Chassis ground and Digital Ground connection using different components

[FreshmanNewbie]

In this reference design schematic, the chassis ground is connected to digital ground using R22. A couple of questions on this.

1. What is the reason for having R22 and why it should be 1210 package and not 0603 or 0402?
2. Can I replace R22 using a capacitor (for connection between the chassis and digital ground)? What would the impact difference between using R and C?

Any typical capacitor value suggestion also will be helpful?

 

[D.A.(Tony)Stewart]

R22 is a jumper to remove Vcm line noise.
Make it a short connect with low resistance that won’t easily fuse open accidentally.

 

[FreshmanNewbie]

R22 is a jumper to remove Vcm line noise.
Make it a short connect with low resistance that won’t easily fuse open accidentally.

Can I replace R22 using a capacitor (for connection between the chassis and digital ground)? What would the impact difference between using R and C?

 

[FvM]

Consider:

Circuit Diagrams utilizing SMSC Products Are Included As A Means Of Illustrating Typical Semiconductor Applications: Consequently Complete Information Sufficient For Construction Purposes Is Not Necessarily Given.

Grounding schemes are selected according to overall instrument design requirements, not particularly LAN interface circuit. Non-isolated interfaces (e.g. USB) and power supply are more critical in this regard.

Can I replace R22 using a capacitor (for connection between the chassis and digital ground)?

Yes. A parallel resistor (Megohm range) is typically required to prevent charging up to high DC voltage by electrostatics.

What would the impact difference between using R and C?

Many effects, depending on instrument function.

 

[D.A.(Tony)Stewart]

The PHY network on RJ45 use the standard Cap to chassis ground for it's spectrum isolation and shunting CM noise to chassis with a CM choke.

This designer chose 0R

Another design might be 1 Meg. or just an RF Cap.

But you can tell, the designer has been very careful in the (DC power delivery network) or PDN design, so let us assume this was done for good reasons and making a change requires analyzing power delivery, noise from e-field susceptibility and peripheral conducted CM noise.


It cannot be analyzed in isolation and is usually about what you have not shown as this is just the network device.

When I was a Test Engineer at Burroughs , every ECN was reviewed by committee had to be analyzed and signed off for EMI risks, QA, Prod Ctrl, Mfg.Eng'ing besides the change done by EE so that one fix did not create a new problem.

Using a cap between a low impedance metallic shield and the DC 0V will have a cutoff where it is gradually ineffective at low frequency.

For the same reason, all desktop computers are hardwired between PE and 0Vdc to avoid mic hum, eliminate USB errors, and prevent VGA horizontal bars scrolling between line f and Video frame rate, even though all VGA cables have a fat ferrite CM choke on them.

Anything with an interface cable becomes an antenna for e-fields from line V/m. Impedance mismatch by

When multiple peripherals powered by floating AC are connected to a PE bonded PC desktop there is a high probability of USB data errors. This is due to SMPS leakage noise that becomes very high levels of CM noise. For USB it is the CMRR of a balanced differential amplifier and the resulting SNR which must be enough above the error threshold that causes a bit error rate (BER) in every message.

Even if you had an mic port with an INA analog spec of 120 dB of CMRR the limiting factor is the imbalanced CMRR of the cable impedance, the CM noise and the length of the cable.

I discovered this USB problem, delivering a servo stepper 1x1m gantry to U of T a while ago and that was the cause due to the interference of the AC charger on a laptop connected to a 12V Arduino supply floating and a PC desktop. It was for antenna pattern testing for 8kW WPT.

Removing the AC laptop charger was one fix. PE bonding the gantry was another.