[TriEmbed] Coffee roaster sanity check

Brian triembed at undecidedgames.net
Sun Dec 29 14:27:45 CST 2019


On 12/29/19 12:03 PM, Pete Soper via TriEmbed wrote:
> Hi Craig,
> 
>     Very few people on this list are going to have KiCad at their 
> fingertips

Being one of the few[1], I'll give this a once-over here in a minute. 
Without looking at the documents, all of Pete's comments sound, well, 
sound.

  + Two amps is way too low for any sort of heating element.  Size your 
switching element for at least 15 amps for a 120-Volt heater; ideally 20 
or more.  A good rule of thumb when selecting current-switching devices 
is to size them for 2x the expected load current.  If your heater is 
1500 Watts (12.5 Amps @ 120 Volts), pick at least a 25-Amp switching 
element.

  + A smaller FET gate resistor is a good idea, but 100 Ohms might be 
too small.  The period of time spent while the FET is in its linear 
region (the time between fully OFF and fully ON) represents time where 
the FET is potentially dissipating more power (i.e. getting hot); 
keeping that period short is a good idea.  The FET gate is a lot like a 
capacitor: current flows in/out of the gate pin while the FET is 
changing state, but stops[2] once it's fully ON or OFF.  Your resistor 
is there to limit the spike of current when the FET starts to 
transition.  A reasonable rule of thumb here is to imagine it represents 
a short circuit; pick a resistor that limits the current to whatever 
your GPIO is capable of sourcing/sinking.  E.g. if the logic level is 5 
Volts and the GPIO pin can source/sink 20 mA, use 5 Volts / 0.02 Amps = 
250 Ohms.

+ Snubber diodes (diodes across windings of motors, relays, etc) 
definitely need to be oriented so that they are reverse-biased (not 
conducting) when the motor is on.  Two handy mnemonics for keeping your 
diodes turned the right way:
   - The "arrow" in the symbol points in the direction of traditional 
current flow (positive -> negative) when the diode is conducting.
   - The "minus sign" of the symbol should be connected to the "minus" 
side of the power supply in order to conduct (in this case, you want it 
"backwards" on purpose).

I'll reply again after I've taken a look at your KiCad files.

Cheers,
-Brian

[1] - I'm surprised so allegedly few members of an embedded-systems 
mailing list would be unfamiliar with KiCad.  It's a great tool for 
folks needing to escape the limits of other "free" options out there.

[2] - There's always a tiny leakage current, but unless you're driving 
large numbers of FETs at the same time, it can probably be ignored.




More information about the TriEmbed mailing list