Super Low Distortion Ultra Pure Audio Oscillators Revisited

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by mediatechnology »

Trimming a Twin T to null its output.
How difficult, it is ?
How many iterations going around at the two trimmers, does it converge smoothly to the final result ?
I just built the second one and it took maybe 30 seconds.
You just dip each one - the first iteration is only a few dB.
Rinse, lather, repeat and it converges pretty fast.
Maybe 5 to 6 iterations.
I use a scope for it - doing it with the FFT is slower.
You should have noticed the two 1K trimmers have different sensibilities. There is a x2 difference in how hard they pull
True.
I had a whole tube of 1K trims and wanted to use them on my Protoboard.
Also wasn't sure how much range I'd need.
I do have a 500Ω in as a gain trim so I should make the BOM have two.
In earlier posts, you mentioned "differential notch filter."
Is it:
Two twin T connected together at a virtual ground node.
Or
A twin H. That would be made of five 16K resistors and five 10nF capacitors.
I want to make a "twin-H" where no filter element is grounded.
You could also call it a "twin-twin-T" or "double twin-T."

What I have now is two twin-T connected together at ground which is indeed not balanced.
For checking the oscillator's balanced output I decided to try them differentially by hanging a CMR stage on the output.

The twin-H INA would obtain a bias current path through the twin-H to the input connector which would have 100K to ground bias resistors.
The 100K input resistors and the post-INA CMR stage would be the only points to touch ground.
This maintains a common mode gain of "1" in the INA and which then provides CMR in a post-INA CMRR stage.
My MM phono preamp board would be a good way to follow the twin-H.

I have enough 10 nF 1% to make a true twin-H.
I think I need scale the resistor values to maintain the same Fc.
Do 5X 15K9 and 5X 10 nF sim at 1 kHz for differential inputs?
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by terkio »

Do 5X 15K9 and 5X 10 nF sim at 1 kHz for differential inputs?
Yes.

I Simed with differential inputs. 5 X 16K and 5 x10nF: Notch -90dB at 994.7191 Hz

Sim with 5 x 15.9K: Notch -96dB at 1.001 KHz.

More info:
The transfer function of the Twin H is certainly the same as the Twin T. From the simulation, I see exactly the same insertion losses at 2KHz, 3 KHz, 4 KHz,

I think the Twin H is pretty cool with, all caps the same value and all resistors the same value.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by mediatechnology »

Thanks for the sim!
I just checked the notch filter as a fully-balanced "double-twin-T," "twin-H" and it works without having to scale any of the components.
Reality and the sim agree.

All that has to be done to switch the filter topology from Left/Right to balanced is to float where VR2/C3/C4 on both channels would common to ground and switching R7 from ground to the other channel's R7 to make the SE 40 dB gain stages a 40 dB INA.
It only takes two switch poles plus whatever switching is required to direct the output monitor to the proper A/D input.

To visualize the balanced version mirror the image below on its horizontal axis.

Image

I realize that there are now four interacting trims but what I found was that once both channels were nulled the null depth, when switched to balanced, really didn't change.

When the channels are switched to balanced C3+C4 Left in series with the RIght channel's C3+C4 is a very expensive way to make 4X 10 nF equal 10 nF but that's OK. They're already paid for and needed for L/R operation. :)
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by terkio »

All that has to be done to switch the filter topology from Left/Right to balanced is to float where VR2/C3/C4 on both channels would common to ground and switching R7 from ground to the other channel's R7 to make the SE 40 dB gain stages a 40 dB INA.
It only takes two switch poles
I think you need three switch poles.
Because for the differential configuration VR2 should be split from C3/ C4.
So:
A switch to float/ground VR2/VR2*
A switch to float/ground C3/C4/C3*/C4*
A switch to float/ground R7/R7*
Here, the * means: other channel
When the channels are switched to balanced C3+C4 Left in series with the RIght channel's C3+C4 is a very expensive way to make 4X 10 nF equal 10 nF....
Making a 10nF with four 10nF :D
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by mediatechnology »

Thinking outloud...

Aren't both the R and C nodes driven to 0V (in the center) with balanced inputs?
(Assuming ideal match.)
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by terkio »

Extremely brain warping.
When fully balanced, indeed both are at 0V so no current in between.
So it should make no difference to split or not to split. But I am not sure, I feel some pit fall here.

Assuming fully balanced inputs ( same level, same impedance ) and fully balanced output ( same load ) this must be true.

I think this must be though considering the DUT.
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by mediatechnology »

Yes its brain-warping.

I think the two nodes may be driven to 0V at resonance. (at reject Fc)
They may not have equal potentials in the pass band.

I'll do a sweep both ways and compare them but the third pole might be needed.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by terkio »

So far, with simulations I found no difference.
Virtual grounds connected or not, no difference detected.
The unbalances, I tried:
inputs amplitudes 0.25 0.75 ( iso 0.5 0.5 )
input resistance 1K 2K ( iso 1K 1K )
With or without differential output load 100K.

I then thought I should try with non ideal twin T that do not have exact C C 2C R R R/2.
There is no need to do this tedious work because a real twin T that has been trimmed for null output does have the same transfert fonction as the ideal one.
If not exacly the same, they are extremely close.
The real one has C1 C2 C3 R1 R2 R3 but once trimmed does have some R'C' strictly equal to the ideal RC.

I suspect the Twin H filter has exactly the same transfer function as the double Twin T filter.

I hope, my thinking will reach the conclusion that you were right with switching with 2 poles ( no need for a third pole ).
User avatar
mediatechnology
Posts: 5437
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by mediatechnology »

So far, with simulations I found no difference.
Virtual grounds connected or not, no difference detected.
The unbalances, I tried:
inputs amplitudes 0.25 0.75 ( iso 0.5 0.5 )
input resistance 1K 2K ( iso 1K 1K )
With or without differential output load 100K.
Interesting.
I found the same thing on the protoboard.
When you sim'd it did you do a sweep?

The point may become moot: I've decided to make it jumper selectable and used three jumpers.
I figure that duplicating the filter, one configured for dual channel SE, and another for balanced might actually be cheaper and work better than one loaded with relays.

I'm in the process of drawing the schematic for the dual channel/balanced version.
I'll post it here soon.

Looking at the schematic-in-progress something became obvious about why no shift in Fc occurs as the filter is changed from unbalanced to balanced.
I had thought it would move an octave but it doesn't.
The net capacitor values in the arms halve because they are in series.
The resistor values double for the same reason.
Their product remains constant.
Therefor Fc remains the same.

Looking at the schematic it's obvious : Not having a schematic and thinking about it is brain-warping.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: Super Low Distortion Ultra Pure Audio Oscillators Revisited

Post by terkio »

mediatechnology wrote: Thu Jan 14, 2021 3:21 pm
So far, with simulations I found no difference.
Virtual grounds connected or not, no difference detected.
The unbalances, I tried:
inputs amplitudes 0.25 0.75 ( iso 0.5 0.5 )
input resistance 1K 2K ( iso 1K 1K )
With or without differential output load 100K.
Interesting.
I found the same thing on the protoboard.
When you sim'd it did you do a sweep?
Yes on the range 10Hz 100 KHz
Doing a Spice "ac analysis".
With in+ AC 0.5. - in- AC 0.5.
The plot of V(out+) - V(out-) displays the Bode curves: Gain and Phase.
They all have the expected shapes with gain 0dB at 10Hz and 100KHz, gain about -90dB at Fc, phase jumping from -90° to +90°
Post Reply