Power controller for a 12v batter/ psu operated amp

[dumle29]

Hello, my friend is making a musik box on wheels, and he will have a 9-16v amplifier in there, for running his stuff. He would also need 5v for mobile devices. He asked me how he would go about adding a power supply, and since he does a lot of computer stuff, I told him to use a PC PSU. So now that that was established, I decided it needed to have a power controller, so that he could plug it in, without the music stopping (or very briefly), so I tried to design that.

So I have some questions about the design, and that's mainly why I post here, as I don't think it would be that useful in other situations

I'm still sorta new on the PCB front, and I have never had anyone teach me, or criticize my work, so I would appreciate your thoughts
One thing I'm a bit iffy on, is the placement of C1, the 470µF, as it is a good bit away from the output. Does it make that big of a difference, and would it pose a problem for a usb powersupply?

Thanks for the criticism

Edit: new design, uploaded the new pictures
EDIT2: Forgot the diode on the relay, removed copper fill from the image (it will be on the board, but it looks ugly in this dpi). C2 is connected to ground through copper fill.

Edit3:
I did a ; ripup (rip all traces) and tried again. That always helps So this time, I tried to focus on the power traces as suggested, and I found a radial 220µF low ESR cap.

EDIT4:
Removed diodes and made the relay toggle between the two sources of power. VMP3203 is 10-20µA at standby.
 

[ve7xen]

- The relay seems excessive to me, you can do this with a PMOS instead I think
- Your +5V isn't switched or diode-ORed. I would not include the input +5V unless you provide some protection for your DC-DC converter
- I'm not familiar with the VMP3203, but keep in mind the +5V zener will have a fairly wide nominal spread and not be all that stable. Probably fine for charging stuff, but if you get one that's 4.9V or something then consider IR drop in the leads you're getting down there. I'd use a TL431 or something and set for 5.25V or so. Assuming this is a reference voltage input.
- As far as the cap, is there a reason you're using axial? A 470uF 6.3V can't be much (any?) taller than your relay. You could put a radial cap right beside the output terminal, which should also shrink your board.
- How much current do you need? Your traces aren't that wide and the ground path is kind of long. I guess you're limited to 1A by the 1N4004, so maybe not a big deal. I'd still move D2 to the other side of the relay so the main power current path is shortened considerably.
- I may be misreading your schematic, but relays are typically shown with their 'normally' contacts closed. When a power source is connected to +12V this would switch contacts and open the connection to your DC-DC converter, and likewise when only a battery is available it would be connected to +12V which has no power.

Keep it up!

[dumle29]

Hi, thanks a lot for your thoughts
-The relay is indeed excessive, as I will pull around 2 amps top (I will change the 1n4004, I just ploped it in there for now).
-Would you say a 1n4004 would do as the diode on the 5v rail from the vmp3202, or should I look for a schottky diode?
-The accuracy of the zener doesn't really mean anything, as the pin on the VMP3202 is just an enable pin that can't go higher than 5v. It should be pulled high internally, however some have had problems with it so I'm pulling it high with that
I'm using an axial capacitor, as that was the one I found to have a low ESR and high enough capacity. It has the dimensions of 30mm long by 10mm diameter.
-For the trace width, I'm using 56 mil, according to circuitcalculator.com, I would only need 26,9 mil in air, if I allow a 10°C increase over 60°C ambient.
-And no, you were not misreading the schematic, I had recently changed the +12v and vbat pins around, and forgot the relay, thanks

[ve7xen]

-Would you say a 1n4004 would do as the diode on the 5v rail from the vmp3202, or should I look for a schottky diode?

You'll need to calculate how much power it will handle (also waste) to know for sure. It's probably fine, depending on the efficiency of the VMP3202 and the amount of current your device is pulling. If you want to charge an iPad at 2A that's 10W or say 1A at 12V counting for some inefficiency. 1A over the ~0.6V diode drop will be 0.6W. IIRC they're rated for 2-3W so you're probably fine. A Schottky will waste less power though, might be important to save battery.

