# simple small sinewave ups

#### AnalogKid

Joined Aug 1, 2013
10,054
The problem is that even though I have a whole-house generator, many devices (clocks and cable boxes) lose their brains in the 15 seconds it takes for the gen to kick in. For a while I have been pondering a small (<10 W) ups project to sprinkle about the house, because I have six 6 V SLA (sealed lead-acid) batteries and a butt-load of electronic components. Of course, per my license plate, it should be a true-sinewave device. This should be more than enough to run a resurrected old digital clock in the garage, the dvr cable box, etc.

This idea might not be new, but it came to me in a flash of lawn mowing: an ebay 10 W class D mono amplifier module as an output transformer driver. Crystal oscillator, binary divider with the last few stages driving stepped resistors to create a 5% distortion sinewave, little lowpass filter, class D module, 6.3 V filament transformer, 115 Vac output. No load switching at power-out - the circuit would be running constantly, with the oscillator/divider being reset constantly by the incoming power line so there is zero output disruption when power fails.

Is this a way to get a true sinewave, online UPS, at near stepped squarewave efficiency, without designing a switching inverter stage from scratch; or am I missing something? Assume I've got the AC input power stage covered.

Thanks.

ak

Last edited:

#### BobTPH

Joined Jun 5, 2013
4,897
Your sine wave has to be 162V peak. A 6.3V to 115V transformer will not do that from a 6V battery. You would get 109V peak or 77V RMS out of it.

Other than that, the idea of using a class D amp for a low power inverter does not sound unreasonable.

I am not sure I understand the part about creating a sine wave from a divider and resistors. Are you talking about using the low bits of the counter and making a customized D/A to approximate a sine wave? That would be a new idea to me.

Bob

#### Alec_t

Joined Sep 17, 2013
12,805
I am not sure I understand the part about creating a sine wave from a divider and resistors.
Here's an example of the sine approximator (albeit using a shift register rather than a divider).

#### crutschow

Joined Mar 14, 2008
29,776
So for 115v out of a 6.3v filament transformer you could bump the battery voltage up to about 10Vdc with a boost regulator, and then feed that to a D audio amp with a bridge output.

#### BobTPH

Joined Jun 5, 2013
4,897
Here's an example of the sine approximator (albeit using a shift register rather than a divider).
Cool, you learn something new every day.

In this case, the resistor network is taking in the last 16 low bits of the counter output. I wonder if there is another network that would use the low bits of the counter to do the same, or maybe AK was planning on pumping them into a shift register like that circuit.

Another way would be to use the counter low bits to address a ROM.

Personally, being micro-kid instead of analog-kid, I would just use a micro to generate a sine PWM sequence.

Bob

#### crutschow

Joined Mar 14, 2008
29,776
Some of you devices, such as the cable box, may take more than 10 watts.

#### AnalogKid

Joined Aug 1, 2013
10,054
Error: 10 Watt *stereo* amp, bridge-tied load.
Are you talking about using the low bits of the counter and making a customized D/A to approximate a sine wave? That would be a new idea to me.
Here's an example of the sine approximator (albeit using a shift register rather than a divider).
I remember that circuit from back in the 90's; it's one of the options I'm considering. Another uses a CD4017 Johnson counter, another uses a binary up/down counter where the count direction and the n+2 bit gates an inverter for a 64-step cycle, another has transistors as log elements in an opamp feedback loop and turns a triangle into a sine. Or start with a 60 Hz square wave and run it through an 8-pole switched-capacitor filter.

I think Ron Mancini had a large app note on sinewave techniques. I know Jim Williams did. So many circuits ...

ak

Last edited:

#### crutschow

Joined Mar 14, 2008
29,776
I would expect that some distortion in the sinewave will have no significant effect on the operation of the devices being powered, so it should not require a lot of steps in the waveform, particularly after running it through a 3-pole active LP filter.

Alternately the integration of a triangle-wave, should give only a few percent distortion.
(I would assume the generation of the triangle-wave and its numerical integration could be done by a microprocessor(?). I don't know if that's simpler than using a look-up table.)

#### AnalogKid

Joined Aug 1, 2013
10,054
I would expect that some distortion in the sinewave will have no significant effect on the operation of the devices being powered
You're right, of course. This isn't an Audio Precision distortion analyzer front end. My guess is that the 16-bit shifter circuit above but scaled back to 8 bits, with a gain-of-two 2-pole filter (my favorite Sallen-Key config) would be more than enough noise and harmonic control. Since I don't care about phase distortion, an active T or twin-T bandpass could do it, or maybe a gyrator. Hmmm, I might have to sim that ...
(I would assume the generation of the triangle-wave and its numerical integration could be done by a microprocessor(?). I don't know if that's simpler than using a look-up table.)
But not as simple as an LM358.

I should have been more clear about the design rules. This is, among other things, and inventory reduction project. Four UPSs, \$0.00 new expense.

Discrete anything - good.

