shadow goblet
Hello, I was designing a custom hotplate and was wondering if someone could review my design please.
Project files: https://github.com/notaroomba/ember
I am still working on making the hotplate itself separately, this is just the control board. SIG/GND/PWR/SIG
orchid herald
so... electrolytic capacitors really don't like heat
like, at all
to the point that if you have an electrolytic cap that's rated to survive, say, 4k hours @ 85C, you double that for every 10C below 85C of your actual working temperature
point being: if you can, at all, eliminate electrolytic capacitors from the design
also: I'd update to a synchronous DCDC buck
shadow goblet
ah okok so something like TPS56339?
that removes all of the electrolytic caps frmo the design and it's a synchronous buck
orchid herald
Honestly, I prefer just buying ready made modules, like the LMZM23601, but I'll take a look at that TPS in a minute
It's not really relevant in a hotplate, but synchronous bucks have better efficiency
shadow goblet
Ah okok, and wow I didn't know that premade modules like that existed lol
orchid herald
yeah, switching converters are annoyingly sensitive to layout if you want to pass EMI
at a glance that TPS you mentioned should work too
if you're not familiar, do take a look at TI's Power Designer https://webench.ti.com/power-designer/
it lets you input the design parameters and suggests designs you can then sort by estimated cost or part count and such
very useful tool
shadow goblet
ahh thats a cool tool, thanks! and yea ima see if I can use that TPS and then I'll send the updated schematic/PCB
shadow goblet
@orchid herald does the rest look good?
orchid herald
didn't really check in, tbh
shadow goblet
ah welp
knotty bobcat
alupoly is often made with better temperature tolerance
if you can insulate your board it might help
like c14 probably cant be replaced with a ceramic cap
depending on voltage you might not want to replace c3 either
orchid herald
I'm on phone, what's the capacitance?
knotty bobcat
120u and 22u respectively
derating @3.3v is not so bad for the 120u but it would be expensive
orchid herald
Yeah, 120u is a poly, 22u is just two-three 1210 MLCCs unless it's high voltage
orchid herald
Yeah, that is tricky
knotty bobcat
probably from usb PD would be my guess so potentially 48v
orchid herald
You can get decent 10u/50V in 1210 without breaking the bank, but that'd be four of them
knotty bobcat
You can get decent 10u/50V in 1210 without breaking the bank, but that'd be four...
derating will still bring it down a lot
even in a larger size
orchid herald
If you assume 50V
*48V
If it's 24V, four of them would be enough, and probably still less area, and similar price
But definitely use polymer caps
For the electrolytic parts
shadow goblet
probably from usb PD would be my guess so potentially 48v
24 V although I was wondering if I would need a controller that supports 48V like TPS26750
Although I'm new to this and had a question for the actual heatplate design. Using KiCad's calculator I can calculate how thin copper traces would need to be to reach a certain temperature but that comes with a voltage drop and power loss, Does that power loss apply to the board as well? Like if I have a hotplate connected in parallel to the 3V3 buck, and the hotplate has a power loss of 95W / the ~100W (20V @5A) that USB-C PD provides, does that mean that only 5W will be supplied to the buck? eg 1V @ 5A?
knotty bobcat
Although I'm new to this and had a question for the actual heatplate design. Usi...
use a resistive material for the hotplate that can withstand the temperatures. if you try to PCB the part that gets hot you will have a very bad time
shadow goblet
Yea I'm planning on getting a FeCrAl based heater from JLCPCB: https://flexheater.jlcpcb.com/capabilities
knotty bobcat
kk
knotty bobcat
Although I'm new to this and had a question for the actual heatplate design. Usi...
the buck needs to be capable of supplying all the current that the hotplate will use
heaters are just resistors, so your loss is going to be I^2*R
shadow goblet
the buck needs to be capable of supplying all the current that the hotplate will...
The buck is just for the microcontroller and logic level stuff, it can handle up to 3V3 @ 3A, the actual heatplate gets 20 V @ 5A
knotty bobcat
The buck is just for the microcontroller and logic level stuff, it can handle up...
then you'll need to make sure your heatplate is designed accordingly. im confused what this has to do with the buck
copper losses should be small
shadow goblet
then you'll need to make sure your heatplate is designed accordingly. im confuse...
It's just that the calculator said that there would be a voltage drop and I was wondering if that applies to the components that come before and rely on the 20V like the buck
knotty bobcat
a 250 degree rise in your copper traces means something has gone horribly horribly wrong
knotty bobcat
It's just that the calculator said that there would be a voltage drop and I was ...
id imagine you want to set this up in parallel
voltage drops apply in series
shadow goblet
ahh okok then
shadow goblet
is this mounting hole fine to place next to those I2C traces?
shadow goblet
would this be better for screws?
knotty bobcat
still closer than id find comfortable personally but ymmv
shadow goblet
ookay i'll move it a bit more
shadow goblet
does this look good? heatbed is a pwm signal
knotty bobcat
no decoupling on the driver?
knotty bobcat
i didnt notice the 1k resistor in line with VDD
why?
that driver is going to throw around some pretty gnarly currents
also adding capacitance to the gate is going to make life very difficult
you're already above 4n Ciss on that FET
adding more is going to stress out that driver
shadow goblet
I was copying a schematic that originally had an optocoupler and had the 1K resistor in line with VDD but then I switched to use a gate driver instead as I was following some advice from a friend. The capacitance on the gate was to follow the datasheet of having an output capacitor.
knotty bobcat
CL is not an output capacitor
that is the gate capacitance Ciss
the longer your FET stays in saturation the more power it will burn
shadow goblet
ahh so I should remove the 1nF capacitor?
knotty bobcat
yes
shadow goblet
ah okok, and is the 1K resistor inline with VDD is something specific to the optocoupler and should I remove that as well?
knotty bobcat
ye delete the resistor in line with VDD
if your driver pulls any current it will cause a huge voltage drop and you risk backpowering the driver or worse
latch up would end very badly 😂
knotty bobcat
you need some way for current to return
maybe a cap between pphv and gnd
unless you have a return underneath?
shadow goblet
ah wdym return underneath? do stiching vias count?
knotty bobcat
the current will need to flow in a loop right
so the current you pull from pphv needs to return along the path you pulled it from
if you have a ground plane or something underneath you'll be OK
but it looks like you split your ground plane
shadow goblet
the current will need to flow in a loop right
Yea but doesn't it go from PPHV to the hotplate and then back to ground? or should I add in the ground plane under PPVH?
knotty bobcat
the latter
the driver is not going to contribute much current to the hotplate
it more-or-less goes to the gate and back again
shadow goblet
what would isolation help with?
knotty bobcat
it would stop the control circuitry from getting damaged (+ anything downstream) if something in the power section fails
knotty bobcat
No
Optocoupler cant throw around the current you need
Otherwise youd just drive it straight from the micro in this design
shadow goblet
ah current to drive the fet?
knotty bobcat
ye
shadow goblet
ah I getcha
and thanks so much for your help!
also is this the right way to add in a current sensing resistor?
knotty bobcat
no your bus voltage should not be after the shunt
shadow goblet
?
