A "Linear" DC to AC converter idea - Thoughts?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Ethan

Administrator
Admin
Moderator
Joined
Jun 3, 2004
Messages
1,602
Location
DC
I had a thought this morning about creating a simple DC to AC converter (simplified schematic attached).
Given a regulated +5V/-5V bipolar supply, imagine the switch symbols are MOSFETs controlled by an MCU that alternately switched from positive/negative rails at near zero-crossings (detected with a window-comparator at the output node indicated by "sine?").
Since the charge/discharge time-constant would vary depending on load, the size of the output capacitors (maybe additional Rs) would have to be "tuned" to be application specific.

Any reason why this would be a bad idea?  I realize it would take some tweaking to get low crossover distortion.
I also realize there are ICs that could help do this more efficiently, but just as an educational experiment, could this work?  Or is it a terrible idea?

Thanks!
 

Attachments

  • Screen Shot 2013-05-27 at 10.19.10 AM.png
    Screen Shot 2013-05-27 at 10.19.10 AM.png
    18.4 KB · Views: 21
Your circuit is missing some parts.

Alternately switching capacitors to DC supplies, further assuming there is some kind of load across the output will create a sawtooth like output waveform as the cap resets to 5V then is discharged by the load.

Perhaps if you added inductors in series with the switches you could shape an output waveform. Crossover distortion would be the least of your problems.

JR
 
Thanks John,

Hmmm... for some reason I was thinking the charge/discharge cycles would be logarithmic, while the load/and capacitor combination would determine the charge/discharge frequency.
 
I seem to remember a  Amp ad(Crown?) that had a picture of a AC drill being powered by oscillator driving an amp.  Might have been an ad from the 70's?  Anyone remember this? 

So what about a class D amp supplying 60Hz AC?
 
Well, my half-baked thought this morning was to experiment with creating a simple method of getting  AC (low current) from solar panels without using an off-the-shelf "inverter".  Aside from that there's no concrete application for this yet, just thought I'd explore the potential for the idea.
Thanks for the comments!
 
In general stepping it up to 120VAC just so some appliance can step it down again is not the most efficient but in the spirit of this thread the DIY solution is along the lines I started you on... A modest voltage class D amp  also known as a DC to AC convertor.  You don't need very low distortion just a 60Hz or so sine wave or sine like waveform.

If you are indeed working from 5 or 10V of DC, the DIY solution is to repurpose a common power supply transformer, but drive it backwards so 5 or 10VAC into the (was) secondary makes 120V output at the (was) primary.

Relatively easy lifting.. You need a simple 60 Hz sine wave, a crude class D amp, and a simple gain element to detect the 120V output and regulate the input voltage to make it so.  For DIY you may be able to cut some corners since the HF PWM square waves won't pass through a 60 Hz transformer you don't need much of an output filter etc.

I think National developed a chip set to regulate the DC from different solar panels that don't all put out the same voltage due to getting less light or aging or whatever so you can stack or parallel and they all play nice together. .

Have fun...

JR

PS: won't bore you with "I'd probably do that with a $2 microprocessor"... but like you know...  8)
 
For small "120V AC" for un-fussy uses, *if* you can get to 12V DC, you can NOT beat the car cigar-lighter inverters. The efficiency is decent (because it reduces costs by making less heat allowing smaller parts). The price for a whole package unit is less than a few MOSFETs and a CPU.

True *sine* from DC is much tougher (and not near what you drew). A classical approach is a tuned-circuit oscillator, but a 60Hz tuned-circuit is a lossy monster. However a simple square-wave works different for lamp-loads and electronic (peak-catcher) loads. Most "sine-like" inverters settle for a 5-step approximation where DC supplies get the 168V peak they expect while lamps get a 120V average (RMS) to keep them happy.

> a AC drill being powered by oscillator driving an amp.

Drill, bah. I had power-amps retired from driving film-projectors at 59.9Hz. I have also run 104Hz through a power-amp and filament transformer to hot-rod a 33/45 record-player to 78.

Yes, this can give a "perfect sine". With classic class-B power amps you are lucky to beat 50% efficiency (78% on paper, but losses lurk everywhere). Anyway nether drills nor turntables *need* a pure sine. (In fact most power hand drills will run very well on 120V _DC_, except the switch will arc worse and may die within warranty.) (True AC sync-type motors will run on square waves, though optimum voltage and torque will differ, and hash will be a problem.)

If you are powering DIY PCs, there are power-lumps that take 12V (clean or car) and directly make the 12V 5V 3.3V etc on the regular PC-plugs. (However I had one in 24/7 service for 5 years and that PC began rebooting randomly.)
_______________________________________

> Any reason why this would be a bad idea?

It won't work. The polarity across the load will *never* reverse. You need at least two more switches to get there.

If your goal is 120V from +/-5V, you also need some lever to jack the voltage.
 
Thanks for all the comments.
My intent wasn't to take 5v and jack it up to 120v. I just happened to draw it with +/-5v because it seemed like a convenient voltage to experiment with the idea.
Although, reading over the comments I'm now curious just how much off the shelf inverters deviate from perfect sine wave output and how much the average consumer device could tolerate.

I've never tried/thought about rectifying a bipolar square wave.  I would imagine with sufficient smoothing, it could be perfectly useable?
Sorry if some of this is overly simplistic. There are big gaps in my elec. theory knowledge.
Just curiosity, albeit without any specific application.
Thanks!
 
