Intercom mic/speaker noise when using step-down converter

[symbianas]

Hi, I am trying to make intercom door open using Morse code. All works well finally except that now I need to find permanent power supply. I have looked into intercom system specification and I am able to get wires from intercom system module to my phone!



"441" - is power supply for talk module which seems to be what I need! It seems to be DC 8V.

The issue I am having is that once I connect step-down converter 20v->3.3V (MINI560) I get noise in mic/speaker of intercom system. I have tried to connect other stuff like 12V LED to that power supply instead of step-down converter and no noise was present in mic/speaker.

Am I using wrong step-down converter for the job? How do I get rid of the noise and get 3.3V for MCU?

[benj38]

Step down DC-DC buck converters like the one you use are notorious for injecting noise into their supplies. Try to put a low ESR electrolytic capacitor of say 470uF at the input of the converter, and see if it helps. If you are lucky, it is all you need. Otherwise, we can guide you to try more complicated solutions.

[symbianas]

I have 25V 2200uF capacitor at hand and it seems to remove noise a lot (maybe completely?). But is it a complete solution?
Or should I look into better step-down converter? Any ideas what would be better step-down converter (or it would be the same just with capacitor included in its circuit)?

Would any of these be more suited for the application?
LT3042 LT3045 Single Power Linear Regulator Module
High-efficiency LDO Regulator Step Down Buck DC-DC Converter
AMS1117 LM1117 SMD DC-DC Step Down Buck Converter

[benj38]

I have 25V 2200uF capacitor at hand and it seems to remove noise a lot (maybe completely?). But is it a complete solution?

The switching noise is inherent to the principle of operation of DC-DC buck converters. The amount of noise and its spectra changes from one circuit to another, but if you filter it (like you just did with the capacitor), and the result is good enough for you, why not move on?

That said, a linear regulator would inject no noise at all. The main drawback of linear regulators is their inefficiency. In this application you would waste ~ 4.7V times the current requirements of the MCU as heat, which will have to be dissipated. Since I guess the current draw of the MCU is quite low, this may be of little concern, and a Linear regulator may be just the ticket.

[symbianas]

The switching noise is inherent to the principle of operation of DC-DC buck converters. The amount of noise and its spectra changes from one circuit to another, but if you filter it (like you just did with the capacitor), and the result is good enough for you, why not move on?

That said, a linear regulator would inject no noise at all. The main drawback of linear regulators is their inefficiency. In this application you would waste ~ 4.7V times the current requirements of the MCU as heat, which will have to be dissipated. Since I guess the current draw of the MCU is quite low, this may be of little concern, and a Linear regulator may be just the ticket.

After further testing it seems that noise (much smaller) still persists. So I guess my only option is linear regulator?..

[Konkedout]

I have no idea how much current you need.  But with that large voltage step down ratio (and voltage difference) you could easily run out of current from your source, or the linear regulator could get too hot.

Any switching converter can make noise.   I would try to figure a way to wrap your converter....maybe in something like an aluminum tube?  Make the tube much longer than the converter.  Or  Aluminum foil but the wrap around should be nicely sealed to make a closed tube.  Connect the metal to your DC- ground.

You might also benefit from common mode chokes at input and output, but my bet is that field from the pcb and inductors is radiating and causing this.

[benj38]

...but my bet is that field from the pcb and inductors is radiating and causing this.

Since the 2200uF electrolytic at the input of the converter eliminated almost all the noise, it is almost certainly injected electrically and not by radiation. In any case, it is extremely easy to find out by simply moving the converter physically away from its supply.

After further testing it seems that noise (much smaller) still persists. So I guess my only option is linear regulator?..

@symbianas, can you please measure the current draw of your MCU?

[PGPG]

you could easily run out of current from your source, or the linear regulator could get too hot.

Everything depends on consumed current. Accepting 10..28V at input I use linear regulator to supply 3V3 microcontroller + RFID reader + LEDs + RS485.

[symbianas]

I have no idea how much current you need.

@symbianas, can you please measure the current draw of your MCU?

Currently I am thinking to run STM32F103C8T6 at 8Mhz and it should consume <10mA. Also I am planing to introduce sleep mode which should consume <1mA on idle. During operation of door opening with Optocoupler it might consume <20mA.

Other option would be to run ESP8266 with WIFI and I guess in such case consumption could jump to <100mA, on sleep mode <20mA.

[benj38]

I would be super surprised if 20mA should be any problem at all. As for 100mA, there is also a good chance, but you will have to test it.

Obviously, if you can find some specs for the "441" supply, it is the best way to go. Alternatively, you can open it up and take some images so we can try to guestimate its power capabilities.

Otherwise (a bit more risky, but you already took some risk when you connected your converter to it), try to connect a 100 Ohm 1W resistor to the 8V supply (this will take 80mA), and see if the voltage remains around 8V or drops, and if anything gets too hot inside the "441". Make sure you are all set up to make a quick measurement, and to quickly disconnect the resistor if something gets hot.

If all is good, you can easily use a linear regulator here.

Just to make it clear, in case it is not, if you use a linear regulator than if your 3.3V load requires a certain current then a little over the same current flows out of the 8V supply; whereas with a DC-DC converter (of say about 80% efficiency), roughly half that current flows out of the 8V supply.

[xvr]

I have 25V 2200uF capacitor at hand and it seems to remove noise a lot

LowESR capacitor preferred here. Also you can add LC filter on input of DC/DC convertor (L-type or π-type)
https://industrial.panasonic.com/ww/ss/technical/b4

But linear regulator will be better, of course

[symbianas]

Obviously, if you can find some specs for the "441" supply, it is the best way to go. Alternatively, you can open it up and take some images so we can try to guestimate its power capabilities.

