Hello! I made a credit card sized addressable/analog led controller (5/12 or 24V) with a built-in buck converter that you can power through the Type-C PD protocol, up to 100W! It's really compact compared to what it can do and how much wiring it saves. Obviously it's completely open source. If you're interested, this is the repo of the project with all the sources: https://github.com/ale1800/YULC
Any feedback is welcome!
#100W Type-C PD Led Controller
68 messages Β· Page 1 of 1 (latest)
This is pretty cool. I am interested in popping out a few of these to manage the lighting in my house. Only thing that I would consider is an add-on so that you can pull and source power directly from outlet.
Very cool indeed! I'll definitely have to look though your repo to understand this better. That's quite the army of capacitors and resistors you have on there! π
Tangential, but I saw the instructions say "Caution: When selecting "USB", do not also connect the barrel jack!."
Would it be possible for the board to handle that itself? I feel like someone will inevitably do that and make a potentially expensive mistake.
I saw that array of capacitors too! I figured this was a build that could definitely live up to the 100w claim
Thanks! Could you explain better what you mean?
Ahhh, I was suggesting something so that one could just wire it directly to 120V AC and get the necessary power/control
Yes, it could but it would need other n-mosfet for 20A and other drivers. Because if you choose P-mosfet for switching the positive side with potential 20A passing through, it would be super hot. Using other two N-mosfet would require other two dedicated IC to drive them and the component price would increase a lot. However it would be technically possible
Aah I see that makes sense. Thanks for the info!
Oh ok, yes it would be useful. So like a tailor-made wall charger
To be honest in the first revision of the board (the main image in the repo) I tried to put a protection for that particular case in a (too) simple way (seems to me almost cheating) but I soon realized that I couldn't beat the physicsπ€. So I just gave up
I wonder if you could make space for one of these converters (or hopefully a much cheaper version π³) on the back side of your board https://www.digikey.com/en/products/detail/cui-inc/PSK-100-24/15964810
Yes there's a lot of capacitance before and after the buck converter. For that amount of current they are a must
Yeah, I see. It would be cool. You would save the cost of an external power supply
Also those dimensions are not so prohibitive for the output power
Yeah I've used something similar for some in-wall devices and they work nicely (although in that case it was 3.3-5V at 1A π€)
I used a Hi-Link converter some years ago. Those modules seem very reliable
That's really awesome! This would be absolutely perfect for me if it had screw terminals for power in π
I know that could have been a solution. I was undecided between terminals or the barrel jack. At the end I opted for the last one. I have a lot of old 12V router power supplies and similar, so I thought that this was the way to go, also to keep everything a little bit cleaner
I wonder if the holes/traces could be designed so either a barrel jack or terminal block could be fitted depending on your needs
Probably yes, I think it could be doable
Maybe it would be easier to add a terminal from the opposite side, where there are all the output terminals. To do that I should reduce the pitch to be able to fit everything
Yeah it seems more likely people would want everything hooked up on one end in that case. And makes it easier to tuck it away in a wall or similar
Hello! This board will hopefully be available soon. So if you would like to have one, I ask you some seconds to fill this form, in order to let you know when they will be ready. https://docs.google.com/forms/d/e/1FAIpQLSeoDaMMP-SH_50COGaSMnnIbudT9Xs9PtZJE55X_nJfeFEupQ/viewform
Thank you very much!
Whatβs the estimated cost?
Probably it will be something around 30/35$. It depends on a lot of factors
Iβm in! Form didnβt ask quantity, but Iβm interested in 5
Thanks! There should not be any problems for the quantity π
Hello everyone! If you are still interested, it's finally available for pre-orders here! https://aaelectronics-shop.com/products/yulc
Hi,
When do you expect to ship the boards?
@sand ruin Hello! I should receive everything this week, so next week I should be ready π€ Unfortunately there has been some delay due to the Golden Week holiday in China.
Ok, good. π
Hello @kind thistle,
Have you some news ?
Hello, yes. The package Is actually at the italian customs, so I should receive everything today or tomorrow. I expect to start shipping next week π It's quite frustrating to have all these delays, but I think we are close π€
Received my board today. π
so pretty
Hope you'll like itπ
I'm waiting for the fuse and the terminal block. I'll replace my current board this weekend. In any case, the card seems to be of very good quality. The only thing missing was a heatsink in the package, but it's not a big deal for my use.
Oh, I'm sorry! Which one is missing?
The one for the inductor. In my case, i will use direct routing power.
@sand ruin great! Happy to hear that, thanks for the update π
I see you put the ESP antenna outside of the board. Did you get any problems with that setup?
Some Chinese claim they can only solder components that are on the board.
Mechanical switches also exist π
Hi! No, luckily I didn't have any problem
Well, yes. But I wanted those switches to automatically switch without any "human" action. If those things are managed by the board itself, there are less chances to make errors blowing everything up
I expect that strategically placing a physical switch would prevent things from blowing up, and instead make the beard do nothing when power is connected incorrectly.
