What's New

Back from Summer Fun - Grounding of PCBs

Well, the summer is over and time to get back to work!

A problem I see a lot in my work is designers not controlling PCB resonances caused by capacitance between one or more PCBs and chassis metal or another PCB and the inductance of the connections between them. Not controlling PCB to chassis or PCB resonances can cause emissions, radiated immunity, and ESD problems.

The trick is to control the resonant frequencies by adding, subtractiing, or moving PCB to metal connections or damping the resonances by putting some resistance, usually on the order of 20 Ohms, in series with one or two of the interconnecting ground conductors.

I am currently working with a client with just such a problem. In his case, it was an easy change. Before this change, the board was completely floating except for connections through a long cable back to the system chassis! The easiest way to do this is to ground a PCB at all mounting points to chassis metal, but allow at one or two (if two, as far apart as possible) mounting points the possibility of a resistor on the order of 20 Ohms in series with that mounting point to ground. The actual value of resistance should be equal to the reactance of the capacitance between the PCB and the chassis or other PCB at resonance (as measured).

Grounding the board this way has solved a lot of ESD problems for me too.

Some additional links to measurements of this kind can be seen in my Technical Tidbit:

http://emcesd.com/tt2011/tt090611.htm

Doug

Large Value Resistors Can Act Like Capacitors!

Do any of your designs use large value resistors, say above 1 MOhm? It is not unusual for analog designs to use large value resistors. I have recently seen a design with a 1000 MOhm resistor. The problem arises from the fact that all resistors have parasitic capacitance and it does not take much to effectively bypass a high value resistor at relatively low frequencies.

For instance, take a 1 MOhm resistor with only 1/4 pF of parasitic capacitance in parallel with the resistance (from end caps, paths on a PCB or descrete leads). The capacitive reactance of 1/4 pF equals 1 MOhm well below one MHz. Above 1 MHz the impedance of the resistor is capacitive and its magnitude is decreasing at 20 dB/decade.

For a 1000 MOhm resistor, the same 1/4 pF of capacitance begins to dominate the resistor's impedance below 1000 Hz! Layout of the circuit becomes very critical to maintain the design intent.

Such effects can throw of your calculated gain of an Opamp circuit significantly as well as affect the circuit in other ways.

Doug

Anodized Metal and ESD

I am in Kansas for a few days with our son in the Army at Ft. Riley, but was sitting here thinking about an ESD problem I was working on last week and thought it might spur thinking.

Having metal parts of a system that are close, closer than 1 cm for sure, is a dangerous physical design. Nearby static charges can induce a spark between the pieces of metal causing a very intense ESD event. Anodized metal creates a very small and intense spark to other pieces of metal. The small the spark, the more intense the higher the di/dt and resulting problems. ALL metal parts of a system must be bonded together.

For an example of a static charge inducing a spark between two pieces of unconnected metal see:

http://emcesd.com/tt2001/tt060101.htm

Have a good Summer!

Doug

The Funnies