Ethan said:
Thanks for all the comments.
My intent wasn't to take 5v and jack it up to 120v. I just happened to draw it with +/-5v because it seemed like a convenient voltage to experiment with the idea.
Keep in mind, power = power so stepping 5 volts up to 120 means 24x the current on 5V side.
Although, reading over the comments I'm now curious just how much off the shelf inverters deviate from perfect sine wave output and how much the average consumer device could tolerate.
Quite a lot. while premium designs tout "true" sine wave output. The rectifiers don't care and an actual square wave would mean zero ripple, but transformers do not have DC response so can't pass perfect square waves either.  Some inverters use a few different taps on transformer windings to make a stepwise sine wave approximation from single DC input.
I've never tried/thought about rectifying a bipolar square wave.  I would imagine with sufficient smoothing, it could be perfectly useable?
Sorry if some of this is overly simplistic. There are big gaps in my elec. theory knowledge.
Just curiosity, albeit without any specific application.
Thanks!
The mains wave shape doesn't matter unless it does...

JR
 
The rectifiers don't care and an actual square wave would mean zero ripple
That's what I would have thought but was a little embarrassed to ask for confirmation on that point.

The mains wave shape doesn't matter unless it does...
Understood. :)

How about the attached drawing? Conceptually, would this work?
Or am I just taking a bassackwards approach to what an actual inverter would do?
(I included the output coupling cap just to indicate that I'm expecting an "AC like" signal)

Thanks!
 

Attachments

  • Screen Shot 2013-05-28 at 12.12.22 PM.png
    Screen Shot 2013-05-28 at 12.12.22 PM.png
    56.6 KB · Views: 14
That would make a 120Vp-p square wave which is less than half of what main voltage sine wave makes p-p, (115x 1.4x 2=322v p-p).

To transfer power at 60Hz that cap in series with the output would need to be pretty large. Think of ripple in a power supply. While the driven side of the cap would look like a square wave the output side, depending on load, would decay toward 0V, just like PS ripple..

Yes it would work if it had more input voltage (or a step up).

JR
 
JohnRoberts said:
That would make a 120Vp-p square wave which is less than half of what main voltage sine wave makes p-p, (115x 1.4x 2=322v p-p).
Ah, yes! Doh! :)

To transfer power at 60Hz that cap in series with the output would need to be pretty large. Think of ripple in a power supply. While the driven side of the cap would look like a square wave the output side, depending on load, would decay toward 0V, just like PS ripple..
Yes, that make sense.

Yes it would work if it had more input voltage (or a step up).
Higher input voltage wouldn't be tough to get, just more cells in series...but I guess this could get impractical pretty quickly.

Thanks for taking the time!
 
If you are talking about switching/chopping it already, not heavy lifting to make a crude PWM class D amp to drive a step up transformer. This doesn't need to be audio quality.

I do not have practical experience with actual solar panels but they do vary quite a bit in output depending on clouds, sun angle (time of day), etc.

If parallel do the bright ones drive current into dark ones? I don't know. It seems stacking in series eliminates that issue.

I'd use a step up, and regulate the voltage to deal with solar output variations.

Perhaps you should apply to the government to get a federal grant to develop this. Say you want to charge an electric car.

Common problem with solar is reliably charging batteries, so should be lots of DC to DC art kicking around for that. Inverters are also mature technology.

JR
 
JohnRoberts said:
This doesn't need to be audio quality.
Absolutely not.  Now that I have an idea of what I'd like to use this for, it's just about power and efficiency.

I do not have practical experience with actual solar panels but they do vary quite a bit in output depending on clouds, sun angle (time of day), etc.
This thought did cross my mind, but I was thinking of regulating down about 10% or so...or perhaps it would be more efficient to monitor the voltage coming from the panels and establishing a hard on/off threshold.  For instance, the system would just cease to operate below a certain voltage, but that seems wasteful as well.  The more I think about this there needs to be some storage system (battery). That would allow for the possibility of stepping up the voltage and storing it even when the panel output sags.
If parallel do the bright ones drive current into dark ones? I don't know. It seems stacking in series eliminates that issue.
If paralleling, it would probably be wise to use some current management scheme, similar to what they use for car batteries in parallel.

Perhaps you should apply to the government to get a federal grant to develop this. Say you want to charge an electric car.
Now I think you're just making fun.  ;)
Inverters are also mature technology.
True. I should probably start by getting a cheap inverter and running some tests to see if I'm just wasting my time trying to reinvent a square wheel.
Thanks!
 
> If parallel do the bright ones drive current into dark ones?

Yes. A panel is a series string. If battery voltage is higher than panel voltage, panels are seriesed. Only very large systems go to parallel connection. A diode in each series-string lead blocks back-current into dark panels. (I'm not sure the diode is "required". Seems to me a dark cell should block its nominal voltage. It may be a practical thing: solar is stupid unless the cells are very cheap. Cheap cells imply dirty silicon in older factories with less-trained workers and loose rejection specs. So leaky cells may be common, and a fat block-diode is cheaper than better cells.)

Remember that each cell is 0.6V so you "must" series to get to useful voltage. And that means a lot of connections, and each connection is tricky (soldering on fine traces). OTOH, the area of each cell is fairly unlimited. Which means the current can be quite large. So the optimum panel is entirely series, until you get to humongous powers larger than standad cells/panels offer.
 
Ethan said:
If paralleling, it would probably be wise to use some current management scheme, similar to what they use for car batteries in parallel.
I don't know how to do that... A shorted cell could cause a battery fire.
Now I think you're just making fun.  ;)
of the government not you...
True. I should probably start by getting a cheap inverter and running some tests to see if I'm just wasting my time trying to reinvent a square wheel.
Thanks!

If completely off grid, charge a lead acid battery, then inverter from that... big dog installations drive power back into the AC mains to run meter backwards and save the cost of storage.

JR
 
Back
Top