I have observed that at system start "441" voltage is 8.04V. Distance between intercom system and my flat is maybe 10-15m and the wire is similar to "ethernet cable".
When I measure voltage at my flat phone it is 7.96V (so already dropped a bit without any load).
If I attach MINI560 step-down converter voltage drops to 7.76V.
If I attach LM317 regulator voltage drops to 7.42V.
(I don't connect any load after step-down/regulator module)

I guess this voltage drop is not good at all? It might also be responsible for the noise?

Would it be any better to use "~" pin (12V AC power supply for lock, button, backlight panel) instead of "441"? I would still need to transfer it through same cable 10-15m and convert it to DC with rectifier?

[benj38]

I have observed that at system start "441" voltage is 8.04V. Distance between intercom system and my flat is maybe 10-15m and the wire is similar to "ethernet cable".
When I measure voltage at my flat phone it is 7.96V (so already dropped a bit without any load).

First, you can not have a voltage drop without a load. There must be something loading it a bit. If the only "load" is your DMM, then this voltage drop of 0.08V means that the wiring and connections have a 200KOhm resistance, which makes no sense. I would consider this difference a measurement error.

Would it be any better to use "~" pin (12V AC power supply for lock, button, backlight panel) instead of "441"? I would still need to transfer it through same cable 10-15m and convert it to DC with rectifier?

I would say that the connection designated "M" is much more interesting than the 12VAC. In your diagram from the original post it says "unstable voltage +14V", I guess this is a translation. Can it be that it means "unregulated" instead of "unstable"?

It is not clear from the image if there is any regulation of the "441" voltage, so the voltage drops you see as the load changes may be quite normal.

What would help a lot is if you:

• Take a picture of the other side of the board.
• Give us the part number of the part designated US-1, the diode D3, and the red diode next to the black wires connector strip.

[symbianas]

I would say that the connection designated "M" is much more interesting than the 12VAC. In your diagram from the original post it says "unstable voltage +14V", I guess this is a translation. Can it be that it means "unregulated" instead of "unstable"?

Documentation is in polish language (did not find anything in english) "niestabilizowane napięcie +14V" I guess it could mean unregulated.

I think I have found schematics of a bit newer version of this intercom system (but newer system does not have M - unregulated pin)





"441" pin is now "MP"

[benj38]

Let's reevaluate the whole thing for a minute.

The board you took a picture of and the MCU, etc. you are trying to power are 10-15 meters apart. Correct?
Why don't you use a separate power supply that plugs into the wall 230V to power the MCU?
I guess the answer is that the MCU is located in a place where you don't have a power outlet nearby, or for some aesthetic reason. Correct?

Where the board you pictured is, it looks like you have some space and that it should not be a problem to add an external supply (say a 5V USB charger). Correct?
i can clearly see in the image of the board that there are unused wires in the bundle that you can use to transfer this 5V. Then, you can use your DC-DC converter on the other side to power the MCU, and you are done.

[symbianas]

The board you took a picture of and the MCU, etc. you are trying to power are 10-15 meters apart. Correct?
Why don't you use a separate power supply that plugs into the wall 230V to power the MCU?
I guess the answer is that the MCU is located in a place where you don't have a power outlet nearby, or for some aesthetic reason. Correct?

Yes, I don't have power outlet near the phone and I don't want to bring new wires through the walls. The only wires that are available are those that goes outside flat to the public stairway intercom system. I am lucky that there is one free wire core which I can use.

Where the board you pictured is, it looks like you have some space and that it should not be a problem to add an external supply (say a 5V USB charger). Correct?

Yes, I guess its possible to attach new power supply to the mains 220V where the system is. And use one free wire for +5V and merge same GND wire as intercom system.

I just wanted to avoid adding my own power supply and reuse intercom power as I need just 20mA and <1mA on idle.
I am still tempted to try "~" 12VAC instead of "441" is it worth a shot?

[benj38]

I just wanted to avoid adding my own power supply and reuse intercom power as I need just 20mA and <1mA on idle.
I am still tempted to try "~" 12VAC instead of "441" is it worth a shot?

If you want only 20mA, and not 100mA, then you can use a linear regulator on the "441" line and that's it. I will be very surprised if it could not handle this load.

[symbianas]

I just wanted to avoid adding my own power supply and reuse intercom power as I need just 20mA and <1mA on idle.
I am still tempted to try "~" 12VAC instead of "441" is it worth a shot?

If you want only 20mA, and not 100mA, then you can use a linear regulator on the "441" line and that's it. I will be very surprised if it could not handle this load.

I have ordered AMS1117 LM1117 SMD DC-DC Step Down Buck Converter maybe it will make smaller voltage drop than LM317 one.

Also I was wondering is there a difference if I get "~" (12VAC) from the system or "~" (12VAC) after the LOCK (because I already have this wire in the phone)? (marked red and green)

[symbianas]

So finally I got my hands on ASM1117 3.3V and 5V linear regulators!

Input voltage 7.96V.

-> ASM1117 3.3V:
input voltage drops to 7.40V (5.5mA); power led removed 7.59V
+ STM32 operating:
input voltage drops to 6.80V (12.4mA).

-> ASM1117 5V:
input voltage drops to 7.31V (6.8mA);
+ STM32 operating:
input voltage drops to 6.70V (13.4mA).

So "ASM1117 3.3V" seems to be performing better than "ASM1117 5V + STM32 onboard regulator".

Removing power LED's on ASM1117 and STM32 should reduce power consumption more.

Though voltage drop is concerning? Sound in speaker changes with connected/disconnected voltage regulator, so not perfect...