Produce 60V AC (25Hz) from a 6V battery - will this work?

[Odysseas]

Hello folks,

I need to produce a voltage of around 60V AC (25Hz!) with a current capability of about 30mA. I came up with the following circuit - my question is this: Can this circuit be expected to work the way it's meant to? The voltage doesn't have to be precise, +/- 20% will be no problem.

TR1 is a small transformer with a secondary voltage of 2x9VAC, and a primary voltage of 2x115V and a rated power of 6VA at 50Hz (Since I want to run it at half the rated frequency, I'll need twice the rated power, correct?). Q2 and Q4 will be controlled by an Attiny, which will output a PWM signal based on a pre-computed sine wave. Q2 for the positive half of the AC voltage, and Q4 for the negative half. V+ is directly connected to a 6V lead acid battery. I will effectively be feeding the transformer half its rated voltage into the secondary side, so I should see about half the rated voltage on the primary side, right? Oh, and I was thinking of a PWM frequency of 1kHz (so the PWM base frequency inside the Attiny will be 256 kHz -> 8 bit PWM) - does that frequency make sense?

Originally, I thought of using an H-bridge with four MOSFETs, but this variant is much simpler as I don't have to fuzz about producing the drive voltage for the high side MOSFETs.

In theory, I think this should work but perhaps I have overlooked something. Any comments are appreciated!

 

[audioguru]

Each 9V winding will have square-wave that is almost 6V peak-to-peak.
Each 115V winding will have a square-wave that is about 77V peak-to-peak for a total output that is about 154V peak-to-peak.
If the transformer is tiny then the output voltage will drop when there is a load.

Are you making a 25Hz sine-wave with 1kHz PWM?
Are you trying to make an old telephone ring? Here they used a sine-wave of 90V at 20Hz.

 

[Odysseas]

Each 9V winding will have square-wave that is almost 6V peak-to-peak.
Each 115V winding will have a square-wave that is about 77V peak-to-peak for a total output that is about 154V peak-to-peak.
If the transformer is tiny then the output voltage will drop when there is a load.

So in theory, I should get about 52V AC out of the primary windings of the transformer as I have outlined in the schematic?

Are you making a 25Hz sine-wave with 1kHz PWM?

That's what I would like to accomplish. The question is, will it work and is 1kHz a reasonable frequency to work with?

Are you trying to make an old telephone ring? Here they used a sine-wave of 90V at 20Hz.

Exactly, that's what I'm trying to do. I have searched the web for all sorts of ideas but to me, this one seems the most obvious since I have 6V available with good current capability. It seems the phone bells here in Germany would require an AC voltage of about 60V, at 25Hz. However, the long wires in those days could remove a significant voltage drop.

 

[audioguru]

A cheap little 230VAC to 18V transformer produces nearly double the voltage (36V) with no load because it is cheap and its voltage drops to half with a load.
If it is used in reverse then it will not produce enough output voltage because it is made with the amount of turns for 36V on the low voltage winding, not 18V.

A good quality transformer has pretty good voltage regulation so it will produce the voltage you expect.

Telephone bells use very low current. I think you can have 5 telephones connected to a telephone line and they will all ring.

 

[Odysseas]

The transformer I have here has an idle voltage multiplier of 1.35. I will try using this transformer. If the output voltage should be too low, I can use a transformer with a secondary voltage of only 6V instead...

 

[canadaelk]

This almost seems like a telephone ringing circuit. If you google that you will find many circuits which, with minor mods, will fill your needs. E

 

[Odysseas]

Okay, I have built and powered up the circuit. I used the 2x9V transformer (it's a Block Trafo FL6/9) I specified in the schematic. The PWM frequency is 1 khZ, the AC frequency is meant to be 25Hz. But the voltage I get out of the primary side of the transformer (both windings in series) is way too high (around 155VAC) - here's an oscilloscope snapshot (I connected it to a :3 voltage divider so the voltages need to be multiplied by three) - look at the horrible shape of the waveform. The circuit was powered by a lab power supply set to 6V.

So what do I do with it? Do I have to increase the PWM frequency? Or lower it? Or filter it after the transformer? Any help is appreciated!

