Low-distortion Audio-range Oscillator - Page 10 - diyAudio
Go Back   Home > Forums > Design & Build > Equipment & Tools

Equipment & Tools From test equipment to hand tools

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 10th February 2012, 06:36 PM   #91
diyAudio Member
 
thaumaturge's Avatar
 
Join Date: Aug 2011
Location: Utah
Bob, he mentions the HP239, not the HP339. I think perhaps I mentioned the 339 (my mistake, I assumed...) . As you no doubt know the HP239 is a variation of the 204.
Doc
__________________
Ne timeas a facie mulierum ea ignorare
  Reply With Quote
Old 10th February 2012, 08:34 PM   #92
richiem is offline richiem  United States
diyAudio Member
 
richiem's Avatar
 
Join Date: Sep 2010
Location: Grapeview, WA
@Bob Cordell -- Hi Bob, thanks for the comments. I do think the HP 239/339 oscillators are very good, especially at mid-band and especially when the JFET feedback is tweaked. Unfortunately, the JFET feedback needs to be optimized for each frequency to get best 2nd H rejection -- which turns out to be true for your oscillator as well. This creates some problems...

I've compared the 239/339 to your oscillator, which I built, incorporating a JFET feedback trim, and to a KH 4500, which I picked up cheap off eBay and tweaked a bit (not the lowest THD when compared to a KH 4402B), and also to an LED/CdS controlled Wien-Bridge unit I built, similar to the Williams design. I have not had a Tek SG505 (phase-shift) to measure, nor the KH 4402B (bridged-t), both of which are considered to be very good, nor have I been able to evaluate an AP oscillator (SVF)/analyzer.

The goodness of the 239/339 design becomes apparent when you get rid of the fundamental and then use a spectrum analyzer to view the results. In my case I've recently used an EMU 0204 USB sound device to feed my PC running the ARTA analysis suite. The noise floor of the 0204 is pretty decent and the measurement BW, while not all I desire or need, is about 90kHz at 192kHz fs, albeit with a noise floor that rises significantly above 30kHz.

I fully agree that the higher frequency THD is not as good, but is still quite good compared to other units I've been able to test. I would love to have a hardware spectrum analyzer with a 1MHz BW and large dynamic range, preferably >90dB. But I'm doing what I can with the less expensive tools I have -- so far, so good.

My friend Davada has been looking at a novel control circuit, one which has the distinct advantage of not needing rectification of the control signal; I'll let him post about his work when he's ready.

As you so correctly point out, the trick is to not need a large dynamic range in the control element -- the less needed the better.

I now doubt that any active-filter auto-tuning THD analyzer, such as the HP 339, 8903, etc, has sufficiently low intrinsic THD to give a good picture of oscillator performance. But an Active Twin-T filter and a good spectrum analyzer can definitely do the job -- in my case, for mid-band audio frequency signals up to 10kHz or a bit more. Audio Precision seemed to think this filter-spectrum analyzer topology was best too.
  Reply With Quote
Old 10th February 2012, 09:12 PM   #93
diyAudio Member
 
Join Date: Sep 2006
Quote:
Originally Posted by richiem View Post
@Bob Cordell -- Hi Bob, thanks for the comments. I do think the HP 239/339 oscillators are very good, especially at mid-band and especially when the JFET feedback is tweaked. Unfortunately, the JFET feedback needs to be optimized for each frequency to get best 2nd H rejection -- which turns out to be true for your oscillator as well. This creates some problems...

I've compared the 239/339 to your oscillator, which I built, incorporating a JFET feedback trim, and to a KH 4500, which I picked up cheap off eBay and tweaked a bit (not the lowest THD when compared to a KH 4402B), and also to an LED/CdS controlled Wien-Bridge unit I built, similar to the Williams design. I have not had a Tek SG505 (phase-shift) to measure, nor the KH 4402B (bridged-t), both of which are considered to be very good, nor have I been able to evaluate an AP oscillator (SVF)/analyzer.

The goodness of the 239/339 design becomes apparent when you get rid of the fundamental and then use a spectrum analyzer to view the results. In my case I've recently used an EMU 0204 USB sound device to feed my PC running the ARTA analysis suite. The noise floor of the 0204 is pretty decent and the measurement BW, while not all I desire or need, is about 90kHz at 192kHz fs, albeit with a noise floor that rises significantly above 30kHz.

