best practice - Using a human finger to test a circuit

In this Linear Tech App Note 13, page 28, Jim describes:

A final form of probe is the human finger. Probing the circuit with a finger can accentuate desired or undesired effects, giving clues that may be useful. The finger can be used to introduce stray capacitance to a suspected circuit node while observing results on the CRT. Two fingers, lightly moistened, can be used to provide an experimental resistance path. Some high speed engineers are particularly adept at these techniques and can estimate the capacitive and resistive effects created with surprising accuracy.

Was he joking? If not, how do the such techniques work in practice to provide accurate estimation? Could anyone describe an actual scenario that those engineers applied such techniques?

The application note, and my question is in regards to low voltage, high speed signals, not mains or high voltage.

Answer

A finger is a very useful item of test equipment. The main problem when fault finding is knowing which bits of the circuit are susceptible, and which bits are OK. As you stare, dumbfounded, at a recalcitrant bit of board, it's good to be able to get any handle on what is going on, just a hint of where to start looking more closely. At one place I worked, its use was known as 'the laying on of hands'. A finger has the following uses and advantages.

1. It is always available


2. It can be used to inject ambient signals into a node, increasing the output of hum of RF breakthrough. Generally useful with high impedance low frequency circuits. Use capacitive (dry skin) or resistive (wet skin) coupling.


3. It can be used to ground a node.


4. At low frequencies it can add capacitance to a node, changing time constants, filter tuning etc.


5. At microwave frequencies, it can absorb and redistribute RF radiation, damp resonances, reduce signal levels, absorb the higher harmonics on an open signal line.


6. It's not generally recognised that high speed busses to memories and FPGAs are now running so fast that a finger on the lines will change rise-times, introduce crosstalk between clock and data, or reduce clock ringing, any of which can affect data integrity. I have dabbed a data bus while looking at the resultant TV picture, and could tell whether I had address setup or data crosstalk problems.


7. It can take the temperature of components, from 'that's OK' through 'it's running a bit hot' to '$$$ it's that one!'


8. It can cool selected hot components, improving the heatsinking, to see if temperature changes matter.


9. The speed. While watching an oscilloscope trace, listening to a demodulated signal, or watching a BER number, you can stroke your finger across a whole board in seconds. Any place it changes, that's a good place to think about whether you expected it to change there.

As a seasoned engineer, there have been many times in my career when several of us have sat round a mis-performing board, and the joke round the table was that if we could 'buy an engineer's finger, and glue it just there', we'd be good to ship.

And the question of accuracy. You don't need 3 digit accuracy, you need to know whether none, or a few pFs, makes a difference here. Once you know where to play with, you switch back to soldering chip components or trimmers.

And yes, all this usefulness comes at a price, you must not use it on circuits above 40v.