Analog anything - best.

ak

#### BobTPH

Joined Jun 5, 2013
4,897
Hopefully, you don't have too many electromechanical buzzers you want to get rid of

Bob

#### crutschow

Joined Mar 14, 2008
29,776
gain-of-two 2-pole filter (my favorite Sallen-Key config)
You might consider a 3-pole Sallen-key, which is basically a 2-pole filter with an added RC LP stage at the input. This RC filter rolls off any sharp edges in the waveform to minimize the bandwidth requirement for the opamp. Otherwise the high frequency components of the sharp edge can feed-forward through the feedback cap.

And a 3-pole filter may be all you need for a good sinewave from a square-wave.

Below is the LTspice simulation of a square-wave through a gain-of-two, 3-pole, 60Hz, Butterworth filter.
The Fourier analysis shows <1.7% distortion, which is negligible for power-line distortion.

#### Attachments

• 2.3 KB Views: 2
Last edited:

#### AnalogKid

Joined Aug 1, 2013
10,054
Yeah, but - the two-pole gain-of-two, when slightly peaked at f0, has all equal valued resistors and capacitors. For something as low rent as 60 Hz power, that's hard to pass up.

In the background I'm still wavering between the sinewave and class-D transformer driver, and a royer-esque two-transistor modified square wave output, driven by two out-of-phase squarewaves. I've got a stack of big fat power MOSFETS that would be very efficient. It might be the transformer output voltage that makes the decision, because I don't want to do a full bridge driver.

Nice table. Your Sim-Fu continues to impress.

ak

#### crutschow

Joined Mar 14, 2008
29,776
I've got a stack of big fat power MOSFETS that would be very efficient. It might be the transformer output voltage that makes the decision, because I don't want to do a full bridge driver.
So bump the DC voltage up to about 20V to power a MOSFET push-pull output, drive the MOSFETs with a sine-wave modulated PWM, and couple the MOSFET's output into the 6.3Vac transformer through a large capacitor to block the DC component.

#### JMW

Joined Nov 21, 2011
137
In the late 70's RCA had a CTCSS circuit that only used precision precision capacitors and resistors. IIRC the models were TacTec portables, Tac 810 and 310. They developed frequencies between 67 and 210 Hz. 60 would not be out of line

#### MisterBill2

Joined Jan 23, 2018
11,877
All of the responses have lost sight of the "simple" part. The simple solution is not very efficient but it is reliable in that any failures develop slowly over time. That is, and on line power system, a motor generator with a big flywheel to provide a couple of minutes ride-through while the diesel generator starts and comes up to speed. In the "normal" mode line power keeps it spinning, at power loss the flywheel holds the speed until the battery drive switches in, and then after a bit the external generators start delivering power. So there is never a millisecond of power outage.Of course it is neither highly efficient nor cheap. But it is good.
" Good, reliable, efficient, or cheap" Pick two.

#### crutschow

Joined Mar 14, 2008
29,776
a motor generator with a big flywheel to provide a couple of minutes ride-through while the diesel generator starts and comes up to speed.
And that's your definition of "simple"?

#### AnalogKid

Joined Aug 1, 2013
10,054
All of the responses have lost sight of the "simple" part.
Well, certainly one of them has.

ak

#### DickCappels

Joined Aug 21, 2008
8,704
One reminder about the transformer: running a filament transformer "backwards"it rarely gives you the output voltage you expect based on the assumed turns ratio. The secondary (in our example the 6.3 volt winding) was not intended to provide magnetizing current to drive the core and as a result of this magnetizing current (which can be shockingly large) times the winding resistance will result in a voltage drop across the resistance of the driven winding and thus provide a lower P-P output voltage than a simple calculation based on the turns ratio would indicated.

Since you plan to build several UPS's you run the risk of having a junkbox full of unsatisfactory transformers, so it is best to only one or two transformers at the start for testing.

A transformer's regulation specification (change in output as load is switched in and out) seems to be a good indicator of which transformers will perform better in this kind of application.

#### MisterBill2

Joined Jan 23, 2018
11,877
And that's your definition of "simple"?
OK, "simple" as in not having a lot of electronics as part of the package.
And now referencing the original post, I have seen all of our cable boxes reset themselves after any sort of power outage, so it does not seem that they have a challenge in that area. And for the digital clocks, we have quite a few in the house with battery backup available. Adding a backup function to digital clocks should not be that complex, unless keeping the display lit at the same time is desired. Really, to ride through a short outage just an isolation diode and a bigger filter capacitor should be able to handle an outage of a minute or so.
Going to a line voltage backup is a lot more complex than just adding a backup DC source for the time-keeping electronics portion. In addition, not needing to deal with mains voltages and switching is a big simplifying change..

#### BobTPH

Joined Jun 5, 2013
4,897
He also said he wants to use parts he already has. I, at least, do not have any excess diesel generators, alternators or large flywheels in my parts cabinets.

Bob