Sometimes it seems there might be something called "Dark Oscillation".
It might be similar to "Dark Matter".
So this "Dark Oscillation" might be lurking in the background and it is long
and slow moving. Sometimes it shows it's head on the 2H sometimes the 3H.
Strange how it has the amplitudes seem about the same at times.
If it is indeed "Dark Oscillation" how do we attenuate it? or shunt it to ground?
With anti matter caps? or Bybee quantum filters? Maybe the anti ground?
Now that is has been postulated how do we measure it?
Sounds to me like we might need to find a theoretical physicist.
If not.
then a trip to San Jose to ask the Rosecrutians.
if not,
then a trip to the Winchester Mansion and see if
we can spook it in another direction so our measurements
and scopes or analyzers wont be subjected to the affects of
"Dark Oscillation".
Cheers,
Sync
It might be similar to "Dark Matter".
So this "Dark Oscillation" might be lurking in the background and it is long
and slow moving. Sometimes it shows it's head on the 2H sometimes the 3H.
Strange how it has the amplitudes seem about the same at times.
If it is indeed "Dark Oscillation" how do we attenuate it? or shunt it to ground?
With anti matter caps? or Bybee quantum filters? Maybe the anti ground?
Now that is has been postulated how do we measure it?
Sounds to me like we might need to find a theoretical physicist.
If not.
then a trip to San Jose to ask the Rosecrutians.
if not,
then a trip to the Winchester Mansion and see if
we can spook it in another direction so our measurements
and scopes or analyzers wont be subjected to the affects of
"Dark Oscillation".
Cheers,
Sync
You'll have to write the college for that.
The lesson appeared as a link to the pdf not a directory.
Bob can you put a twin T at the input to your analyzer or maybe use your DM?
No need, its a reference to Horowitz & Hill's 'The art of Electronics'
It just covers basic ways of stabilising simple Wien bridge oscillators using a lamp or diodes and a FET.
Hi Bob:
Just so it's clear to me, this is using your original oscillator design from the article with the published jfet and resistor values, right?
thanks,
mlloyd1
Just so it's clear to me, this is using your original oscillator design from the article with the published jfet and resistor values, right?
thanks,
mlloyd1
Did you know that Dick Moore built and measured your oscillator with a notch filter:
IG-18 #2, the BIG-18
@EssB, Thank you for that reference. I have their 2nd edition, perhaps
I can find it in there. That should get me the information that I need.
Cheers from this side of the pond.
I can find it in there. That should get me the information that I need.
Cheers from this side of the pond.
If I put a 1kHz twin T between the oscillator and the analyzer, the analyzer would have no fundamental to lock onto. Also, a passive twin-T would cause differing amounts of attenuation at different harmonics. Maybe an active twin-T with a gain of 10 and maybe de-tuned enough to have only 20dB of attenuation at 1kHz would make the analyzer 10X more sensitive to oscillator harmonics. This all assumes that the active twin-T with gain does not introduce any distortion not well less than that of the oscillator. A better approach would be to put in an active bandpass filter with extremely low distortion, so that the analyzer could be evaluated with a more perfect input signal.
Cheers,
Bob
Yes, that is essentially correct - the analyzer published circa 1981 in Audio magazine as a 3-part construction project. A couple of years ago I added some minor trim to the gate feedback ratio on the JFET VCRs (nominally 50%).
The whole design used NE5534AN op amps. The oscillator just used a full-wave rectifier followed by an integrator to feed the gate of the amplitude stabilizer JFET. The integrator path did include some speed-up circuitry that comes into play when the leveling error is large. No sample-hold or anything else fancy. The THD analyzer article and the circuit can be found on my web site.
Cheers,
Bob
Hi Pete,
Yes, I do recall it. He did a great write-up. The key was to put a notch filter in front of essentially a spectrum analyzer instead of a THD analyzer - the spectrum analyzer does not need a fundamental to lock onto. Quite awhile back I built passive twin-T filters at 1kHz and 20kHz to do some of that kind of stuff.
Cheers,
Bob
I don't recall where I read it but I thought either the notch depth was held at 40 or 60dB
or if it was deeper some of the fundamental was injected back in at a known level.
