levels compared to results using PVA ballast adhesive. This fact should be better publicised in the model railway media!

The 5mm thick Woodland Scenics Track Bed provides a significant improvement on both the noise level and the frequency content of model rail running noise, when compared to 2.5mm cork underlay. It is reputed to have an indefinite life compared to the soft foam underlay produced by Hornby and Peco, which I understand bio-degrades, starting to disintegrate after circa 2 years of use.

The 5mm height of the Woodland Scenics material is probably prototypical for high speed main lines, but may require some camouflage on branchlines and through stations.


Decisions for the new loft layout:

The Woodland Scenics foam without ballast gave the best results. However, for visual realism, the Copydex fixed ballast is essential. Although the ballast made a small degradation to the sound characteristics, this combination offers the best compromise.  This approach will be used on the high speed sections of the new layout. The 5mm depth of the foam strip means that sheet material will be needed in the station areas to avoid the need for excessive quantities of ballast material in between the tracks.  Also platforms may need to be mounted on 2mm plasticard foundations to achieve the correct height with respect to the rails.

See below for the differences in profile of a high speed up and down line pair with 2.5mm cork and 5mm foam




Supplier website links:
Trains 4 U Peterborough A good (almost local) source of Bachmann and Dapol product. The Woodland Scenics track Bed and 2.5mm cork underlay used in the above tests  were obtained here. Always helpful and often an excellent source for advice on railway modelling issues.
Model Junction Bury St Edmunds A good (almost local) source of Bachmann and Hornby product. The Code 75 Peco track was obtained here.
Homebase DIY shops Copydex adhesive is available in half litre containers from these stores.

The photos of real track were taken on the Settle-Carlisle line at Ribblehead. The photos of test tracks were taken on the kitchen worktop at 200 ISO, using a tripod.    


Click to move to Model Rail Index Page