 

[kubeek]

Looks to me more like aliasing than a horrible waveform, especially those bits alone away from the wave. Try looking at it with a faster timebase to get a proper look. Or use an analog oscilloscope.

 

[alec_t]

LTspice simulation shows a similar messy waveform, so I don't think a faster timebase or an analogue scope will help

Edit: Further sim shows an improvement if the phase of one of the input windings is reversed.

 

[tcmtech]

Post filtering with a AC capacitor on the transformers output side will clean that up considerably.

Without knowing anything about the transformers inductance and whatnot it will be a bit of trial and error testing to find the best capacitor value though.

 

[audioguru]

Nearly all the pulses have the same width. The waveform looks like Pulse Amplitude Modulation, not Pulse Width Modulation.

 

[Ramussons]

Would'nt it be simpler to use a 25 Hz Collpitts Oscillator (Phase Shift Oscillator) as the source and drive the mosfets in a linear class ab mode? After all, the power requirements are very small.

Ramesh

 

[Odysseas]

Thanks everyone for your responses so far! The input signal to the MOSFETs is a PWM signal coming from an Attiny44, using the Timer0 module. I will snap a Screenshot of the signal coming out of the tiny, and post it here once I am home again. I could also post the software I have written for the μC but I doubt that would help anything.

I have also considered using a simple, integrated BTL audio amplifier and feed it with a sine signal. The problem was, where do I get such a signal from? All the circuits I came up with are built to oscillate well into the kHz range, certainly not at 25Hz. But if someone can suggest such a circuit, I am open to that idea.

@alec_t: if I would reverse the phase on one of the inputs, wouldn't I then also flip over one half of the sine wave? And then, wouldn't the transformer eventually enter saturation? I would effectively place the two input windings in parallel.

 

[alec_t]

if I would reverse the phase on one of the inputs, wouldn't I then also flip over one half of the sine wave? And then, wouldn't the transformer eventually enter saturation? I would effectively place the two input windings in parallel.

Indeed. I was only playing around with the sim. My bad.
Here's the result for a revised simulation (with correctly phased windings, I hope!). PWM FET drive is simulated by voltage sources (a bit clunky...someone out there may have a more elegant way). PWM frequency is 10kHz. D3/4 are freewheel diodes. R1C1 provide smoothing. Waveform distortion is predominantly 3rd and 5th harmonics.

 

[Odysseas]

Thanks a lot for your efforts! Could you do me a favour real quick and see if there is substantial current flowing through the freewheel diodes? I also thought about installing them, but doesn't producing an AC signal in one secondary winding inevitably induce voltage in the other winding? If so, the freewheeling diode would short circuit that voltage and waste a lot of power.

 

[ericgibbs]

hi alec,
This is my LTSpice PWM folder, thought it may be helpful.

E.

 

[alec_t]

@Odysseas
Here are the currents:

(Bear in mind that I've only guessed inductance etc values for the windings, so you would need to scale the currents according to your actual transformer. But at least the relative magnitudes of the currents can be judged.)

@Eric
Thanks for the folder. I thought you might have something up your sleeve

 

[audioguru]

I have also considered using a simple, integrated BTL audio amplifier and feed it with a sine signal. The problem was, where do I get such a signal from? All the circuits I came up with are built to oscillate well into the kHz range, certainly not at 25Hz. But if someone can suggest such a circuit, I am open to that idea.

ANY half-decent audio amplifier produces a 25Hz output. Most audio amplifiers go down to 20Hz. A Wien Bridge, Phase-shift or Bubba oscillator produces a good sinewave at 25Hz.
Since you need 60VAC then an audio amplifier that produces 68W into 8 ohms will work (LM3875 with a 70V supply) and it will heat with about 7W.

 

[Odysseas]

@alec_t: Am I correct in thinking that when one secondary winding is active, the other one will induce a voltage in phase with the active voltage? If so, that voltage would be short circuited be the freewheel diodes. I am hesitant to install them...

 

[alec_t]

I'm not sure the sim paints the whole picture, but when L1 is active there is no current in D4. Likewise when L2 is active there is no current in D3.