How about the PicoScope 4262? Real 16 bit, 20 bit "enhanced", 5 MHz. And won't break the bank like the AP.
I've used the older Picoscope stuff (10 MHz, 8 bit) and found it very handy indeed. It's worth a look.
Low distortion oscillator tests measurement circuits - 2012-05-03 04:33:52 | Test & Measurement World
I was thinking of making an expresspcb layout for it in my copious free time. It does use readily available parts and the distortion spectrum shows the differences in parts used for the filters. I like the LDR for the gain adjust. However it would be slow to settle and needs to be used very carefully since LDR's are not very linear on this scale.
The one I use is a variation on this: http://www.shibasoku.com/download/avc/ag15c_e.pdf. They are rare so not really practical to find. Its about 10 dB better than the KH4400. And 4X the size so not too convenient.
I was thinking of making an expresspcb layout for it in my copious free time. It does use readily available parts and the distortion spectrum shows the differences in parts used for the filters. I like the LDR for the gain adjust. However it would be slow to settle and needs to be used very carefully since LDR's are not very linear on this scale.
The one I use is a variation on this: http://www.shibasoku.com/download/avc/ag15c_e.pdf. They are rare so not really practical to find. Its about 10 dB better than the KH4400. And 4X the size so not too convenient.
super low distortion oscillators aren't the only path to high resolution distortion measurement
I keep pointing to indirect multitone IMD testing as allowing relatively low quality oscillators - or DAC outputs to give very high resolution of DUT nonlinearity
basic premise - create different frequency test sines with separate osc or DAC - they only have each fundamental and its harmonic distortions (and possibly spurs)
then combine in a low distortion summing circuit, apply to amp/test circuit/device, look for IMD sum and difference distortion products in output
filters can knock down test signal amplitude, boost specific IMD frequencies - then even cheap PC soundcard ADC/DAC can manage 160 dB distortion measurement floors with averaging
I keep pointing to indirect multitone IMD testing as allowing relatively low quality oscillators - or DAC outputs to give very high resolution of DUT nonlinearity
basic premise - create different frequency test sines with separate osc or DAC - they only have each fundamental and its harmonic distortions (and possibly spurs)
then combine in a low distortion summing circuit, apply to amp/test circuit/device, look for IMD sum and difference distortion products in output
filters can knock down test signal amplitude, boost specific IMD frequencies - then even cheap PC soundcard ADC/DAC can manage 160 dB distortion measurement floors with averaging
Good point. Below is a HP 8903B output vs M-Audio soundcard output, both going into the unbalanced input of the soundcard. M-Audio is green, HP 8903B is orange. IIRC, this was 1V RMS.
The 8903B has a monitor out that is the signal after the notch filter, and it can be scaled. It's handy for seeing what's being measured as THD+n. I could post a graph of that on Monday, if anyone is interested.
Yes, I would be interested. Just hooked my monitor out up to my spectrum Analyzer. I would be interested in understanding how you scale the monitor output.
Ken
I'm with jcx -- a very interesting road to travel when the goal is superior performance of audio amps.
I've also pointed out my (ahem) improved build of Bob Cordell's SV oscillator in a Heath IG-18 with 1kHz THD (not THD+n, of course) much less than 1ppm (< -120dBu), and 10kHz THD only somewhat higher. These are pretty dang good results.
Simple magnetic shielding of the toroid power transformer has reduced all line-related noise peaks to less than -120dBu as well.
If you want to build this as a single frequency oscillator, you can get much higher performance (under 0.00003% or -130dBu) by tweaking the AGC JFET's operating point:
IG-18 #2, the BIG-18
I've also pointed out my (ahem) improved build of Bob Cordell's SV oscillator in a Heath IG-18 with 1kHz THD (not THD+n, of course) much less than 1ppm (< -120dBu), and 10kHz THD only somewhat higher. These are pretty dang good results.
Simple magnetic shielding of the toroid power transformer has reduced all line-related noise peaks to less than -120dBu as well.
If you want to build this as a single frequency oscillator, you can get much higher performance (under 0.00003% or -130dBu) by tweaking the AGC JFET's operating point:
IG-18 #2, the BIG-18
Below is a shot of the HP 8903B monitor out set at 0dB & 20dB Post Notch Gain.
The generator is set to 441Hz, 2V RMS which was adjusted to read -6dBFS on the spectrum analyzer.
Default Post Notch Gain seems to be set to Automatic, but you can change that. Just type in these number on the keypad and then hit the SPCL (special) button
The generator is set to 441Hz, 2V RMS which was adjusted to read -6dBFS on the spectrum analyzer.
