FFT question ...

[summitville]

Can anyone interpret the attached FFT File? This is the FFT analysis of a Bubba Oscillator making a 60Hz Waveform. I see a single peak at 60 Hz, the lower frequencies drop below -30db and the higher frequencies drop below -45db. Is that GOOD? Does the FFT graph indicate that the Bubba Oscillator is creating a true sine wave with less than 1% Distortion?

Thank You,

 

[mneary]

For starters, notice the 60Hz waveform is -8db. There is no horizontal scale; is it log or linear? Would be nice to see your .asc file including a model of the transformer, and the model of your load.

 

[summitville]

"... There is no horizontal scale ..."
I see "LOG Frequency" at the bottom of the FFT Graph when I scroll down.

"... model of the transformer ..."
There is no transformer, only Op-Amps & R-C filters.

"... model of the load ..."
The Bubba Sine Wave output is taken from the Low Impedance output of the Op-Amp at the Final (5th) Stage.
The Load will be High-Impedance Inputs of two other Op-Amp circuits.
The next Stages (Absolute Value & Zero Cross) are not designed, yet.

I can upload asc file later tonight.
I used this bubba schematic on page 5 as my starting point ...
https://www.electro-tech-online.com/custompdfs/2009/12/aug_07.pdf
But I added a 5th Op-Amp (Unity Gain) between
the "Cosine" R-C stage and the next "Rg-Rf" Stage.
This way, no R-C circuit was "loaded" by the next stage.
I took my Bubba Sine Wave from the output of this new 5th Op-Amp.
It is very clean when compared to the Sine Wave of the LTSpice Component.
I changed the R-C to 25.5k & 0.1uf = 60hz, if I remember right?

I had read some articles about computing %THD from FFT Graphs.
But their FFT Graphs only had 6 clean spikes at the Harmonic Frequencies
not the squiggly FFT graph I get from LTSpice.

 

[mneary]

I had to save the document, change 'ownership', and open it for edit before I could see the horizontal scale. Maybe that's why most people use pdf.

No transformer? Then this isn't an inverter then, is it? I'm surprised to see this thread in the "Renewable Energy" forum....

The document goes from page 33 through 37; do you mean Figure 5? I see a fairly well designed wein bridge circuit. Surprised that it needed any filters at all.

 

[RCinFLA]

Just an oscillator with not too stellar sideband noise.

Depending on your use, it might be just fine.

Noise in the active devices, resistors, and jitter caused by cycle to cycle slight variation in the transition threshold causes the frequency domain sideband noise.

 

[summitville]

nmeary, The best Wein Bridge Sine Wave circuit that I could design in LTSpice was much worse than the attached 5-Stage Bubba 60Hz Pure Sine Wave Oscillator. Do you know of a Wein Bridge (60 Hz) Osc design with less than 1% THD?

5-Stage Bubba Osc:

The first four Harmonics of the attached Bubba Osc computes to (if my math is correct) only 0.7% THD (Total Harmonic Distortion). Which I think is great!

R1=R2=R3=R4=26.45k (ideal value)
Adjust R's for your actual circuit as best you can measure 60Hz or count pulses over a very long but accurate time period.

C1=C2=C3=C4=0.1uf (Temp stable)

Add By-Pass caps to Op-Amps & a well regulated Voltage Supply, etc.

If circuit does not Oscillate then slightly Increase R10 and/or decrease R11.
You only need a slight clipping of the Peaks at "U1-Output" for a stable & pure Sine Wave.

The FFT analysis & harmonics are ...

Low Freq Attenuation = < -60dB

56Hz & 64Hz Cut-Off = -50dB
60Hz Peak = -2dB

Harmonic peaks ...
120Hz = -47dB
180Hz = -56dB
240Hz = -53dB
300Hz = -47dB
360Hz = -70dB
420Hz = -61dB
480Hz = -60dB
540Hz = -56dB
600Hz = -64dB

Any harmonic at or below -60dB is basically insignificant.

Attached are two PDF's ...
1) 5-Stage Bubba Osc Schematic
2) FFT Analysis

 

[mneary]

I re-read the TI document you sent, and found the "Bubba" oscillator on page 37 instead of 5 as you said. I see now that Texas Instruments has re-named an age-old phase-shift oscillator "Bubba" and it has no connection to the "Bubba Inverter" thread here, in this Renewable Energy topic.

I was confused until I realized that this thread does not have anything to do with Renewable Energy.

 

[summitville]

How the 0.7% THD Bubba Sine Wave Oscillator will be used in my Multi-Cell Pure Sine Wave Inverter is made apparent in the attached jpg file. Three indentical Low-Power High Frequency DC-DC Converters are connected in Parallel to the 48V Battery but their Secondaries are in connected in Series. This makes the total I^2R Losses of the three Low Power DC-DC Converters only 1/3rd of a single DC-DC converter, 66% LESS Heat Loss. The DC-DC Converters also run at 120 Degree Phase Shift to smooth out the current surges.

To double the Output Watts (1KW -> 2KW) just add three more DC-DC Modules, parallel connect the rectifed DC Outputs at C1 and upgrade the H-Bridge, as needed, for more amps.

To double the Output Voltage (120V -> 240V) just add three more DC-DC Modules, series connect all 6 rectifed DC Outputs and upgrade the H-Bridge, as needed, for more voltage.

This is a very modular, small, lightweight and cool running design.
A true Modular N-Cell Inverter would have separate H-Bridges for each Secondary.

If you don't need a Pure Sine Wave output then drop the Bubba Osc and use a simple SQ Wave PWM chip on the output.

A very similar design was reported to have 96% efficiency.

 

[Mr RB]

...

Three indentical Low-Power High Frequency DC-DC Converters are connected in Parallel to the 48V Battery but their Secondaries are in connected in Series.
...

The DC-DC Converters also run at 120 Degree Phase Shift to smooth out the current surges.
...

Sounds good in theory.

Have you actually tried running 3 high frequency transformers in series and 120' out of phase??

 

[summitville]

No, I have not tried running 3 high frequency transformers in series and 120' out of phase. At first glance, I do not think your suggested design change would work for my Inverter unless you also changed the secondary portion of the schematic to a Wye or Delta config - that might work for you. A Wye Design? Hmmm, that could be an interesting alternative, though. You could rectify with 6 Diodes just like the output from a PM Alternator.

In my schematic (in previous post) the the Secondary side of each DC-DC Converter (the Rectified DC Outputs at C2, C3 & C4) are all wired in Series to make a Voltage Tripler. I think I will stick with that design for now.

 

[mneary]

To maintain the impressive THD from the "Bubba" (known outside TI as phase-shift) oscillator, you'll need an equally perfect Bubba PWM subsystem.