Given that now you have to think about blowing things up when connecting power, and with a switch you would have to think about the board getting powered at all when connecting power it would be a slight improvement.
I will consider that. My goal is to minimize the physical interaction, leaving only a few manual settings
If you can get a switch for the task.
While switches rated for 20A do exist they are typically switches rated for mains switching, designed for panel mounting, and cost about $5. That is a bit steep for a switch even if it replaces a couple of MOSFETs and a MOSFET driver. Some PCB mount variants exist but when they are not part miscategorization typically you have to wonder what the designer of that switch was thinking.
I see that you use some MOSFETs for the PD path which limits the current according to the PD specification but say you cannot do that for the power jack path because there the power is not limited in this way, and the pMOSFETs for the high side switching need to be much bigger than the nMOSFETs for the LED low side switching.
You are using a MOSFET driver IC, and when looking for MOSFETs I found some manufacturer promo that in this case they have you covered: they have MOSFET driver ICs that pump up the input voltage to switch the nMOSFET on the high side. I did not research how that actually works in practice. Obviously some additional passives are needed. Another question is cost of such driver, and yet another how well it deals with variable input voltage.
When looking at commercial solutions I found that typically the LED switching controller datasheet says it provides the maximum output power at the maximum voltage, and the output power is proportionally lower at lower voltages. It is probably because of this problem. When providing equivalent power output at lower voltage higher current is needed, and the parts that can handle such current are limited.
Thinking about realistic applications I do not think I would ever want more than 100W of led lighting controlled as one unit.
Limiting yourself to 10A you get about 100W at 12V and 200W at 24V, and get much wider selection parts to use.
Of course, for the commercial smart LED solutions it is fine to limit to 10A or even lower because the voltage regulation is outsourced to a separate power supply, and the smart controller typically does only the switching. When you also do the voltage regulation and want to limit the heat generated by the PMIC you might want lower input voltage, and corresponding higher current.
Another thing about mechanical switches is that the mains rated switches are rated for AC, and where the manufacturer bothers to state the DC rating it is much lower. This is probably not an accident. DC is more prone to arcing when switching, increasing switch contact wear.
Disclaimer: I am not an electrician nor electric engineer, and know next to nothing about this stuff.
For completeness, USB PD trigger board costs about $1
https://aliexpress.com/item/4001243047536.html
which makes outsourcing of voltage regulation look wise.
I believe it's already doing PD triggering when using usb power
I understand what you mean. I'm already using those drivers to switch the high-side N-mosfet for both of the channels. They are handy but also quite expensive. Also adding MOSFETs on the barrel jack path that can go up to 20A means more heat to be generated. And you would need two MOSFETs in back to back configuration, so another driver and other two warm/hot components that increase the overall board temperature (and cost) adding the need of two more heatsinks. So there are a lot of pros and cons that I had to consider. In the end I chose the other way. So you have to pay attention to that. If you take dig-uno or other led controllers they all say "match the voltage", "do not use both power supply and esp32 USB power and so on". Considering that this board can do a lot more things, I think that a minimum of attention is the price you can definitely pay to have the possibility to configure the board exactly how you want.
And that's the source of the problem. Because it does PD triggering onboard there are two power sources to switch between. With an external PD trigger power always comes from jack, no switching required.
Sure, but then you wouldn't have the possibility to program the esp while powering the led strips with a completely different voltage
Why wouldn't you? You could use dual redundant LDOs hooked to the jack input and the USB VBUS for the ESP power only, that's 3.3V 500mA. You would only lose the option to power the LEDs from the computer from which you are programming the ESP when you are programming it.
Or at least using one port, you could always use another cable if your computer has more than one port.
Does the provided ESPHome firmware support non-addressable RGBCT LED strips (eg like https://www.amazon.com/dp/B0C49NBDC2 )?
RGBCT Light
Documentation
Source
Or more likely
RGBWW Light
Documentation
Source
Ah, cannot love ESPHome enough π Thanks!
As you can see ESPHome can control those strips, but you'll need to change the ESPHome configuration
@kind thistle Very cool project! I found a potential problem, FYI. The MCP1416 mosfet driver has a maximum VDD of 18V. If you're requesting 20V through USB-PD, it seems like you're gonna damage that chip?
I see an absolute maximum rating of 20V on the datasheet but not sure what the implications are of running it at that, or if it's possible for PD to spike above that
@rotund cloak I missed that abs value, good catch. I've read that USB-PD sources will actually output a few hundred mV (at 20V) to account for voltage drop in the cable so it's actually 20V at the end of a 3-6ft cable (haven't tested that yet).
Hi! As @rotund cloak said, I'm taking advantage of that value to make it work. I extensively tested that configuration on my side (a lot of hours) and honestly I've always measured a bit less than 20V (19.5V/19.6V) and I never faced a problem. Then the ch224k (the PD IC) can have an input max of 22V that the PD 3.0 protocol can't reach. So from this you know that the voltage can't go higher and will be at a safe level.
I'm building my own USB-PD device at 20V and I've been looking what others have done for ideas. Your USB power-path is very well designed. Nice work!