3.7 VDC to 50 VAC inverter for ringing old 1000 Ω phone bells

I have this old black rotary dial phone which I want to connect it to a GSM module and microcontroller.
The bells will ring at 24 VAC but they sound better with 48 VAC (no sound at 6 VAC). Inside the phone there is a schematic telling the bells has 1000 Ω resistance. I calculate that to 50VAC/1000Ω = 0.05 A, and 50VAC²/1000Ω = 2.5W.
Now I want to make 50VAC from a 3.7VDC battery based on this inverter schematic with two MOSFETs.

I have found this 1x240VAC to 2x18VAC 50mA transformer and when connected to a square wave of 3.7VDC at 50Hz it should give 240/18*3.7 ≈ 50VAC.
On the battery side it should give a load around 2.5W/3.7VDC = 0.68A.

Should I just choose MOSFETs which can handle 1A?

Is that how it works with transformers and inverters or have I misunderstand/miscalculated something?

The bells are suppose to ring for ½ a second (I think) and then with 1 second pause. I don't expect the transformer to became too hot.

It is similar but not identical to a system I used to use for testing the volume of telephone ringers. It will work to provide ringer voltage but probably produce much higher voltage than you need. If I remember right, the system I used had a pair of bipolar transistors instead of MOSFETs but that make little difference, it used an 18-0-18 transformer though so the output voltage was much lower. The danger of producing 230V is most phones, even old ones, had a voltage clamping circuit in them which will effectively short out any excess voltage and overload your circuit.

Brian.

With just the center tapped xfmr and a pair of NPNs,
pair of resistors and pair of capacitors you could make
a self-oscillating push-pull inverter. NPNs will work down
to near a volt, no problem there. But I think you want
a much lower frequency than mains transformers are
meant to operate at. That may result in more heat
than you expect.

If you need 2.5W (plus inefficiency) from a 3.7V battery,
that's looking like about an amp. Might be a higher-end
lithium battery needed.

According to ETSI EN 300001 (General technical requirements for equipment connected to an analogue subscriber interface), ringing frequency is specified differently in different countries. 25 Hz is more common than 50 Hz, but both should work. I guess that a lower frequency might sound better with old telephones like W48.

betwixt: "higher voltage than you need"
Well that is better than my Falstad simulation (some values must be wrong). As I can do with 24VAC I think I will try with a 24 VAC 240 VAC transformer (ratio 1:10 should give 37VAC minus inefficiency) to start with.
I don't have any power transistors so I will just buy some MOSFETs. And a fuse to protect my circuit.


dick_freebird & FvM: I will have a microcontroller to do other things anyway so I will let that do the ringing.
It is a phone from a danish manufacturer year 1954 (not a W48, but it looks like it), so I'm quite sure it is made for 50Hz. To my knowledge it is designed for 60 VAC but less can do.
If the LiPo can not drive the bells I think of adding a supercapacitor as the power consumption come in burst.

Telephone ring frequency (in North America) is 20Hz, not mains frequency. The ringer sometimes has a resonant-capacitor. The armature cannot move fast enough at 50/60Hz to get a good loud ring. 50/60Hz transformers will do poorly at 20Hz. It maybe easier to step-up DC and then pulse the ringer with H-bridge.

^^^ This, what Prairiedog said ^^^

Unless you derate the power transformer by 2 1/2 to 3 times.