Input circuits, shunts

I started looking more closely at the input circuit to this meter, and I realised something rather scary.

The Plan is to use reed relays to connect the stages from the input potential divider to the adc or rms converter, which looked like a Good Plan, when I did my first design, so I stuck with it through all the little tweaks and re-drawings and working things out (for which there has been an awful lot, for something so simple).

Then I started looking at reed relays, and bought a batch of 2 position ones which were made in 1985, so were very cheap and adequate for all the low voltage switching except the current shunts.

At some point I realised that the switch for the lowest range, which connects the meter's input directly to the inner circuits, actually needs to be a high voltage switch, because it's the only thing between the full maximum input and the delicate innards.

Whilst I was at it, I also started looking at current shunts. I'd planned to have a 2.5A shunt and a 25A shunt. I quickly discovered that whilst reasonably precise resistors can be bought for the 2.5A shunt, for higher currents it all gets a bit strange, you get a A/mV rating. There's a range of cheap shunts on eBay for various currents, all at 75mV. Okay, so I can scale that with a change in μC code - but since the current ranges are having their own amp anyway, it would be better to use an amp with appropriate gain (x40, for a 30A shunt) - save myself some coding!

I discussing my quandary on this forum, I learnt something about clamping diodes (thanks, ronv). I also decided that instead of just having a line of switches down the potential divider from input to the last tap, it should go from the first to the last tap, so there is always a 9M resistor in series with the input. I'll then need an amp to scale the input back up.

This raises the interesting possibility of using a gain other than x10 for the amp, and having the input ranges other than multiples of 2.5, and changing the scaling in the μC - I could have a gain of 12.5 and get ranges in more "standard" multiples of 2, or x8.3333333 to get ranges in multiples of 3. Seems a lot of work to change the scaling though - it was a sod to work out what numbers to use to do integer arithmetic. Thank goodness for spreadsheets!

Other schemes keep rampaging through my wild imagination...