I fully agree that the higher frequency THD is not as good, but is still quite good compared to other units I've been able to test. I would love to have a hardware spectrum analyzer with a 1MHz BW and large dynamic range, preferably >90dB. But I'm doing what I can with the less expensive tools I have -- so far, so good.

My friend Davada has been looking at a novel control circuit, one which has the distinct advantage of not needing rectification of the control signal; I'll let him post about his work when he's ready.

As you so correctly point out, the trick is to not need a large dynamic range in the control element -- the less needed the better.

I now doubt that any active-filter auto-tuning THD analyzer, such as the HP 339, 8903, etc, has sufficiently low intrinsic THD to give a good picture of oscillator performance. But an Active Twin-T filter and a good spectrum analyzer can definitely do the job -- in my case, for mid-band audio frequency signals up to 10kHz or a bit more. Audio Precision seemed to think this filter-spectrum analyzer topology was best too.
Hi richiem,

We think a lot alike. I've been using active twin-T filters for over 30 years. I've always loved twin-T's. In fact, before I built my THD analyzer, I measured THD at 1 kHz and 20 kHz with a pair of frequency-dedicated twin T filters with feedback. When last year I did some tweaking of my THD analyzer, I also used twin-Ts to look at oscillator distortion. As you correctly point out, the analyzer section usually cannot be as good as the oscillator. That is definitely the case with my THD analyzer.

BTW, what is interesting, but which was hard for me to prove, is that phase noise in the system seems to set the floor in my now-tweaked analyzer (close-in noise sidebands about the fundamental). It is not THD or flat noise in the SVF notch circuitry of the analyzer. I used to think that ordinary circuit noise set my floor, but that appears not to be the case. These were very difficult things to sort out by measurement, so what I am saying may just be intelligent speculation. Actually, to be more precise, the SVF notch filter in the analyzer is ultimately formed by a subtraction circuit, and any phase jitter or other close-in spectral noise near the fundamental impairs the notch.

Looking with a spectrum analyzer at the residual from my THD analyzer when it is driven by its own oscillator, the signal energy is dominated by fundamental and close-in sidebands, not flat noise or harmonic spectral lines. Looking at the same residual on a capture with a DSO shows that the phase of what little there is of the fundamental is all over the place, so it is not just an imperfect static tuning null.

At one point I thought I had some 1/f noise in my tuning control loops, but I could not find it and or reduce it. Of course, even ordinary circuit noise can be magnified about the fundamental in an SVF.

I probably could have learned more about it by doing lots of experiments, some involving an active twin-T, but the whole thing began to become a time sink.

Life for me would not be as good without my trusty old HP3580A analog spectrum analyzer, although unfortunately it only goes to 50 kHz and only has 85dB dynamic range (that's why twint-Ts or my distortion magnifier in front of it are very helpful).

Interestingly, I found that substituting modern LM4562s for the 5534s in my analyzer did not reduce the floor. That was a bit of a surprize. For its age, at least for instrumentation, the old 5534 is remarkable in distortion performance.

BTW, maybe I read you wrong, but the Tek SG 505, designed by Bruce Hofer before he helped found AP, is definitely an SVF oscillator. There are many aspects of the AP1 that are very similar to my THD analyzer. Bruce Hofer, Rich Cabot and I became good friends while they were still at Tek.

Cheers,
Bob
  Reply With Quote
Old 10th February 2012, 10:00 PM   #94
richiem is offline richiem  United States
diyAudio Member
 
richiem's Avatar
 
Join Date: Sep 2010
Location: Grapeview, WA
Quote:
Originally Posted by Bob Cordell View Post
Hi richiem,
...
Interestingly, I found that substituting modern LM4562s for the 5534s in my analyzer did not reduce the floor. That was a bit of a surprize. For its age, at least for instrumentation, the old 5534 is remarkable in distortion performance.

BTW, maybe I read you wrong, but the Tek SG 505, designed by Bruce Hofer before he helped found AP, is definitely an SVF oscillator.
...
Cheers,
Bob
They are still among the best, and TI/BB seem not to have made them worse than the few old Signetics parts I still have. These days, I'm tending to use the LT1468 (or OPA134) unless the source Z is very low -- then the LME49710 looks pretty good.

