12 VAC step down

[oliv3r]

Hey all,

I'm dealing with something I'm not quite sure how to resolve.

I have a battery operated chime (doorbell ding dong) that runs on 3 (or was it 4?) AA batteries. They probably use these as there was plenty of space (the speaker is more or less limiting) and to yield a reasonable runtime of a year or so.

I also have a 8 VAC transformer installed in the metering cabinet, specifically for the doorbel, though I know other variants exist, 12 VAC is also not uncommon. Changing the AC transformer is not something that is desired or planned

The PCB of the doorchime has a 3v3 LDO SOT-23 with the label HT36. Getting a datasheet for this part is probably impossible, bit it follows the same pinout as one of the microchip ones. Anyway, I don't know what its max input is, but I also don't think it's great to pump 15V DC into it, heats gotta go somewhere.

So with these two fixed systems, I have to work. What is now the best way to get the voltage dropped efficiently, without a costly step-down circuit.

Obviously a full-bridge rectifier; but then what. I think a capacitive dropper should be fine, we already have the AC at safe levels and isolated. Current also shouldn't be an issue, as the device can run on batteries after all (I'll do some power measurements one of these days to know for sure of course).

Any other efficient idea's? Just chaining a few diodes? Or just push in a beefy zener after the bridge?

Here's what I drew up; note that R1 is missing in the actual device, so anything before BATT is my addition. VBus is a USB connector.

[langwadt]

7805 ..

[pcprogrammer]

Get a cheap buck converter from aliexpress. Depends on the current the chime draws to what you need. You do need to rectify the AC to DC and add a capacitor to smooth it a bit. For example: https://nl.aliexpress.com/item/1005005058186802.html

They have all kinds on there and they work just fine. Adjustable or fixed voltage.

[DavidAlfa]

Skip the tranformer entirely, just get an isolated converter for less than $2, best efficiency:

www.aliexpress.com/item/1005004826780459.html
www.aliexpress.com/item/33012749903.html

[madires]

7805 ..

Should be the best solution for a long lifetime.

[oliv3r]

So I did some measurements and already blew up something

I soldered on some wires to the battery points to my DP832, and was working great. I saw power draw of almost 1Watt, so 250 mA. While I was trying to figure out how to do a histogram of the power draw, to see what the max was, I accidentally pushed 14V's. So I removed the HT36 which wasn't working anymore as it should (voltage wouldn't go above 2.65 V anymore) and shorted R1, which now makes sense, PCB reuse without a regulator It still works somewhat, but not sure if I may have blown up the speaker coils a bit. Anyhow,

I don't really want to remove the transformer from my metering cabinet, as I have a very thin solid copper wire running through walls to where the bell will be placed. I'm not gonna run 230 VAC on these wires So I'll always end up with 8 - 12 VAC as input.

So I could replace the transformer in the metering cabinet with a dinrail 5V supply (https://denkovi.com/hdr-15-5-industrial-din-rail-power-supply-5V-2.4a-out for example) though voltage drop over those thin wires is probably not ideal. And again, I don't want to really change the metering cabinet

Also, I have no faith in those ali-express units for serious use :p and they are not DIN mountable.

So 8705 as an input to the LDO is actually not a bad idea. Will have to be a tiny SOT one of course, as I don't have that much space.

So from those measurements, if I get a 350 - 500 mW unit it should be fine, even if someone is 'spam calling' generally, power draw is minimalistic, and the audio playback only lasts for 5? seconds. So 5 seconds of heat shouldn't be an issue i reccon.

What about efficiency? 99.9% of the time, the unit will just idle (lets say 10mA for arguments sake) doing 15 VDC -> 5VDC. Negletable?

I'm lookinag at AMS1117-5.0 https://jlcpcb.com/parts/1st/Power_Supply_Chip_79853 which should be great, no?

Am I correctly interpreting the datasheet however, that the AMS1117 will only really do 100 mA, because 12 - 5 = 7 V?

[pcprogrammer]

Depends on the space you have of course, but a single diode, 100uf capacitor, 100nf capacitor, 10uf capacitor and a LD1117V33 might do the trick. Depends on how high the DC voltage after the diode becomes. It is not that big of a power draw and there is a lot of overhead between the input and the output voltage. The LD1117V33 can handle 15V as an absolute maximum.

With the transformer in your fuse box having 8VAC you might be fine with this setup. As you already blew up the original 3.3V regulator no need to replace it with two. (8V, 3.3V or 5V, 3.3V)

LD1117V33 datasheet.

