It is not recommended to add solvent to UV dielectrics to thin them. Because UV dielectrics cure so quickly when exposed to UV light sources, the solvent will be trapped and migrate slowly out of the material over time, causing problems.

Too many people rely on the setting that regulates the power to the lamp (typically100, 200 and 300 watts) without measuring the energy at the belt surface. The watt setting is only a measure of the electrical power going to the lamp to power it up. It is essential to invest in a small radiometer (sometimes referred to as a “hockey puck”) to check the actual output of the lamp. These radiometers have a series of cells that will read the intensity of UV energy coming from the lamp. The radiometer is turned on, placed on the belt and run under the lamp. When it exits it will have a numerical reading on a digital LCD. This reading is in millijoules per square centimeter (mj/cm2). Typically the reading will be in the range of 400 to 900 for a UV curable material.

One thing to keep in mind is that the radiometer will measure the overall output of the lamp across the entire range of UV energy, and will not distinguish the differences in wavelength. The general range for UV radiation on the electromagnetic spectrum is wavelengths between 200 and 400 nanometers. Below 200 nanometers starts the x-ray range, and above 400 nanometers starts the visible light range. A UV lamp will give a “spectral distribution” of different wavelengths of UV energy, much like a unique fingerprint and this spectral distribution will change as the lamp ages. Below is a representation of the spectral distribution of a typical fusion type UV lamp, showing the lamp output at different UV wavelengths.


One other essential factor is that over time, a UV lamp will degrade and the UV spectrum output of the bulb will begin to shift. Keep in mind that the photoinitiators in a UV cured material need to be exposed to specific wavelengths of UV light in order to start reacting. As a lamp ages and the UV spectra shifts, the photoinitiator will not react as readily. One common mistake is that as a lamp ages, an operator will continue to adjust the belt speed to get the specified reading on the radiometer, and eventually there will be problems caused by the UV material not curing properly. Even though the energy reading on the radiometer may be the same on the lamp when it is new and the aged lamp with the adjusted belt speed, it represents a different “spectral distribution”.

A simple analogy to explain this would be to picture someone standing on a tall ladder holding a rock in one hand that weighs 10 pounds, and a bucket with 10 pounds of sand in the other. If the person drops the rock, and pours the sand out of the bucket all at once, they will both have the same kinetic energy when they hit something, but would you rather be standing under the rock or the sand?

Another consideration is that different types of UV bulbs, such as mercury vapor and fusion, will have different spectral distributions so they each work more efficiently with different photoinitiators.

When setting up a UV curing system, a general recommendation is to set the lamp at the highest power setting (usually 300 watts), and vary the belt speed to control the output of the lamp, as measured by the radiometer. Regardless of curing efficiency, it is a good practice to change lamps after a certain number of hours as recommended by the manufacturer.

Focal length of the lamp is also important. With a focused lamp, the distance from the bulb to the substrate is crucial because the reflector behind the lamp focuses the light to a specific distance, much like when you focus a camera lens to sharpen the picture. If the bulb is too close or far away from the substrate, it will affect the energy reaching the substrate.

The first thing is to be sure that the dielectric is completely cured. If the dielectric feels tacky or sticky on the surface, it is not cured completely. However, it is possible to have the dielectric not cured completely, but still not feel tacky on the surface. The simplest way to evaluate degree of cure and how well the dielectric sticks to the substrate is to perform a crosshatch tape adhesion test and a hard crease test.

Cross hatch tape adhesion testing can be performed simply by using a small razor blade to cut a cross hatch pattern into the dielectric and substrate, but not cutting through the substrate. The cross hatch pattern is made of lines about 1/8” apart. A series of lines (6 to 10 usually) are cut in one way, and then another series of lines are cut 90 degrees to these lines, intersecting them so that you are left with small squares of material. Then, tape is applied to the cross hatch pattern and lifted off quickly. The dielectric should remain attached to the substrate. If only one or two squares lift off, this is not a cause for concern.

One thing to be cautious of is that there are some additives in dielectrics that are used to improve flow characteristics and prevent formation of bubbles. These additives will also migrate to the surface of the dielectric and create a layer that will not allow tape to stick very well to the surface of the dielectric. This would result in a “false positive” test for passing the cross hatch tape adhesion.

If two layers of dielectric are printed on the tail area of a circuit, the cross hatch tape adhesion test will indicate if the two layers are bonded together well. If the top layer of dielectric lifts off of the bottom layer in a tape test, this is an indication that the bottom layer of dielectric was cured completely before the top layer was applied. When printing the first layer of dielectric, the UV belt speed should be increased so that the first layer does not cure completely. This will leave molecules of dielectric on the surface that are unreacted. When the second layer is printed, the belt should be slowed back to normal speed. Now when the second layer crosslinks, molecules of the second layer will tie in with molecules of the first layer, and the layers will be bonded together well.

When a circuit is folded to a hard crease, you should not see the dielectric separate completely from the substrate beyond the line of the crease. If it does, it could be an indication of the dielectric not being cured completely, or incompatibility between the dielectric and the substrate.

