OPA547F Waveform Generator

[_-_-_Adam]

Hi All,

I'm new here and thought I'd share what I'm doing.

I'm designing a 73.5v Pk-Pk 1kHz 50mA waveform generator (sine only).

This is the output stage (it still has the OPA452, but I've decided to use the OPA547 as its more easily available).

An externally hosted image should be here but it was not working when we last tested it.

🙂

 

[_-_-_Adam]

Here's the PSU.

An externally hosted image should be here but it was not working when we last tested it.

 

[analog_sa]

The only thing missing seems to be the waveform generator 🙂

You can obviously also use any of the "chip amps" in your bridge amp.

 

[_-_-_Adam]

here it is, it's a slightly hacked Miniosc Miniature Audio Oscillator

An externally hosted image should be here but it was not working when we last tested it.

I've been nutting out why I get 2.5% THD.

I've placed C? across R10 (10nF MKT).
All other caps are polystyrene (except bypass).

D1 ~ D4 were 1n4148 but the temperature stability was woeful so I changed them to some 11DQ03's that somehow magically arrived this also greatly reduced the unwanted Harmonics.

I've managed to bring it down to 0.6%

Amplitude stability is still pretty bad 🙁 though and I'd like to get the THD down a bit.

 

[Mooly]

Dare we ask why you need 73.5 volts output 😉 and from a bridge output too as that must limit it's connectivity.

Just wondered 🙂

 

[Samuel Groner]

I've been nutting out why I get 2.5% THD.

There are several reasons why your plan shows poor performance. The most important ones:

* You need a better plan for the level stabilisation. These clamp diodes necessarily introduce frightening amounts of distortion. As a simple fix you could probably replace them with a diode bridge which drives a zener (there's something in one of Jung's audio opamp books, details on request).
* I'm not particularly familiar with the oscillator topology you use (I'd rather start with a state-variable filter) but probably you need some sort of trim to pre-trim the Q of the filter. Otherwise you might get excess drive to the diode clamp.
* TL064 likely show rather high levels of common-mode distortion, particularly at low supply voltages. Use +/-15 V and a decent opamp type (I suggest you try TLE2704).

All other caps are polystyrene.

The choice of capacitors becomes important (assuming a *reasonable* cap to start with and not something like ceramic X7R) once you approach 0.001%.

Samuel

 

[Samuel Groner]

Another one: How does an opamp with +/-24 V rails provide an output of 36.8 Vp..?

Samuel

 

[_-_-_Adam]

Opamps are gained for 18v Peak with 180 degree phase inversion.

So the differential voltage is 36v Pk (but still only 18v pk to GND).

When Opamp 1 is at +18v, Opamp 2 is at -18V. (My bad labelling the 26Vrms, as it's actually the differential voltage, not the pin-gnd voltage)

 

[_-_-_Adam]

a diode bridge which drives a zener (there's something in one of Jung's audio opamp books, details on request).

I would have thought that would just raise the clamping voltage, or are Zeners less noisy? Any info appreciated.🙂

TL064 likely show rather high levels of common-mode distortion, particularly at low supply voltages. Use +/-15 V

I'll bump up the rails and have another look at the TL064 before I rip it out of the circuit

Thanks for the feedback.

 

[_-_-_Adam]

Dare we ask why you need 73.5 volts output 😉 and from a bridge output too as that must limit it's connectivity.

Just wondered 🙂

Sure I'm working with these things
resolver - Google Search

 

[Mooly]

Sure I'm working with these things
resolver - Google Search

Thank you 🙂
I have not come across those before.

 

[Samuel Groner]

My bad labelling the 26 Vrms, as it's actually the differential voltage, not the pin-gnd voltage

And it's the peak-peak voltage, not RMS... The 1 Vrms net label ain't particularly meaningful either. At first these things are unimportant, but they make reading of the schematic difficult for others (which means you'll get less help here).

I would have thought that would just raise the clamping voltage, or are Zeners less noisy?

Reverse-biased junctions are far more noisy than forward-biased ones, but that doesn't matter. The key point is that the limiting element should have a very hard characteristics. Soft limiting generates more harmonics.

After a fresh look at your diagram I suggest the following steps:

* Remove the 3.9 nF cap labeled C?; I presume it may have a significant effect on distortion cancellation. Any reason why you introduced this part? It ain't in the original schematic.
* Large Schottky rectifiers are surely the wrong parts for D1-D4; try something like 1N914B.
* If this still gives insufficiently low distortion levels its probably time to add a zener from the anode of D1 (cathode of the zener) to anode of D4. Try a 6.2 V part or an IC reference such as LT1634-5. This will raise the level but this is easily handled by reducing gain in the output stage. Perhaps there are other changes needed to make this work, you'll need to try it out.

Samuel

 

[Samuel Groner]

Large Schottky rectifiers are surely the wrong parts for D1-D4; try something like 1N914B.

I see you already experimented with these. But try again after the C? cap has gone, and at least try small-signal Schottkys instead of the power rectifiers.

Samuel

 

[beun]

When you need extremely low distortion use a two-integrator oscillator like this example:
Operation explanation of the sine/cosine wave oscillator

For amplitude control change R1 with a low voltage lamp which will act as a PTC resistor thereby stabilizing the voltage. You may have to experiment with the lamp and R2 to get an acceptable voltage. Distortions down to 0.001% are possible with this scheme.

 

[beun]

Or alternatively use a CD player with a test disc to generate a sine.

 

[_-_-_Adam]

I see you already experimented with these. But try again after the C? cap has gone, and at least try small-signal Schottkys instead of the power rectifiers.

Samuel

I'll post some Spectrum results from the different configs.

 

[Samuel Groner]

Another simple improvement: put a 1 nF cap in parallel of R2 and R4. This will also somewhat reduce level (reduce R1/R3 to correct this) and might affect capacitive load driving capability of the output stage.

Samuel

 

[Samuel Groner]

Any news on this? I've looked a bit more into diode limited oscillators and I think I've been wrong that the zener bridge would reduce distortion. I think it will provide more stable levels at least. Distortion-wise there's not much to do with these approaches; 0.02% is about the lowest figure reported for SV topologies.

Samuel

 

[_-_-_Adam]

Not yet, I'll post some stuff within the next couple of weeks and the spectrum plots as promised.

 

[_-_-_Adam]

Hi all,

First up I replaced the TL064 with an LM348 (what I could get me hands on).
I'm rebuilding the circuit using 15V zenners in place of the 5.1V zenners, this bumps the + & - rails for the Opamp up to +-15 Volt supply.

Now I can use 5.6V zenners for the clamp in the Weinbridge, this will give me the best amplitude stability versus temperature (It's bad with the 11DQ03's but much worse with the original 1n4148 diodes)

The 5.6v zenners balance between the Zenner effect and the Avalanche effect which both have opposite temperature coefficients.

The output of the OPA547 power op amps some DC offset so I placed some 4k7 resistors to cancel the DC offset out and this worked a treat, I've got some 1nF caps on order to place across the 120K feedback resistors to limit the bandwidth, as the 3.3nF I had lying around affected the gain too much at 1kHz.

updated schematics to follow soon 🙂.