Default Post Notch Gain seems to be set to Automatic, but you can change that. Just type in these number on the keypad and then hit the SPCL (special) button
- 3.0 Automatic
- 3.1 0dB gain
- 3.2 20dB gain
- 3.3 40dB gain
- 3.4 60dB gain
Attachments
Interesting discussion. I'm not up to speed on the whole FFT thing, but I have an AP System 1. Below is a graph of distortion vs oscillator level with a cable from output back to input. Balanced I/O was used with adapters to a 50 ohm BNC coax cable. Three different frequencies are shown. The red graph is with a 10X 50 ohm attenuator inserted in the cable. Note that measurements were made with a VGA monitor sitting on top of the AP system. I'd be interested to see what newer AP systems look like.
Attachments
Makes sense. The lower gain will leave the post notch amps noise floor as the limit. The ranging may not be able to make sense of that (or why have the extra gain?) With the extra gain the noise and distortion will be above the noise floor and the calculations will make sense.
@ Loudthud -- well, just as you might expect, lowering the signal 20dB with the attenuator raises the level vs THD plot by 20dB. So the red curve isn't real helpful. I'm not familiar with the specs on the Sys 1 -- is it 50 ohm output Z? If it's not, I would avoid using 50 ohm attenuators or terminated cables....
Loudthud,
Try plotting the power spectrum with an input voltage of 1 Vrms at 1kHz over a range of 0 to 10kHz. It will show the immunity of the system to low frequency noise from a typical power supply. Then try a plot of 1 Vrms at 10kHz over a frequency of 0 to 100kHz. This will show immunity to switching power supplies used in computer monitors, computers, fluorscent lights, etc.
Ken
I tried it out myself last night, used the -40 setting on the 8903 and used the math function on the spectrum analyzer, just divided the spectrum by some power of 10 (I forgot which power
) of a known spectrum until I got a good match on the 2nd H from my other methodology. Worked well at 10kHz fundamental, but had a lot of noise at 1kHz fundamental. Thanks for the tip!
Ken
The AP system one does not have an internal FFT. However connecting one to the monitor output works well. We used a Picoscope 8 bit unit and it worked very well the Picoscope software is as good as any digital scope.
"...The AP system one does not have an internal FFT..."
-------------------------------------------------------
What a System One has depends on the options it was ordered with.
The SYS-22 units are analog only with no FFT.
The SYS-222 units [System One + DSP] add generation and analysis of multitone signals, FFT spectrum analysis, and MLS quasi-anechoic acoustic testing.
SYS-322 units [Dual Domain] also have AES/EBU balanced and unbalanced digital inputs/outputs as well as consumer coaxial and optical I/O for digital domain generation and analysis.
In addition there are other useful options for generation of tone burst, square wave and noise signals, A-weighting, de-emphasis & low pass filters, wow & flutter analysis, and IMD.
If considering one of these units be informed!
-------------------------------------------------------
What a System One has depends on the options it was ordered with.
The SYS-22 units are analog only with no FFT.
The SYS-222 units [System One + DSP] add generation and analysis of multitone signals, FFT spectrum analysis, and MLS quasi-anechoic acoustic testing.
SYS-322 units [Dual Domain] also have AES/EBU balanced and unbalanced digital inputs/outputs as well as consumer coaxial and optical I/O for digital domain generation and analysis.
In addition there are other useful options for generation of tone burst, square wave and noise signals, A-weighting, de-emphasis & low pass filters, wow & flutter analysis, and IMD.
If considering one of these units be informed!
Low distortion oscillator board project
I was doing some research on the project, particularly on the AGC circuits and got sidetracked by my Fluke 510A. Here are some notes on my efforts: http://www.diyaudio.com/forums/blogs/1audio/904-upgrade-fluke-510a-ac-voltage-reference.html I got it down to .0003% THD+N in a 30 KHz band.
In short, I thought I would look at the options and alternatives before I layout the one from the article. There are three typical oscillators circuits used, two wein bridge forms, one with a single amp and another with two (both inverting to make them more linear) and the state variable oscillator. There are several AGC control variations in use for this, the FET (used by Fluke above and KH in the 4400) the LDR in the article linked above, using an analog multiplier, and finally using a variable gain amplifier. Each has pluses and minuses. I wanted to do more research before committing.
Now that the Fluke distraction is back on the shelf (until I can get a cal done on my voltage reference) I'll get back to the project as time permits.
I was doing some research on the project, particularly on the AGC circuits and got sidetracked by my Fluke 510A. Here are some notes on my efforts: http://www.diyaudio.com/forums/blogs/1audio/904-upgrade-fluke-510a-ac-voltage-reference.html I got it down to .0003% THD+N in a 30 KHz band.
In short, I thought I would look at the options and alternatives before I layout the one from the article. There are three typical oscillators circuits used, two wein bridge forms, one with a single amp and another with two (both inverting to make them more linear) and the state variable oscillator. There are several AGC control variations in use for this, the FET (used by Fluke above and KH in the 4400) the LDR in the article linked above, using an analog multiplier, and finally using a variable gain amplifier. Each has pluses and minuses. I wanted to do more research before committing.
Now that the Fluke distraction is back on the shelf (until I can get a cal done on my voltage reference) I'll get back to the project as time permits.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.