-The accuracy of the zener doesn't really mean anything, as the pin on the VMP3202 is just an enable pin that can't go higher than 5v. It should be pulled high internally, however some have had problems with it so I'm pulling it high with that

Gotcha. Most of these regulators I've seen can take the full Vin(max) on their enable pin as well. You might not need it.

I'm using an axial capacitor, as that was the one I found to have a low ESR and high enough capacity. It has the dimensions of 30mm long by 10mm diameter.

Weird, I'd expect availability of axial caps these days to be crap. Anyway, ultra low ESR probably isn't critical in this spot, any cap will work fine (check the DC-DC datasheet of course, but most of these modules will be stable with any output capacitance). I'd just go for a good quality standard cap. Axial will make your layout better .

-For the trace width, I'm using 56 mil, according to circuitcalculator.com, I would only need 26,9 mil in air, if I allow a 10°C increase over 60°C ambient.

As long as you're checking you're fine. I'd still maximize it and minimize your power trace length at the expense of your other traces.

[dumle29]

-The reason I'm asking if a 1n4004 would be sufficient for the 5v rail, is that I don't know if devices will charge at 4,3 volts, or if they absolutely need 5v or close to 5 volts. In any case, it should be easy to drop a schottky in instead if it proves a problem.

-About the VMP3202. It's a cheap (well 5$ a pop, but they are easy), it is a switching DC-DC adaptor with coil and the schtuff, and comes in different fixed voltage variations (tho it seems the voltage is set with a resistive devider if you take of the casing). A rather stupid thing about it, is that the enable pin is logic level, and no higher. Going higher than 5v on the enablepin, will burn it. This shouldn't really be an issue, as it is weakly pulled high internally, tho people have said that it is unstable if relying on the internal pull-up.

-The ESR does need to be as low as possible, as this is a high frequency switching supply, and a normal cap will heat up and waste too much power / explode / dry out I did however decide that 220µF would be more than enough, and at that value, I found a reasonably priced Panasonic radial cap, along with some other brands. (just noticed the panasonic one )

-I ripped all the traces, and went again. This time trying to make the runs as short as possible, and it made the board more compact, saving me 8$

[ve7xen]

-The reason I'm asking if a 1n4004 would be sufficient for the 5v rail, is that I don't know if devices will charge at 4,3 volts, or if they absolutely need 5v or close to 5 volts. In any case, it should be easy to drop a schottky in instead if it proves a problem.

Sorry I didn't realize you meant on the output. You want at a very minimum 4.75V at the device-side connector under full load current, as that is the USB minimum spec. Most devices will charge a bit faster and be a bit happier at a little over 5V, so I recommend targeting 5.2-5.3V at the USB port, as most commercial chargers do.

-The ESR does need to be as low as possible, as this is a high frequency switching supply, and a normal cap will heat up and waste too much power / explode / dry out I did however decide that 220µF would be more than enough, and at that value, I found a reasonably priced Panasonic radial cap, along with some other brands. (just noticed the panasonic one )

I would expect a self-contained module to include enough low-ESR capacitance that significant ripple currents don't really flow in your external circuit, but like I said, I don't know the module you're using.

-I ripped all the traces, and went again. This time trying to make the runs as short as possible, and it made the board more compact, saving me 8$

Nice work

[dumle29]

Okay, the usb should now get 5v, and no need to fuss about high amp, low forward voltage transistors anymore I Simply used the Dual throw of the relay

About the ESR, the datasheet recommends 47µF low/ultra-low ESR cap, so the ripple current must be rather high. Other users of it has said not to go lower than 220µF cap.

So the change now, is that I removed the diodes. They were the bottlenecks when it came to current, and they had an annoying forward voltage, that caused problems.

This too meant a slight decrease in size, and the price for 3 boards at oshpark is now just short of 21$ for 3 boards

[brainwash]

The LM2596 modules go for peanuts on ebay and are able to sustain 1.5A@5V. With some heatsinking they will do 2-3A, otherwise they will get quite hot. I assume this is for charging phones and tablets, so the lone module would do just fine. Cost: 1-2$.