You re right, the SG505 is an SV unit -- I'm not sure what unit I was thinking of.

It seems whatever filter is used, that everything else being equal, filter Q doesn't seem to matter in oscillator goodness -- rather it seems that overall maximum circuit gain (and GBW at higher freq's) is the determining factor -- and I'm sure that's no surprise to you. Someone here mentioned the need to control circuit gain vs frequency, and I'm guessing that controlling integrator Tau in the ALC is an important part of that.
  Reply With Quote
Old 11th February 2012, 10:48 AM   #95
PChi is offline PChi  United Kingdom
diyAudio Member
 
Join Date: Jan 2012
Location: Staines, UK
Hello Richiem,

Thanks for your inputs and for your offer of detailed measurements. I'm taking nothing for granted until I have managed to put something together. I'm not claiming that I will be able to achieve 0.001 % or less just that I have taken a (very) slightly different approach. Certainly the HP239A circuit is good.

Also thanks for Bob Cordell's inputs and for his book 'Designing Audio Power Amplifiers' which was most interesting and money well spent.
  Reply With Quote
Old 26th February 2012, 08:55 AM   #96
diyAudio Member
 
Join Date: Jun 2005
Location: Zürich
Quote:
I used to think that ordinary circuit noise set my floor, but that appears not to be the case.
Just as a side note: It is important to understand that the noise gain of the opamps in the oscillator becomes very high near the oscillation frequency. This results in a noise floor which looks similar to a jitter source with 1/f spectrum.

Quote:
At one point I thought I had some 1/f noise in my tuning control loops, but I could not find it and or reduce it.
The demodulators look like a prime suspect. Most auto-tuning notch filters I've analyzed used switching demodulators which might be better with this respect (but are sensitive to harmonics, so might not null as nicely without further precautions).

Samuel
  Reply With Quote
Old 13th March 2012, 05:15 PM   #97
PChi is offline PChi  United Kingdom
diyAudio Member
 
Join Date: Jan 2012
Location: Staines, UK
I thought that I would revive this thread because I now have a PCB up and running using the Texas Instruments / National Semiconductor LME49990 in a phase shift based circuit.
The sample and hold amplitude circuit works OK after I had some fun with the control loop. I didn't find it easy to get the loop to behave over the Oscillator frequency range of 10 Hz to 100 kHz.
The gain control works OK with the multiplier output attenuated by 100 which minimizes the noise contribution.
I am now at the distortion measurement stage. Performance is OK at low frequency 0.002 % at 100 Hz, 250 Hz and 1 kHz. But distortion rises rapidly when the frequency is increased above that, 0.003 % at 2.5 kHz. 0.01 % at 10 kHz and 0.04 % at 25 kHz.
It appears that the main problem is the Distortion Analyzer but it isn't easy to modify it the way I 'designed' and constructed it with capacitors soldered directly between the back of the rotary switches and the PCB relying on the case to hold them apart.
A nice readily available easy to build distortion analyzer design would be useful.

I haven't populated the PCB with the output attenuator and buffer because I am waiting until I am satisfied with the Oscillator.
I didn't choose a balanced output driver because I intend to use a bench power supply with an isolated output so the the output ground can float.
I have AC coupled the output with a physically massive 40 uF Polypropylene capacitor and am regretting the choice. The HP239A has a DC coupled output.
Has anyone any comments on what they would choose? AC coupled or DC? Single ended or differential?

I also went for a low impedance output using a National Semiconductor LME49600 buffer with LME49710 which I now think was excessive. The HP239A has a 600 ohm output impedance which I don't like but it does save on a buffer amplifier.
  Reply With Quote
Old 13th March 2012, 06:59 PM   #98
richiem is offline richiem  United States
diyAudio Member
 
richiem's Avatar
 
Join Date: Sep 2010
Location: Grapeview, WA
@PChi -- No need for AC coupling if the buffer has even fair DC stability -- *and* if the instrument or circuit being driven doesn't try to push DC back into the buffer. My opinion is that the 49600/49710 combo will raise THD -- others may have different opinions.

Balanced output is good if you work with professional audio gear, otherwise it's not needed or helpful. But I have found that being able to lift circuit ground off the chassis ground is very helpful. I have two oscillators that allow this -- an HP 239A and a K-H 4500, and it's been quite useful.

