Super Low Distortion Ultra Pure Audio Oscillators Revisited

[mediatechnology]

Since the point in time I committed to laying out ULDO and Nacho I think I've built about 17 boards for customers.
This week was no exception and I'm thankful for the orders and the break from doing the layout.
ULDO is basically finished and as you can see a lot of work needs to be done on Nacho.
I'm not sure the line of demarcation between ULDO and Nacho will be on the final layout - I drew it in for dramatic effect.


ULDO-Nacho Layout Work In Progress

ULDO's schematic looks like a giant hairball with the THAT340's transistors hanging off the op amp but it laid out pretty well.
I mirror-imaged one THAT340 which is also symmetrical around the 5532.
Those big honkin' radial electrolytics are the differential bypass for the 5532.
They serve not so much to bypass the 5532 but to isolate its supply currents, when driving balanced loads, from the ±15V rails.
The return current for single-ended loads is a "home run" all the way back to the DC 0V connector.

[mediatechnology]

I'm still triple and quadruple-checking the layout but it looks like I'll be submitting a small board order in the next day or so.







Looking forward to getting that Protoboard back...

[mediatechnology]

I've got a prototype run of boards on order so in a week or so we'll see if this dog hunts.

[mediatechnology]

ExpressPCB got boards ordered on a 5 day turn-around produced in 2 days.

Looks like its time to build one and see if it runs...


ULDO Nacho PC Board

[mediatechnology]

ULDO-Nacho checks out without any board errors.

Will post some distortion FFTs.
+20 dBu into 604Ω balanced is measuring below -141 dBc.
Found that the 49R9Ω output resistors add a slight amount of third harmonic likely due to signal-induced self-heating.
With 0Ω build-out third harmonic is even less.


ULDO-Nacho Assembled

[georgeonthenet]

I am familar with dbu/dbm and dbv, but dbc? Is that a typo?

This board assembled is very deja vu for me as it looks so 1990s industrial. I love that style of circuits. I have done some audio that way, however mine are mostly surface mount these days. Nice work Wayne.

As precision oscillators go, at the moment the best I have is my tweaked HP 339 ocillator that gets me down to 0.0004% consistiently and that handles most of my work thanks to digital scope averaging to get the noise out of the way.

[georgeonthenet]

OK, I did my homework on dbc. Interesting

[fredbloggstwo]

Hi Wayne,

Looks a very neat layout

When do you go into mass production

Regards

Mike

[mediatechnology]

Thanks George and Mike!
To answer your question George I have a couple of extra boards.

I've been running a bunch of tests and updating the schematics to match the PC board.


ULDO Ultra-Low Distortion Oscillator Schematic

(Right click to open on a larger window.)


"Nacho" Notch Filter Schematic for "ULDO," the Ultra-Low Distortion Oscillator

FFT of Nacho's Output

I ran 1 M point FFTs with a 5 sample average.
The Y-Axis scale includes a -40 dB offset to reflect the post-filter gain.


ULDO +20 dBu 604Ω Un-Balanced Load Notch Out Y Axis Offset -40 dB

The FFT does not factor in Nacho's insertion loss.
A spreadsheet compensates the data measured off screen.
Since the screen has the 40 dB offset the spreadsheets post filter gain is set to 0 dB.


ULDO +20 dBu 604 Un-Balanced Load THD

The measured THD is just under -141 dBc.

I didn't enter harmonics above 7th since their reduced level don't weight the final measurement much.

Other loads and configurations tell a very similar story.

10KΩ loads at levels from 0 to +20 dBu typically run -140 to -141.
604Ω balanced loads come in around -138.5 dBc.
With the build-outs shorted it's a little lower suggesting that either the build-outs or test load introduce a little non-linearity of their own.

[JR.]

Sweet,,, I know I couldn't hear that.

JR