• 3 Posts
  • 34 Comments
Joined 1 year ago
cake
Cake day: June 11th, 2023

help-circle




  • The standard way of looking at this is to consider a capacitor-resistor series combination going to ground. Connect a 10v (wrt ground) supply to the capacitor and the voltage across the resistor rises to +10v, then decays. Now connect that capacitor to ground and that same resistor gets -10v across it, which then decays. Whatever is connected to the capacitor “top” terminal has to be able to sink current as well as source it.

    That’s what generators in simulators do - they have zero internal impedance (usually). They sink currents as well as source them.




  • Assuming that the ribbon cable is standard - you could consider adding TWO IDC connectors, side by side. Then slice the cable through between them. Then add a standard extension cable to link the two. Indeed if one of the two is male, the other female - the extension can be removed if the thing is relocated to where the extension isn’t needed - or a longer one is needed.

    I confess to having done this sort of mod several times, myself. It’s also quite an easy way of sticking a protocol analyser/sniffer between the two and/or modifying the data that is sent on its way. Or adding an additional sensor (even of a different type) and converting its output to something suitable.




  • How I would do it is to use two digital IO pins on the processor to generate the reference square wave. Put the sensor plus a series precision resistor between them and just pull one IO pin high as the other is pulled low and then swap them. That presumes that IO pins can both source and sink IO current. Then take the junction between the two to an Analogue in pin. You get two measurements each cycle. Use a lookup table of values and interpolate between them. If you wanted more precision - add more series resistors of different values covering the range of humidity that you want to sense, going to different IO pins. So you can choose the IO pin pair that brings the centre point between sensor and resistor closest to the mid-voltage point. It’s effectively a balanced half bridge arrangement - using the precision of the resistors to determine overall measurement precision. OK it ties up several IO pins - but microcontrollers are so cheap, I’d probably just dedicate one to this sensor and that’s all that it would do.




  • Depends on your programming abilities. You can set up a “1-Wire” bus and put the lot on that. You could have one microcontroller as master and the rest as slaves - your external USB connection would go to the master and it could, in theory at least, then pass on program upgrades to the others. Clearly they would all have to be programmed individually, originally. But, quite frankly, the investment in time and effort on setting all that up? If you had dozens, hundreds - then yes. But you only have a small hand full. I’d just bring out their USB connections and program them individually, via those. I think each one has a unique identifier, stored in ROM - so they could all run the same software and include that identifier in messages on the bus.






  • It’s usually a good idea to look at what’s normally connected, when breaking a circuit, and replicate that. Which I seem to remember is a 10k resistor with a parallel capacitor (being too lazy to go back and look again). You could try the same combination, in place of the added input cable, on the lines that you plan to use and see if it whines. If not, add the cable(s) and try again. That may stop it happening on all channels.

    There are two muting methods - open circuit the input (via a switch or a gate) - in which case there will probably still be some signal transfer through capacitive coupling - especially if the amp side of the open circuit is high gain. Or short the signal path to ground - and that short will have some impedance and thus the signal is only attenuated and not removed all together.

    Turning the amp gain to far higher than it ever would be in practice is hardly a fair test. There aren’t many amps that will be noise free under those conditions - and, in this case, there will be a signal to amplify, albeit highly attenuated.

    It’s good to be cautious - cascade failure can be waiting to bite. Never direct connect unless unavoidable - add a series capacitor, if you can. Yes, not usually a good idea to provide a dc path unless essential and, even then, current limit it if possible.