No conventional THD analyzer will show you the true distortion of the oscillator, which makes a low-noise active filter (like a Twin-T) and spectrum analyzer combination the best way to look at distortion. Of course hardware spectrum analyzers aren't cheap, but they are needed to see products of fundamentals above 20kHz. Software analyzers are cheap and effective up to 18kHz or so with a good audio ADC. I'm waiting for someone to do a DIY 16-bit ADC system that has a bandwidth of 500kHz (fs = 1MHz) and low noise here on the Forum... I'm not up to it myself.

As to the rapid rise in THD above a few kHz, this is almost surely due to the GBW limitation of the oscillator amp and the resulting loss of open-loop gain.
  Reply With Quote
Old 13th March 2012, 08:14 PM   #99
benb is offline benb  United States
diyAudio Member
 
Join Date: Apr 2010
Quote:
Originally Posted by kevinahcc20 View Post
Depends on what low distortion is...you can pick up a Heath IG18 (or the later IG5218) on eBay for under $50. Richiem on this forum has reported on upgrade to improve the performance substantially. Since the expensive parts are the case, multiple rotary selectors & decading components, this is a reasonable way to get there.
I have two IG18's from hamfests, they're barely operational, and I decided not to try to fool with them. I recall there was two articles from the first decade of Audio Amateur that describe improvements for lower distortion.
Quote:
Originally Posted by dirkwright View Post
I need to buy some 8 position SOIC to DIP adapters. I found a few in different places.
There was this group buy that was a great deal, but it looks dead at the moment, maybe you can post there and start it up again:
Group buy SO08 -> DIL08 adapter ADP0r0
Quote:
Originally Posted by dirkwright View Post
I have an idea for making a distortion analyzer. It's not an entirely new idea though. I'm still reading about different techniques for measuring distortion. My idea, which came to me out of the blue, is to use a differential amplifier with really high CMRR and extremely low distortion as a difference amplifier. The oscillator output goes to one input and the output from the device under test goes to the other input after going through an attenuator. The DUT has to be non-inverting for this to work, and the phase error has to be nill as well. Basically, if you feed a common mode signal to a differential amplifier, what's left at the output is the junk that's the difference.

The problem that I see with the typical notch filter type of distortion analyzer is that amplifiers under test don't just amplify and distort the fundamental frequency from the oscillator, they also amplifier and distort every harmonic that the oscillator produces, as well as the noise and IM products. So, the output of the DUT becomes a mess of harmonics, noise and IM products that you can't separate out with a notch filter.

Anyway, I'm still working with this idea and it's clear to me that in order for my device to work I need a very low distortion oscillator and a differential amp with crazy high CMRR and vanishingly low distortion. The LME49990 seems to fit the bill so far. CMRR is 137 dB and distortion is 0.00001%.
I've seen this before, in Jung's "Audio Op Amp" book, and it was mentioned for power amps somewhere in the "Sound Quality Vs. Measurements" thread (the circuit/technique has a name, but I forget). Instead of going through an op amp, the input and output of an inverting amp are resistively mixed, making the whole thing passive (except for amplification of the resulting distortion components). There's R-C additions to the resistors to compensate for phase shift in the amp as well.
  Reply With Quote
Old 13th March 2012, 08:33 PM   #100
richiem is offline richiem  United States
diyAudio Member
 
richiem's Avatar
 
Join Date: Sep 2010
Location: Grapeview, WA
Search for Bob Cordell's DMA (I hope I'm remembering that correctly) for a very good way of doing what Dirk is thinking about. Unfortunately, it doesn't work for measuring oscillators directly unless you have a VERY high performance oscillator to compare to (THD less than 100ppb?)...
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Radford Low Distortion Oscillator Series 2 audiomik Equipment & Tools 21 19th February 2014 10:46 AM
ultra-low distortion audio oscillator geekysuavo Analog Line Level 16 26th March 2013 03:04 PM
Low distortion oscillator? rjm Equipment & Tools 30 4th May 2011 10:45 PM
Can we improve this low distortion sine oscillator ? gaetan8888 Solid State 22 29th March 2009 12:30 PM
Simple, low distortion 1kHz oscillator jackinnj Solid State 4 6th October 2003 03:58 AM


New To Site? Need Help?

All times are GMT. The time now is 02:09 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2