@SyncTronX -- as Demin mentioned, the passive twin-t filter wants to see a high impedance load on it's output -- the QA400 may be too low a load for the parts you've chosen for the filter, which is causing severe attenuation of the signal going thru the filter. Try using a DMM on AC to measure the input level and the output level of the filter -- most of them have a 1Mohm or higher input impedance that won't load the filter. I think the QA400 may have as little as 10kohm input impedance -- too low for a passive notch filter.
What would do (as a buffer) 1Meg in 1K out (impedance)? With an THD of > -160dB at 1st and > -170dB at 2nd harmonic?
Theoretically the signal going into the buffer after notch is very small. I'm not too concerned about distortion specs of the buffer. Any of the low distortion op amps will do as long as the input resistance is high. Jfet input op amps are the best for this application. The bipolar op amps tend to load the network too much. The TT is very sensitive to output loading.
@richiem,
I measured again.
Input to the notch, 6.58 VacRMS.
Output to the QA400 .518 VacRMS.
Changed.
Added 1 meg ohm input BNC
Added 1K ohm output BNC
Result
Input to the notch, 6.58 VacRMS.
Output to the QA400 .107VacRMS.
This doesn't make sense to me that it is not much different.
I lost 400 mV with the change and I don't see much difference
in the QA400.
Question should I move the 1 meg resistor to the output BNC?
I measured again.
Input to the notch, 6.58 VacRMS.
Output to the QA400 .518 VacRMS.
Changed.
Added 1 meg ohm input BNC
Added 1K ohm output BNC
Result
Input to the notch, 6.58 VacRMS.
Output to the QA400 .107VacRMS.
This doesn't make sense to me that it is not much different.
I lost 400 mV with the change and I don't see much difference
in the QA400.
Question should I move the 1 meg resistor to the output BNC?
series or parallel for the Rs?
what freq are you measuring? the notch freq?
how about NOT at the notch?
you should do a sweep, broadband noise, psuedo random, whatever the QA has to do a wide band characterization of the notch, you're going to need to know that anyway.
You want to know the f0, f2, f3, etc to be able to correct the figures you see on the FFT.
Alan
what freq are you measuring? the notch freq?
how about NOT at the notch?
you should do a sweep, broadband noise, psuedo random, whatever the QA has to do a wide band characterization of the notch, you're going to need to know that anyway.
You want to know the f0, f2, f3, etc to be able to correct the figures you see on the FFT.
Alan
SyncTronX,
If you have not done so already, read Dick's article about his Twin-T. There is great information in every sentence!
Active Twin-T notch filter
If you have not done so already, read Dick's article about his Twin-T. There is great information in every sentence!
Active Twin-T notch filter
The difference is a loss of 13 dB with the 1K resistor. The input Z of the QA400 is 10K. The 1M and 1K resistors can both go. They are not adding anything.
It suggests a source impedance of 10K in effect. Its probably different with frequency. The load on the notch should be at least 10X the internal impedance or 100K. With the 10K input of the QA400 you will affect the performance of the notch. It really needs a low noise hi-Z buffer.
Attached is an LT spice sim of a notch with 40 dB of gain based on some work by JensH. It should give an idea of what is possible.
Attachments
If the Q was lowered on the second one then only the first would have to fine tuned.
@Sky, thanks, yes I've read most of Dick Moores stuff a few times.
It doesn't hurt to keep reading until I understand it.
@Demian okay, then I guess I'm stuck between
.1v and 50mV input to the notch at the Oscillators
full output of the HP339a.
Not there for the active notch yet. When ready
I'll try to build it into the B&K1607 as y'all
noted and in that thread and somewhere burried
here.
Had I been thinking, I just would have put the 1Meg
resistor across the output on the banana jacks and
saved some time.
When i remove the 1 meg, it should be a simple process of
putting a 100K across the input.
When I soldered the 1k on the BNC output it didn't have much
affect. Why then does a simple 75 ohm termination BNC cleans
up everything compared to the 1 meg/1K ohm cmbination.
Unless I need to look at it again, just to check the termination
affect again. I've got 75 ohm and 600 ohm BNC terminations.
It doesn't hurt to keep reading until I understand it.
@Demian okay, then I guess I'm stuck between
.1v and 50mV input to the notch at the Oscillators
full output of the HP339a.
Not there for the active notch yet. When ready
I'll try to build it into the B&K1607 as y'all
noted and in that thread and somewhere burried
here.
