Why touch screen voting machines have problems

Analysis: saving democracy from stupidity

SemiAccurate logoThere has been a lot of talk recently about voting machine problems, especially touch screen voting machines, and it has a very easy technical explanation. That explanation is crappy hardware, crappy programming, and electrical interference.

Electrical interference is one of the biggest problems with modern electronic devices no matter what they are, and it is everywhere. Modern chips, sensors, and most other electronics expect a very clean and steady electric power source. If you don’t have it, the problems can range from immediate and obvious to intermittent and subtle. You are also probably familiar with vacuum cleaners causing snow or static on analog broadcast TV or radio back in the day. That is electromagnetic interference (EMI), fluctuations in the power supply, or both.

Most modern electronic equipment is tested to see that it does not generate excessive interference and EMI by the FCC, various other bodies in other countries. Because of this, anyone making electronic equipment sweats the details to make sure their devices don’t cause problems, if they do, they can’t be sold in most countries. The combination of stringent oversight, penalties, and companies actually wanting to make good products ends up doing a pretty good job making sure a device does not cause either problem. If you look at almost any electrical device, you can see a little FCC logo on it, it looks like this.

FCC logos on two of three power supplies

FCC logos on two of three adapters

The flip side of the coin is how a device deals with EMI or dirty power from an external source, be it on the power line, or simply electrical fields radiating from a nearby object. High quality devices sweat the details, use expensive filtered power supplies, good sensors, quality electrical components, and put a lot of time into software tricks that can filter out any spurious noise that doesn’t get caught by the hardware. Cheap devices usually don’t bother with any of that, demos can be rigged, and that no name thingy from WalMart has no one to call in the case of problems anyway. Resistance to interference is one of biggest differentiators between higher priced and lower priced electronic devices, and it can make a big difference to the end user.

Getting back to touchscreens, electrical noise can have huge effects. Why? Mainly because modern capacitive touch screens work by putting out an electric field at one point and reading it back at another. When you put your finger in to the field, it changes the field strength and shape a bit and the sensor reads this as an object. We go in to the technology in a lot more depth when we covered the last batch of Synaptics controllers. Other types of touch screens use different types of sensors but they are all very sensitive and susceptible to EMI and power problems. Visually, it looks like this.

Synaptics touch sensor explanation

How a Synaptics touch sensor works

The fields that these controllers generate are very weak, partially to save power in mobile devices and partially because there are thousands of them packed in to your phone screen. If you make the field too strong, it can be read by a neighboring sensor and lead to incorrect results. For both reasons, you want to make the field as weak as possible, doing so is a win/win for the device and end user. Until there is noise.

As you can probably guess, thousands of weak electrical fields are very susceptible to both strong electrical fields and variations in the power supply. Fluctuating power levels mean fluctuating field strengths and that confuses the receivers. An old vacuum cleaner, refrigerator compressor, fluorescent lighting ballasts, and power lines running through a wall or in the room next door that you can’t see may put out uncontrolled electromagnetic fields thousands or millions of times stronger than your phone does. All of this is invisible and can lead to massive electrical interference in sensitive devices like a touch screen.

A high quality touch screen controller will filter out a lot of the noise on the power line, a lower quality controller will not. A good controller usually has high quality software to back up quality sensors. These may be more expensive but are vastly better at resisting interference than their cheaper brethren. The Synaptics controller linked above senses atmospheric interference and recalibrates itself many times a second to filter this out. In the same situation, cheaper controllers have big problems that tend to be immediate and obvious.

In the video below, I demonstrate how a cheap touch screen controller has problems with my phone, a Motorola Photon. It uses a cheap Atmel controller that is probably the worst touch screen controller I have had the displeasure of using in years. Luckily for our demonstration it is the perfect example of a bad touch controller. I also use two USB power supplies, a good one and a bad one. The good is a Cyberpower CPTUC01 car/home USB power supply, the bad is a generic Augustar device with no model number and no FCC certification that I pulled out of my parts bin. The software used is Synaptics TouchExplorer, specifically the Draw test. Please note that the software is a generic touch screen tester, the controller is not by Synaptics, a Synaptics controller would probably not have had the problems seen below.

A quick demonstration cheap touch screen controller problems

As you can see, the phone works fine and draws a smooth line when not plugged in. The tracking is precise, the lines steady, and the slide to unlock tracks perfectly. Although not shown, the other tests in TouchExplorer show similar results. When plugged in to the ‘good’ CyberPower charger, there is no change, the power supplied is clean and causes no problems. When plugged in to the ‘bad’ Augustar charger, the power is noisy and the Atmel controller does not filter this out. The result is choppy lines when it works, and it often doesn’t work.

