I jost ordered an DPH8920
96v output with 20A
was torn between the two since my supplies are just a bunch of stacked 12v/63A server psus so I have the Amps.
But I’d rather be future proof for those 20s days #stormcore
Don’t have one (yet). Haven’t had time to figure out the models yet. US wall power is about 1500w, it’ll be hard to feed these babies. Nice problem to have
Yeah but I dont plan on running the thing at full tilt ever anyway.
its more about the range. Efficiency of my DC supply is gonna be maybe 95% about the same as these unit so over all 1.5KW*0.903 gives me 1300 watts to play with.
Thats assuming I don’t find just plug it into the 220v line in the garage. You may also be able to find a 30A line around. possibly by the dryer or boiler
I’m facing some issues related to the poor quality of the power supply output.
What’s weird is that my prototype worked flawlessly straight forward.
I’m also reconsidering a lot of things in my personal life.
So to be more aligned with what really motivates me, I decided to release everything under Open Source licence.
This will also people to contribute as I’m in a self-learning process on many topics (including dev and electronics ).
Back to technics : I was working on the casing when my 3D printer decided to piss me off.
So I ordered an Ender 3 Pro and started a (big) project of re-building my proprietary 3D printer with standard parts.
To sum up : skills limited + no 3D printing ability + less time for the project + personal life concerns = hard time to find motivation and keep this project moving forward.
I hope that the opening to other contributors may help the project to reach more people because I’m really convinced that this is a really useful device.
As I get weird noise on the comm between DPS and my MCU, and because I didn’t even push my prototype to the limits (charging at 20A) I was concerned to even more troubles.
So I decided to isolate galvanically the link between them. I moved to 2x Broadcom HCPL-0600 (tiny package for 6N137 optocoupler).
I still need to test it on breadboard but at least I managed to squeeze everyhting on the board to embed the extra components without changing the width of the board (thus fits my enclosure without changes).
If experts see something wrong on my device or do you any recommendation regarding the layout, things to know about optocoupling or anything, feel free.
The more I’m going on this project, the more I feel like swimming in an ocean of unknown.
Do you share ground between all PCBs? While the uart interface is pretty rigid, keep cabling to a minimum and away from any switching sources. I can’t really tell much from that picture, would have to look at gerbers, whats your PCB stackup? Do you have a dedicated groundplane? Shielding cables and adding ferrites/LP-filtering on power ports to supress high frequency noise will prob help as well.
Also if thats a tantal cap on your PCB, keep in mind that rating should be 3x what you use it for. Else you wont have much cap left.
I do not share an extra GND with TX and RX between my PCB and the DPS because they are powered with the same main power supply. So they’re already sharing GND by this way so I wanted to avoid a GND loop.
For more details about the noises, see the scope shot I posted above :
You can see a crosstalk from TX to RX (although not in the other way).
2 layers
Yes, Sir.
I may switch to another technology of cap.
TBH, I know barely nothing in that field.
What would you recommend ?
I havn’t plowed thru this thread yet, so might have answered stuff you already figured out, just said stuff off the top off my head that are usual problems. If you see noise all over, try to check your voltage rails and see if you have stability problems, finding the source of the noise could solve alot headache.
regarding the caps its always good practice to decouple all ICs/supply lines with 100nF near inputs, the esp module probably draws alot of current on the wifi transmission so a bigger cap there would be beneficial, on the 3.3V regulator I would deinatly use MLCCs, one 100nF and one 10uF. (or whatever the datasheet specifies). If its for 3.3V use 10V rating as cost isn’t major factor at these levels.
Crosstalk can be caused by a majority of things, I’d have to look at a schemtic to make any assumptions.
Edit: to meassure noise in a good way you need to solder a coaxial connection with a solid ground point, thru f.e a SMA connector and then to a BNC into your oscillioscope. Those probes are pretty good at picking up florescent light, 50hz from main power and so on.
Thanks for your inputs !
Regarding the caps, I based my design on reference designs so I had plenty of caps, like you suggested.
However, it ended to bring issues because the MCU wasn’t powered in an enough short time.
The ramp up caused by all caps made it enter in a “ghost” state (powered but stay asleep).
By removing caps one by one, I finally could have it booting up at each powering on.
The development is still ongoing.
However, it’s quite hard to find motivation because it seems endless…
I was really confident for the v0.3 board but unfortunately I missed some basics about voltage regulator requirements and learnt that the one I used (AMS117) was an old and picky one.
This was confirmed on another project I’m working on and based on the same design.
Let’s see the bright side : it makes me learn a LOT about voltage regulators, especially on 1117 designs.
Also, I twisted RX and TX on an optocoupler. I hardly hated myself on this.