Isolating polarities on a series wired battery bank

If you short out one battery you get sparks because of the high discharge current actually melts metal which flies-off from where the short is made/enacted.

If you wire two batteries in series and short the combined series battery out you'll get sparks, maybe even bigger ones.

Now, think about what you did - you put two batteries in series and then created a short. I know you think you put two batteries in parallel but draw out the circuit and you'll recognize that wiring one one wrong way round (in parallel) is like shorting two in series out.

If you wired two batteries in series but they were "opposing" (+ to + and - to -) you'd just get a small current flow as they equalized in voltage - this is called putting them in parallel BUT you can also see that they are in series with a short added.

There is no one single strategy, and I don't think you can expect to get the design absolutely right without some trial and error.

Some blocks may need buffering from others. It is common to minimise external influence on the LO, and that often means a buffer between it and the mixer. An emitter follower (or source follower with a FET) running at fairly high current (10-20ma rather than 1-2ma) for good linearity may be called for.

Other times you may simply need to impedance-match between stages, using the "maximum power transfer theorem" (loosely stated, make source and load impedances equal) to minimise added noise by using as much of the source power as possible. This would apply between antenna and input filters, between these and RF amplifier, between mixer and IF amplifier etc, i.e. before the gain stages where signal levels are smallest.

For this reason a lot of RF circuits standardise on specific impedances (50 ohms is commonest, but also 75 ohms, 300 ohms or 200 ohms).

Sometimes you need to match circuits with different impedances. Then you can:

• tolerate an impedance mismatch. A low source impedance into a high load impedance is safe, but can (a) change the frequency response of a filter (either before or after the interface) or (b) worsen SNR by wasting signal power
• Add an impedance matching network (can be as simple as a termination resistor). This fixes (a) but not (b) above
• Match impedances with a transformer. Going from 75 to 300 ohms for example, use a 1:2 step-up transformer ( = 1:4 impedance ratio as the secondary has twice the voltage, half the current). This fixes both issues; if the next stage was the dominant noise source, feeding it twice the input voltage can improve SNR by 6dB!

You will see all these techniques employed.