On the touch test, not shown, the bad charger makes the phone unable to sense a single steady finger on the phone, it jumps all over the place. Unplug it and things track perfectly, plug it back in to the bad charger and it immediately goes crazy. With the phone plugged in to the Augustar charger, it takes many tries to simply unlock the phone with a swipe, and sometimes you cannot unlock it at all. The difference between a good power supply and a bad one when using a cheap touch controller is very clear, one ends up as a device that works, the other does not.

You can do the same demonstration with strong electrical fields instead of unclean power with similar results. Put your phone near some older fluorescent lighting ballasts, if they flicker you may have better results. Same with older motors, high current AC power lines, and old vacuum cleaners. If there is no FCC logo on a device, it has a much higher likelihood of causing problems with a touch controller. A good touch controller will usually filter interference caused by electrical fields out, a low quality controller will fall over like my phone did in the video.

Back to the original point of the article, touch screen voting machines. These machines are usually built by companies that win a low bid contract, not companies that build the best product. Sometimes a high quality company may win a low bid contract, but a quick trip through WalMart will give you an education on cost vs quality in commodity electronics even when comparing brand names the lines sold at different outlets have different component quality.  Low cost usually means low quality hardware but the inverse in not always the case. Worse yet, the voting machines probably didn’t have much attention paid to making sure the touch screens are resistant to noise, interference, and other electrical problems. When tested, it is unlikely that they had anything close to the scrutiny that a decent cell phone does, this is a low bid win after all.

It would absolutely shock SemiAccurate if there was anything more than cursory attention paid to making sure the touch screen controllers in modern voting machines have anything more than basic functionality. A quick scan of the news shows dozens of reported problems with touch screens used in today’s US elections. Some reports show that touching the area for one candidate registers a touch in a different location. Others show that some areas of the screen simply do not respond at all. Some have intermittent problems, others are repeatable. Does this sound like a familiar set of issues?

The US being the US, division and suspicion is the order of the day. People are quick to jump to the conclusion that the ‘other side’ has rigged the machine because a person has problems voting for their chosen candidate. As this story is written, it is about 6pm on the November 6, 2012 and there are lots of reports of touch screen problems happening for and against both political parties, and the rhetoric and confusion it brings is spreading rapidly.

Unfortunately for those looking to find a conspiracy behind the touch screen voting machine irregularities, the cause behind the overwhelming majority of the cases is much easier to explain. Cheap voting machines that use cheap touch screen controllers, cheap power components, little if any EM shielding, and precious little software to mitigate problems can easily cause all of the reported problems. Voting machines in the US come from a variety of manufacturers, vary greatly in quality, and have all kinds of publicly documented problems. One thing you can say about all of them is that it is unlikely that any of them have had the same scrutiny that most consumer electronics get, much less thorough analysis by an unbiased independent quality tester.

If you put one of these cheap voting machines on a power strip with a fan, refrigerator, or fluorescent light, you will probably be feeding it noisy power. If the building is old, the likelihood of old, stressed wiring goes way up as well. Same with placement, put it by a wall with power lines running behind it, and voila, a varying electromagnetic field that the user is not aware of. Does the floor below have fluorescent lighting? By a kitchen? You get the idea, unclean power and EMI is everywhere. Some of this is constant, some of it is intermittent, but all of it will wreak havok on a cheap touch screen controller. Anyone think that touch screen voting machines that win low cost bids use anything but the cheapest parts available?

If not, you very likely have your answer as to the source of the problems with touch screen voting machines in the US today. While it doesn’t preclude actual conspiracies, cheap hardware will readily explain the obvious problems of touching one area and having another reported. Those same machines that didn’t bother to guard against commonly known touch screen problems probably didn’t do a good job on the rest of the device either. Unfortunately for voters in the US, only the touch screen side is easy to notice, document, and reproduce. And now you know why.S|A

Author’s note: Like the article itself, the video above was done quickly as a response to current events, then redone later in the evening for quality reasons. We didn’t have time to make it all that pretty or overly scripted, so sorry about any quirks. You should be able to re-create it fairly easily with the right components, try it out for yourself and see.

Author’s note 2: YouTube kept rejecting this video when we tried to put it up with any quality. This version is the best we could get them to not dump, and we have NO idea why. Their feedback blows at best.

The following two tabs change content below.

Charlie Demerjian

Roving engine of chaos and snide remarks at SemiAccurate
Charlie Demerjian is the founder of Stone Arch Networking Services and SemiAccurate.com. SemiAccurate.com is a technology news site; addressing hardware design, software selection, customization, securing and maintenance, with over one million views per month. He is a technologist and analyst specializing in semiconductors, system and network architecture. As head writer of SemiAccurate.com, he regularly advises writers, analysts, and industry executives on technical matters and long lead industry trends. Charlie is also available through Guidepoint and Mosaic. FullyAccurate