Low-distortion Audio-range Oscillator

AX tech editor
Joined 2002
Paid Member
My oscillator is a 3kHz unit with the OPA1656. For some reason I have noticed (before doing this last measurement) that it appears to be a trifle cleaner than my 1kHz oscillator. I don't think the 1kHz has the OPA1656 so that may be a factor. And I used a twin-tee at the input, did any of you use that too?

But I also look very intently to the unit when measuring, maybe I scared it ;-)

Jan
 
Last edited:
AX tech editor
Joined 2002
Paid Member
Ja, I think you should use the latest version, beta 50. I use that and it is stable. You might also want to use 64bit FFT.

I did this VictorRev5-1kHz and your notch measurement in REW yesterday. You can see my settings in the image. 48kHz. 24bit. Tracker Pre.
Calibration file for notch used.

Sigurd, question please. I see you used the dBc scale, which I think makes sense. What I don't know is how REW determines what the 0dBc level is. The level coming into the RTX cannot be used because that has been squashed by the twin-tee. I have been looking for some sort of setting but didn't find anything.

Jan
 
I use your balanced Twin T with f = 1002,8 Hz and Victor´s version 5 sine wave generator with F = 1000,7 Hz.

My Victor´s generator is not in a shielded box but the notch is.

My oscillator is a 3kHz unit with the OPA1656. For some reason I have noticed (before doing this last measurement) that it appears to be a trifle cleaner than my 1kHz oscillator. I don't think the 1kHz has the OPA1656 so that may be a factor. And I used a twin-tee at the input, did any of you use that too?

But I also look very intently to the unit when measuring, maybe I scared it ;-)

Jan
 
Yes, the new 555B has a much more capable software (and a few dB less distortion) but starts at $ 30k. And in its wisdom AP now decided that with the new machines you need to pay a yearly software license between $ 1k and $ 2k depending on options ...

Unobtanium for us mere mortals.

Jan, I'm considering buying an AP off the secondhand market. Can you suggest which model is best for our purposes in diy audio? I find the range of options confusing.
 
AX tech editor
Joined 2002
Paid Member
I just saw the ticket for you: an AP 2712. That is identical to the 2722 except for the digital I/O, which you probably don't need.

This is the link:

Audio Precision sys-2712 DSP leistungsstarke Audio Analyzer getestet TOP | eBay

It says it has ended but he still has it. You'll need an AP APIB to USB adapter setting you back another $ 300. But I think he is willing to negotiate as he couldn't sell it at the asking price.

Worth a look.

Jan
 
Member
Joined 2004
Paid Member
I think it depends on what you need it for. An AP system 1 can be had for less than $1K, sometimes significantly less. However if you need to test SOTA ADC's and DAC's in volume you may need an APx555. For normal R&D a sys 1 or any of a number of other solutions should be more than adequate with additions like Victors oscillators and some twin T filters. If you are developing new products for a variety of markets (Bluetooth, HDMI etc.) you may need the premium products but even there you can find work arounds.

The portables seems to be something of a dead end from watching the market.

If you were in the US I have 2 Sys 1's I would like to "rehome".
 
It's how many secs of samples at the given samplerate must be accumulated to reach the FFT length. The actual FFT calculation takes millisecs with libfftw.

I didn't notice, I thought you were simply processing from the file in memory. I also found the three missing samples in Jan's file, but remain a little confused by the claimed 48k sampling rate of the original since there would be no 9th harmonic in the data.
 
I think it depends on what you need it for. An AP system 1 can be had for less than $1K, sometimes significantly less. However if you need to test SOTA ADC's and DAC's in volume you may need an APx555. For normal R&D a sys 1 or any of a number of other solutions should be more than adequate with additions like Victors oscillators and some twin T filters. If you are developing new products for a variety of markets (Bluetooth, HDMI etc.) you may need the premium products but even there you can find work arounds.

I'm just a hobbyist so production testing isn't an issue. Being retired I have no plans to start a new business and audio was never my main line of income, despite doing some serious development work on HDCD and electrostatics.

The portables seems to be something of a dead end from watching the market.

Are they just limited in functionality? I have never used any sort of AP unit and to tell you the truth having to read the manual for every model seems a bit daunting just for a basic comparison of features.

If you were in the US I have 2 Sys 1's I would like to "rehome".

Whatever I buy it will have to come from overseas. There are none in Australia for sale.
 

When I was shopping for an AP audio analyzer, I asked AP which of their products would work with their APIB-USB adapter. Here's their response:

System Twos that won’t work with USB interface:
2022
2222
2322 (48k)
2300 (48k)
System Twos that will work with USB interface:
Cascade:
2422
2522 (96k)
2500 (96k)
Cascade Plus:
2122
2622
2722 (96k)
2700 (96k)
2700 Series:
2702
2712
2722-192k
2720-192k

So basically, get a 2400 or above. Or an APx-series. The APx is newer and they all have USB built in.

Another thing to consider is the control software. I was never a fan of the SYS-2700 software. The new APx software, by contrast, is a true joy to use (and I don't say that often about software!)

AP has noticed that the various external sound cards are catching up in performance - at least at select operating points. They now offer their APx software for use with external sound cards via the ASIO driver. A license is ~$5k.

Tom
 
AX tech editor
Joined 2002
Paid Member
- if you feed osc with battery power, does it have the same amount of low frequency (150, 250, 350 Hz etc) noise?

It is on battery power. I am pretty sure those mains spikes are from the RTX, they typically look like spurs from rectifying process, what with the higher harmonics.
A simple mains incursion would be 50Hz, maybe also 100Hz, and not much more.

Jan
 
Looks a bit dirty. Here is my usual Asus Xonar Essence RTX (on PC, powered from ATX) for comparison (input of cable is shorted):
 

Attachments

  • Noise_InputShorted_Scale200dB.png
    Noise_InputShorted_Scale200dB.png
    20.6 KB · Views: 322
From the REW help file:
"A dBc Y axis option is offered which places the peak level of the input at 0 dBc, or places the peak of the fundamental at 0 dBc when the distortion panel is active."

Clear as crystal?


Sigurd, question please. I see you used the dBc scale, which I think makes sense. What I don't know is how REW determines what the 0dBc level is. The level coming into the RTX cannot be used because that has been squashed by the twin-tee. I have been looking for some sort of setting but didn't find anything.

Jan
 
When I was shopping for an AP audio analyzer, I asked AP which of their products would work with their APIB-USB adapter. Here's their response:

...

So basically, get a 2400 or above. Or an APx-series. The APx is newer and they all have USB built in.

I have no idea why USB would be important. Remember, I've never used one of these things.

Another thing to consider is the control software. I was never a fan of the SYS-2700 software. The new APx software, by contrast, is a true joy to use (and I don't say that often about software!)

AP has noticed that the various external sound cards are catching up in performance - at least at select operating points. They now offer their APx software for use with external sound cards via the ASIO driver. A license is ~$5k

Outrageous. That's completely out of the question for a hobbyist, and would likely scare off most small companies as well.
 
Yes but as far as I know the data on the distortion panel is either shown in dB or %. And that dB is not to the same reference as the dBc Y-scale.
But I can be wrong, only started on REW yesterday.

Jan

Distortion dB is relative to the fundamental. So the distortion is just the level of fundamental (whatever db type it is at the top of the distortion box) + the listed distortion dB.