Suspicion of electronic vote tampering is widespread, and those suspicions are reasonable, if perhaps not immediately validated by recorded incidents (that I am aware of).
The history of election fraud, with its many techniques and variations, is pretty rich -- people have been trying to steal elections since elections were invented. We can safely assume the motivation for electronic vote fraud exists.
Even a properly marked paper ballot is susceptible to problems or tampering. They can be destroyed, hidden, faked, miscounted (intentionally or not)... so on, so forth. Or there can be technical difficulties, like the infamous Hanging Chads of the Bush-Gore election in the USA. But with proper physical security, transparency, carefully managed processes, etc. it is possible to have an election result that most participants (and election monitors) are satisfied is reasonably legitimate.
Electronic voting machines are a relatively new voting mechanism, and they introduce several additional layers of complexity, reduce transparency, and they rely on proprietary private systems. Every one of these factors is reason enough for extra caution/suspicion.
Setting aside intentional tampering for a moment, some potential flaws with electronic voting machines may stem from: programming errors, data corruption, non-obvious component failure or misalignment (such as a touch screen not responding appropriately or being miscalibrated), data transmission issues, power outages... you get the idea.
As far as intentional tampering, various models of electronic voting machines have been shown to be pretty easy to mess with if you have the know-how. Plenty of talk about this is easy to turn up with some searching, but here's a blog post by Symantec from a little while back discussing their findings and thoughts. Where paper ballots generally take time, effort, manual forgery, etc to tamper with, electronic ballots are potentially just as easy to alter/insert/delete as any other piece of electronic data, and if those changes are not logged, there would be no trail to follow. Even worse, the proprietary "black box" nature of these voting machines prevents a larger pool of security researchers from looking for weaknesses and recommending changes. No system of digital security is foolproof, and given the above (and the many proof-of-concept examples discussed by security firms around the internet), we know the technical means for electronic vote fraud exist.
Major elections are high-stakes situations for all involved; if there is any plausible reason to doubt the outcome, especially if you're the party that lost... the accusations will fly.