Automated tests


Does anybody know a way to do some automated tests? I mean i would like to test the application that i wrote in some automated way because i suspect a memory leak that i can’t find or something wrong with the logic because for some reason after some undefined time the application freezes and needs to be reset


Write a program or script that sends the appropriate stimuli to your unit, and leave it with the TMT logging trace information…

There are many companies that make many tools and facilities for automated testing - you need to be a lot more specific about your requirements!

I use Procomm-Plus for scripting serial port stuff - but it’s old and unsupported now :frowning:
SecureCRT look like a suitable modern equivalent:

LabView and HP-Vee are names that spring to mind: … 99soft.htm

HP-Vee is now Agilent-Vee: … kt=1&k=vee


My application reads gps data and puts in on a ftp server and somehow it ends up in a state in witch it doesn’t send data anymore although it has gsm signal and gps signal. These doesn’t happen on a regular basis and when it does i need to reset the device(sometimes with a hardware reset). So i would like to be able to simulate(if possible) the gain/loss of gsm signal and run the application in this condition. I don’t think TMT could be a solution because this error occurs sometimes after a week so it not a viable solution. I need something that can simulate a week of running application in a reasonable time so i can see the error


How do you know that?

To simulate loss of signal, you need to get an RF switch, and switch between your antenna and a 50-ohm termination.
If you get an electrically-operated RF switch (or a “coaxial realy”), you can obviously control it automatically.

You can simulate low GSM signal by inserting an RF antennuator in the antenna connection;
but this only simulates a steady reduction in the signal level - not the fading and interference found in real life…

Why is that not viable? Just leave the TMT running and logging for a week!
If the problem takes a week to manifest, then you’re going to have to leave something running for a week, aren’t you? So why can’t it be TMT?

That is almost certainly impossible, I’m afraid: this is going to be related to real-time, real-world events: you’d need to know what those events were to be able to simulate them - and if you knew that, you’d probably know the answer already!

The only other option is to build some sort of logging into you application so that you can retrieve the logs after the problem has occurred - then look for clues in the logged information.


For another post on this topic, with illustrations - see: … ife#p29005