It is always recommended to print two separate layers for crossovers so that if there are any thin areas or pinholes in the first layer, the second layer will cover them and prevent shorts.

Each layer of dielectric should be between .0005 and .0006” thick, for a final combined thickness of .001 to .0012”. If the final thickness is any thinner, then any small silver bumps in the ink trace underneath can short out with the crossover ink on top.

Printing a single thick layer is not recommended. A single printed layer .001” thick or more will have difficulty curing. This will cause curing issues with the dielectric, and the potential is increased for shorts from voids or thinning areas.

While crossover shorting is not a concern in the tail areas, it is recommended that two layers of dielectric be used here. The provides an extra margin of safety if the tail is abraded during handling and final assembly, exposing the silver ink traces underneath to potential shorting to metal housings or components on the final assembly. Care should be taken to assure that there are no bubbles, pinholes or “fisheyes” as a result of spot contamination formed when the dielectric is printed on crossovers.

The first thing to do, when you find holes, is to determine what type of hole or void you are seeing. It is normal to see an occasional small entrapped air bubble in a printed dielectric – especially glossy dielectrics. If there are lots of air bubbles, or pinholes that appear as trapped bubbles that have burst, then some of the printing process parameters need to be investigated. The first parameter is squeegee speed and pressure. Printing too fast and with high pressure will form bubbles in the dielectric as it is printed.

The main culprit for bubbles or pinholes is screen snap off or offset. If the offset distance from the substrate is too little, then the entire mesh surface will be lifted out of the printed layer all at the same time and leave behind bubbles. In extreme cases, this will show up as an almost perfect pattern of lines matching up with the mesh.

If the holes appear as larger craters, sometimes called “fisheyes”, then the likely culprit is either contamination of the dielectric or substrate surface, or incompatibility between dielectric and substrate.

Whenever attempting to get a wet material to stick well to a substrate, it is important to get the material to “wet out” onto the substrate. Picture drops of water flattening out on a car that has not been waxed for years versus those same drops of water beading up on a car that has just been waxed.

If spot contamination gets onto a substrate or into the dielectric, the dielectric will tend to run away from the contaminant during printing, much in the same way that the water beads up on the waxed car surface.

One of the most common causes of contamination is solvent from screen cleaning. If screens are not dried completely, the ink placed on the screen will pull the solvent into the ink. This can lead to fish eyes in the print layer at the start of the run.

Other potential causes of contaminants are airborne dust, oil from fingerprints or hand lotion on the substrate, lint from clothing and small pieces of emulsion breaking off from the screen over time. When using lubricating oil or penetrating oil (such as WD-40), care must be taken to not allow the material to become airborne in the area where screen printing is taking place, or where it can get onto substrates.

If there is an incompatibility between the dielectric and the substrate, the dielectric will not flow out as well and will tend to start forming small craters. In almost all instances, if there is a problem with incompatibility between dielectric ink and substrate, it will show up as an adhesion failure during tape testing after printing.

It is not recommended to reuse dielectric. Once it is on a press it will pick up contamination. If a dielectric is reused, it should be collected and stored in a separate container after filtering it through a filter or mesh, and then added in small amounts to fresh material on the press.

This is one of the most common modes of failure in a membrane switch, and there are many potential causes. The first one is that the dielectric is being printed too thin. Recommendation is printing two layers for a combined .0012” to .0014”thickness. It is also possible that the dielectric is not cured completely. Sometimes when this happens, a failure will show up later, after the switch is exposed to humidity or other materials that might attack the dielectric because it is not strong.

Another area to investigate is completeness of drying of the silver ink layers, especially the bottom silver ink trace. If it is not dried, the solvent remaining in it will slowly migrate, over time, out into the dielectric that has been printed on top of it, weakening the dielectric and allowing for the possibility of an eventual short with the top layer of conductive ink.

Pinholes, fisheyes or voids in dielectric pads for crossovers are another potential cause for shorts – especially if the dielectric is being printed too thin to begin with.

When investigating potential causes for crossover failure, one important question is whether the failures are sporadic, in random patterns, or specific to a particular area on switches.

If they are located in the same areas on switches, there may be a defect in the screen itself. It is also possible that there is something inherent in the circuit and screen layout itself that contributes to the problem. There have been instances where crossover failures occur in areas where the squeegee runs parallel to the edge of a conductive ink trace when printing the dielectric layer. As the squeegee bumps up to the edge of the silver ink trace, it causes starving of material on part of the crossover dielectric pad. If failures occur close to the edge of a print pattern, it is possible that screen tension differences contribute to the problem, or the print pattern may be too close to the frame so that a proper screen snap off is not possible after printing.

One of the biggest contributors to cross over failure is a switch design that allows for long sections of crossover, that run parallel with and on top of the silver ink trace below. This is really tempting fate by increasing the odds that a printing defect will allow for a small area to create a short. Crossovers should occur at a 90 degree angle with the silver ink trace below, to minimize the area required to complete the crossover, and minimize the chance for something causing a failure.

When testing switches for high humidity and temperature exposure over a long time, it is even more crucial that the dielectric crossover patterns be printed at the correct thickness, without pinholes or voids, and be crosslinked completely.