In a message dated 3/25/2003 4:11:56 PM Central Standard Time, [email protected] writes:
most interesting data. I have two questions: What software are you exercising the PC's with and is the spread spectrum enabled? I assume you are using an RBW of 120 kHz and using Quasi-peak detection. Hi Dave, the PC's are not Spread Spectrum based. What makes me say this is that in real time, the discrete frequencies are rock solid. I don't see the modulation that an SS clock would show. The RBW is 120 kHz. The PC's were tested first with a DOS based SW, that was written by my client to excercise his card ( It's a 100 base T Ethernet card. Then, once I'd pulled his card, the PC's were booted to windows, and I left them at the desktop. I did not bother with scrolling H's. I have friends at Matrox and Nvidea that have suggested that is a bit bogus therse days, so I really have not used that. Besides, pulling the CRT ( LCD and CRT ) didn't make a huge difference. The LCD though I believe is a better monitor ( my back likes it better ), and is usually quieter if I use it as a second monitor on a laptop. In my limited experience I have found that the software that is exercising the PC can make quite a difference. It is my understanding that many PC's are tested with H's printing to the screen. When running a game such as Doom the emissions will go up several dB. And if the spread spectrum is not enabled there will be an increase of 8 dB or so. I'm curious as to what the failing frequencies are. The 2 mainstream PC's had emissions from 30 MHz to 50 MHz. I attributed this to the power supply, it looked just like BB noise from reverse recovery times. These two PC's looked good over 100 MHz The clones touched the spec line to about 150 MHz, then, discrete spurs could be seen all the way past 1000 MHz. I stopped at 1 GHz. From memory I don't recall wht they were, but a quick guess would be about 30 MHz spacing. Don't hold me to that. Cheers, Derek.
