How to get rid of noise during digital line transitions

[VinceD]

A spike in the ground line occurs when there is a transition from high to low in the digital lines.  The peak of the spike increases as the number of lines that go from high to low increases.  The spike is large enough to alter the functionality of my circuit.  The spike in the ground line is observed directly from the SCC-68 junction box.

[K C]

Hi Vince,

A few questions for you:

What kind of cable are you using

What is driving the cable

Do you use twisted pair

What is at the end of the cable

How did you measured the spikes

Depending on above, with a series resistor and/or a proper termination could help

[VinceD]

Hello KC,

Thanks for responding.  Here is my setup.  I send 20 lines of digital signal using a PCI-6259 to a SCC-68 I/O connector via SHC-68-68 Shielded cable. After sending the digital data, I acquire an analog signal from one of the inputs. Data acquisition is "synchronized" such that the rising edge of the ai/sample clock sends the digital data, while the falling edge acquires the analog data.  Clock frequency is in the order of 100 to 500KHz. The digital data is more or less a counter. Whenever there is a large number of digital lines going from high to low (eg counter values 255 to 256 or 511 to 512), there is large spike in the ground in the order of 500 mV.  The height of the spike is proportional to the number of lines going from high to low, eg: 15 to 16 will have a lower spike compared to the transition from 63 to 64.  The problem worsens if i have a transition from 1023 to 1024 (or higher) and the spike can sometimes go up to 1 V. I can measure the spike (~20-40 ns) directly from the ground of the SCC-68 connector.

Hope this gives a better description of my problem.

VinceD

[K C]

Hi Vince,

Well this helps but it does not answer all of my questions and I have a few more.

Please go over the question one-by-one and try to answer them.

Additional question:

How do you measure a spike at the ground level i.e. with respect to what

Message Edited by K C on 10-24-2008 08:36 AM

[VinceD]

What kind of cable are you using?

NI SHC-68-68 as mentioned in the detailed setup described earlier.

What is driving the cable?

NI PC-6259 sends digitial signal to the SHC-68-68 to the SCC68-i/o connector as mentioned in the detailed setup described earlier.

Do you use twisted pair?

No. I use NI SHC-68-68 as mentioned in the detailed setup described earlier

What is at the end of the cable

The end of the digital lines drives two CMOS IC's(74HC573 - octal latch).

How did you measured the spikes?

The spikes were measured using an oscilloscope with the postive and negative terminals connected to the ground of the SCC-68 connector.

[K C]

Hi Vince,

 

Sorry for all the question but I just want to get a complete picture.

 

What you can do/try:

 

Be sure that you use all the ground lines in the cable and connect them all together at both sides.

 

Add a resistor of 10k at the HC573 inputs (signal to ground)

  

Measure the signal at an input of a HC573 before and after adding the resistor

Maybe you need to do more but first see the result of this action.

 

Message Edited by K C on 10-27-2008 09:35 AM

[VinceD]

Hello KC,

Thanks a lot.  If I understand your solution right, I tried to place pull-down resistors (18KOhm) before the 74HC573 inputs.  The PCI-6259 data sheet (http://sine.ni.com/ds/app/doc/p/id/ds-22/lang/en) suggested a 20K to 50KOhm pull-down resistor.  The resistors did not affect the spikes.  I took a picture of the spike (see below) and its the same spike with or without the resistors.  The pink line is connected to one of the digital lines while the green is probing the ground (500mV/div and 100ns/div). Ill try it with 10K later.

Vince

PS. I also removed all the ICs connected to the lines but the spikes are still there.

[K C]

Hi Vince,

The pull-down resistor specified in the datasheet is the resistor present on the 6259, I think.

The signal looks OK. I still do not understand how you measure the spike. Between ground and ???

The IC's will no have any (or very low) influence on the signal.  

[K C]

Hi Vince,

Just spent some more time to read your previous post.

How do you wire the IC from your SCC-68 box ?

Depending how you measure the spike it maybe possible that you look at the wrong place for the problem.

How does the signal at the inputs of a HC573 looks like if you switch more inputs together (i.e. the fault situation)