High Speed Digital Communication Bus Probing
For high-speed digital communication busses, what attributes do you generally advise to add to a PCB to help with validating the communication bus as well as troubleshooting the design? In particular, I'm interested in adding features to a PCB with a parallel bus for SDR SDRAM with a clock of 167MHz, and a quad SPI bus for Flash memory with a clock of 133MHz. The rise/fall times are roughly 1ns. I suspect I will need to use an active probe to monitor a given signal with an oscilloscope. Is it realistic to probe multiple lines of a parallel bus to validate relative timing and signal integrity of the signals? Are there any features on a PCB that are helpful for holding the probes, without adding significant inductance, so that you can use your hands for interfacing with the test equipment. Other than an oscilloscope, what other instruments are useful for probing high speed digital busses?
1 answer
One of the first things you should do is provide a connection for the scope probe ground. If you think you'll be regularly looking at signals on this board, then one of those little loops or pins to clip a scope probe too would be useful. If it's a one-off problem, then just a thru-hole pad would be good enough. You solder a small ground wire to the pad, then clip the probe to the other end of the wire.
The high-speed signals are different. At these speeds, you are probably delay-matching the data lines. You want something that doesn't interfere with the delay. Unfortunately, there is no such thing. In this case, looking at the signals will affect them.
I might add one of those clips to the clock line only. That way you can see the clock signal hands-free. Then manually hold the probe for the other channel to whatever data line you are interested in. I'm assuming you can arrange for particular read and write sequences to eventually see the skew of every data line relative to the clock.
Be prepared for the skew being a little different when the scope probes are removed.
0 comment threads
0 comment threads