Here is Dick's Active Twin-T write-up and IIRC he did not have an issue with the notch
adding significant distortion:
Active Twin-T notch filter
Alright then take your Twin T and put it in the FB of a low distortion op amp. That will give you a very high Q band pass filter.
Excellent.
Eagerly waiting your further investigation.
(Yes, I still have my Cordell Analyer......) 🙂
Patrick
Thanks, Pete. Excellent piece by Dick.
BTW, in the past, I have always done the active twin-T differently. While Dick's uses positive feedback, mine used negative feedback to sharpen the notch.
In other words, I would take the buffered output of the passive twin-T and feed it back to a summer that drove the input of the passive twin-T. The amount of NFB would decrease the gain droop on either side of the notch approximately by the amount of the NFB. For example, at a point away from the notch where transmission was down 3 dB, 20dB of NFB around the twin-T would reduce the droop to about 0.3dB. Of course, the notch depth due to an imperfect passive twin-T element would also be reduced by 20dB.
I think the positive feedback in the other approach also sacrifices some notch depth, but I'm not sure how much. I'm not sure which approach is better overall, but I suspect that the positive feedback version is better. A good project for a couple of SPICE runs on a rainy day with a slightly imperfectly implemented and/or tuned twin-T.
Any thoughts?
Cheers,
Bob
BTW, I have sometimes used a trimmer in series with the shunt resistor of the twin-T to make very fine center frequency adjustments. If the change from the ideal value is not much, and the shift in center-frequency is not much, the depth of the notch is compromised very little.
A low-value 4-gang pot could also be used to trim all three of the twin-T resistors (two of the gangs in parallel for the shunt resistor) in a slightly more ideal way. Pot tracking might not be too much of an issue if the % change of the twin-T resistors is kept fairly small.
At 1kHz if everything is set up just right you shouldn't be able to put more than 2 to 4 mV into it. At that level the distortion is minimal.
I have some spice files around on it. if your interested I can put them up here.
I've not built one yet or even thought about the designs too much, have only followed
them on the surface since way back in Wireless World, so I don't have anything to add.
Dick reported such good performance that I'll probably just follow what he did.
He suggested going to spot frequencies, at first I thought it was not good enough but
now I'm thinking of 1,2,5 with a multiplier switch as being pretty good enough since
10KHz and 20KHz are available and there would probably be enough data to make a
good enough plot versus frequency. Could even go to 1,2,3,5,7,9 just to get a few
more points per decade. I'd include a fine trim as you also suggest.
Where does a state variable notch fall in all of this performance wise?
I think the Boonton uses an SV osc and notch, the schematics are here if you are interested. I like their use of FET switches and computer control, in case you are interested:
http://boonton.com/~/media/Boonton/Manuals and Software/1121 Instruction Manual.ashx
Hi Pete,
Thanks for that manual. I looked at the specs on distortion residual and it looks like it is spec'd at -80dB (0.01%) if I interpreted that correctly. Unless it does much, much better than spec'd, it is not that impressive. But the automation is nice. Looks like it came out in the late '90s.
An SV notch may not typically be as good as an active twin-T. First, it has several more op amps; second, it probably almost always needs to be auto-tuned, so you have the same type of distortions introduced as in an audio oscillator, those including distortion of the (JFET) agc element and distortion introduced by the agc level-detect/control circuit.
Cheers,
Bob
The designer is on the forum and simple mods get the analyzer to .0006% 1KHz THD:
http://www.diyaudio.com/forums/equi...distortion-analyzer-tweaks-9.html#post4642016
Mainly OP amp swaps, and a few other minor things.
http://www.diyaudio.com/forums/equi...distortion-analyzer-tweaks-9.html#post4642016
Mainly OP amp swaps, and a few other minor things.
Similar results from the filter topology used in the HP339A produces better than -100dB THD + N. . Only 2-3 opamp changes... updates ... are needed;
Just some care in making a stand-alone filter having low drift and stable null is needed. And, some flexibility in tuning.
THx-RNMarsh
Just some care in making a stand-alone filter having low drift and stable null is needed. And, some flexibility in tuning.
THx-RNMarsh