Had I been thinking, I just would have put the 1Meg
resistor across the output on the banana jacks and
saved some time.
When i remove the 1 meg, it should be a simple process of
putting a 100K across the input.
When I soldered the 1k on the BNC output it didn't have much
affect. Why then does a simple 75 ohm termination BNC cleans
up everything compared to the 1 meg/1K ohm cmbination.
Unless I need to look at it again, just to check the termination
affect again. I've got 75 ohm and 600 ohm BNC terminations.
I was thinking of building a second TT notch out of different componants.
I'm not sure how to change the Q of it. Or could I just plug it into
the BK and run the BK to the QA400?
The Orange drops I had are big at .01 600V caps. Are the losses
within them? Or the Bourns 20K trimmers?
I got a pair of clarostat UK 20K 20+ turn trimmers, but I was
saving those for something else. I don't have any additional
Male to Chassis BNC connecters though. Just putting thoughts
out here.
I'm not sure how to change the Q of it. Or could I just plug it into
the BK and run the BK to the QA400?
The Orange drops I had are big at .01 600V caps. Are the losses
within them? Or the Bourns 20K trimmers?
I got a pair of clarostat UK 20K 20+ turn trimmers, but I was
saving those for something else. I don't have any additional
Male to Chassis BNC connecters though. Just putting thoughts
out here.
@SyncTronX -- I think you're confusing the RF spec'd impedance of connectors and cables (in the Megahertz) with actual impedances at audio frequencies. Do not parallel a resistor across any of the connectors, no "matching" is needed. Feed the oscillator into the filter and connect the output of the filter directly to the QA400 -- no resistors needed -- the filter is the load on the oscillator, the QA400 is the load on the filter.
Look at a frequency well above the notch frequency (10X higher) and see what your signal levels are. Then you can tune your notch at the filter's fundamental frequency to get minimum signal out of the filter at the notch frequency -- even 20dB of notch filtering (1/10th) of the fundamental will let the QA400 look at distortion products pretty accurately. 40dB (1/100th) is even better.
Then note that with a passive filter, the 2nd harmonic is reduced by about 9dB, so add 9dB to the number the QA400 shows to see the actual level. The third is attenuated about 5dB, so add that -- you can mostly ignore the rest for now.
Look at a frequency well above the notch frequency (10X higher) and see what your signal levels are. Then you can tune your notch at the filter's fundamental frequency to get minimum signal out of the filter at the notch frequency -- even 20dB of notch filtering (1/10th) of the fundamental will let the QA400 look at distortion products pretty accurately. 40dB (1/100th) is even better.
Then note that with a passive filter, the 2nd harmonic is reduced by about 9dB, so add 9dB to the number the QA400 shows to see the actual level. The third is attenuated about 5dB, so add that -- you can mostly ignore the rest for now.
That is essentially what the Shibasoku is. Here is a two stage buffered twin T. The fundamental is notched -40 dB in each stage and boosted 40 dB by the buffer. The composite is 80 dB of notch and 80 dB of boost of the harmonics. Minor challenge- getting precise enough parts to do this in the real world.
Attachments
@SyncTronX -- you can't change the Q of the passive T-T filter, only the notch depth by fine tuning. To change the Q requires using an active filter like the one I describe in my web page. So my question is, do you think you're not getting enough notch depth? That is, the filtered fundamental is not low enough? You can't -- and don't need -- to remove all of it, just enough to let the QA400 see the actual harmonics of the input signal instead of it's own harmonics.
Notch Depth & Talking Fingers
@richiem,
No, I dont thinkI'm getting enought notch debth and
I have too great insertion losses I think.
I can fine tune 2nd and 3rd harmonic from the rear two pots.
Not much but I can.
Then I can adjust the notch slightly using the front pot.
I can't only get down to -40 thru -60 dB when the signal
is very low, say 200uV to 500uV. I can't only get into
the miliVolt range by maxing out the HP339a, Using 10V
Oscillator scale.
Or
Is that what it's supposed to be? I guess I have to find
some preamp signals and run them through the HP339A
or direct to notch, than QA400.
I'm concerned with having too low a signal using the notch
filter. I've seen Demian running 500 - 800 to 900mVrms
but I can't get anywhere near that. Maybe I need some other
Oscillator, or is that what the HP339a limits are?
I don't have any experience to draw from. I've never done
this before, I don't know what I'm doing, for me it is all research
as I don't even know what to expect...however I do have minimimal
expectations based on what y'all have done here and on your
respective and others web sites.