[oliv3r]

I did blow it, but this was on my test board :p I have 6 others

The 2.5v above requirement is tight. If we take 8VAC, as output, that's after rectification 11.3 right? Assuming that 8 VAC is RMS of course, and why wouldn't it be? take off 2x 0.7 and there's still a good 9V left.

As for using a 3v3 regulator, the design currently uses 'vbatt' to power some parts, mostly the LED.

I think the difference between the AMS1117 part and the 7805 (https://jlcpcb.com/partdetail/Stmicroelectronics-L78M05ABDTTR/C58069 I've found) is that the 7805 handles more current, at the cost of a bigger voltage drop. Though in this case, that'll be quite alright I recon. The AMS1117 is much smaller in size, but I have room for the 7805 in TO-252 it seems.

Doing some quick math though, This should now take as input VAC 6.5 - 25 (which would be about 8.5 - 37 DC).

So in short, it now looks like this; though I forgot the electrolitic cap after the bridge :p

[DavidAlfa]

Remember it's not only the voltage, but also the power.
You could put 37V to it, yeah... But 100mA would disssipate about 3W, a heatsink-less linear regulator will struggle, specially in a small board.
For best power dissipation, pour copper planes at the top and bottom as large as possible, connected to gnd, make the 78m05 ground pad solid and put lots of vias.

[Randy222]

"efficiency" and "3W" and "linear" don't really seem like words that go together.

Limited space, and efficiency are two things I heard mentioned.
A linear fits in the space, but not so efficient.

For efficiency, dc-dc converter seems logical, they make small dc-dc's like size of std 8pin dip.

If you want the linear route for less space, and need to sink out some heat, top solder only parts would be best, you can then mount that pcb a pcb-sized 1/8" aluminum, then screw down the 7805 back to the aluminum (drill tap). You can even add a thin piece of silicone sheet between board and aluminum to keep heat from going in that direction.

[oliv3r]

So there's deff. no space for the THT part. The SMD version works though in the space I have. There's a reasonable amount of copper front and back too.

As for power, that will only happen in short bursts. E.g. 99.9% of the time the system will be in idle, and my measurments showed something like 10 mA @ 3v3.

So then the question comes, how fast will the 7805 heat up, heat up the PCB, under 'full load' and stall. (it has temperature protection right?)

Proto will tell I suppose, but again, short bursts short of 1 Watt, and during those burts, playing the various notes caused different power spikes, so even then it wasn't a constant 1 Watt load. "It will be fine" is my gut feeling. Also the HT36 is not breaking a sweat either, and that is also dealing with 4.7 -> 3.3 conversion, which granted is quite a few volts less in delta of course.

[Randy222]

Watts are watts, joules/s

With all loads, it's gonna be temp that finally does it it. Quick fast ns pulses of 'on' time with high amps, the device will still survive, as long as no significant reactance is to worry about.

But, you can perhaps test using PWM driven fet on a fixed dummy load(max) and see how the reg temp reacts to various %pwm, maybe just decade it 10,20,30,40,50,60,70,80,90 for some period of time, like 5/10/15/20/25/30sec
Durations (time) also allows you to perhaps profile the pesty finger, pressing the doorbell over and over and overe again.
That will at least get you some temp rise data.

Then you can perhaps esitmate where in that data the chimes load kinda is, which then narrows down where the regulator operates.

[oliv3r]

Watts are watts, joules/s

Durations (time) also allows you to perhaps profile the pesty finger, pressing the doorbell over and over and overe again.
That will at least get you some temp rise data.

Then you can perhaps esitmate where in that data the chimes load kinda is, which then narrows down where the regulator operates.

Since this is not a professional project, I suppose just doing some imperical tests (spam the output over 10 seconds, 30 seconds, 5 minutes and watch the temperature.

But yeah, thanks! Watts are Watts.

[NiHaoMike]

You'll almost certainly find that the idle loss of the transformer dominates the overall energy use. Therefore, if you wanted to save energy, replace the transformer with a switching supply, a common 12V wall wart works nicely and gives lots of overhead for voltage drop.

[oliv3r]

You'll almost certainly find that the idle loss of the transformer dominates the overall energy use. Therefore, if you wanted to save energy, replace the transformer with a switching supply, a common 12V wall wart works nicely and gives lots of overhead for voltage drop.

What transformer? The transformer that is used in the metering cabinet? How would that be any different, just that there's now a wire between the transformer part, and the full-bridge-rectifier, the result is more or less identical. Also, as I wrote before, I can't replace the metering cabinet transformer But for those curious, the transformer is the following: https://datasheet.eaton.com/datasheet.php?model=1740298&locale=en with a nice pic also attached.