So I don't know if I've reached the limits already of the HP339a
and the QA400 w/ and w/o notch.
Y'all have worked in and around this stuff for a while now.
You have a feel for what is right or wrong and can SWAG it
or have enough of a big picture to kind of know whats going on
and can reasonably know if something is right or wrong or strange
without having done it before.
Now imagine for a minute you are undertaking something
that is completely inconceiveable...didn't have a clue about,
you gather some data but don't exactly recognize it yet.
I'm at the Monkey See Monkey Do point.
I'm struggling to get beyond that and I don't know
if my reading are normal, abnormal, if so in what way
and what can I do?
I have to get to the point where I know if it is a common
cause that is part of the process. If so then I work the process.
Or
Is it some other cause that isn't part of the process? If so, then
working the process doesn't get me there. Something has to change
outside the process.
Statistically speaking, is it inside the control limits or outside
the control limits?
However, because I don't even know the process yet, it is unstable,
and I don't know that I don't know so I'm asking questions.
With more experience, you might say, that is a known issue
or the HP339a is within spec and do you understand that this is
pretty high end stuff? Or middle end stuff.
I don't know if what I have is working the way it should work
or it's okay and with mods you'll see some improvement, yoda,
yoda, yoda.
Ya know they didn't promise me a rose garden so I had to grow my own.
I've killed them off several times trying to "protect" them. Now I've
got a few that I'm starting again. Use caution with -ides for plants.
---trying to shut my fingers up. : )
@richiem,
No, I dont thinkI'm getting enought notch debth and
I have too great insertion losses I think.
I can fine tune 2nd and 3rd harmonic from the rear two pots.
Not much but I can.
Then I can adjust the notch slightly using the front pot.
I can't only get down to -40 thru -60 dB when the signal
is very low, say 200uV to 500uV. I can't only get into
the miliVolt range by maxing out the HP339a, Using 10V
Oscillator scale.
Or
Is that what it's supposed to be? I guess I have to find
some preamp signals and run them through the HP339A
or direct to notch, than QA400.
I'm concerned with having too low a signal using the notch
filter. I've seen Demian running 500 - 800 to 900mVrms
but I can't get anywhere near that. Maybe I need some other
Oscillator, or is that what the HP339a limits are?
I don't have any experience to draw from. I've never done
this before, I don't know what I'm doing, for me it is all research
as I don't even know what to expect...however I do have minimimal
expectations based on what y'all have done here and on your
respective and others web sites.
So I don't know if I've reached the limits already of the HP339a
and the QA400 w/ and w/o notch.
Y'all have worked in and around this stuff for a while now.
You have a feel for what is right or wrong and can SWAG it
or have enough of a big picture to kind of know whats going on
and can reasonably know if something is right or wrong or strange
without having done it before.
Now imagine for a minute you are undertaking something
that is completely inconceiveable...didn't have a clue about,
you gather some data but don't exactly recognize it yet.
I'm at the Monkey See Monkey Do point.
I'm struggling to get beyond that and I don't know
if my reading are normal, abnormal, if so in what way
and what can I do?
I have to get to the point where I know if it is a common
cause that is part of the process. If so then I work the process.
Or
Is it some other cause that isn't part of the process? If so, then
working the process doesn't get me there. Something has to change
outside the process.
Statistically speaking, is it inside the control limits or outside
the control limits?
However, because I don't even know the process yet, it is unstable,
and I don't know that I don't know so I'm asking questions.
With more experience, you might say, that is a known issue
or the HP339a is within spec and do you understand that this is
pretty high end stuff? Or middle end stuff.
I don't know if what I have is working the way it should work
or it's okay and with mods you'll see some improvement, yoda,
yoda, yoda.
Ya know they didn't promise me a rose garden so I had to grow my own.
I've killed them off several times trying to "protect" them. Now I've
got a few that I'm starting again. Use caution with -ides for plants.
---trying to shut my fingers up. : )
The shabasoku uses at least 3 TT. Anyway, using two cascaded to get -80 may have benefits.... less distortion, noise, parts tolerance and drift?
-RM
SyncTronX -- let's keep it simple -- you are putting the output of the 339A oscillator into the T-T, then looking at the spectrum with the QA400, yes? Ignore for now using the 339A analyzer function and just focus on the T-T and QA400.
First, Set the 339 osc. for the filter design center frequency -- maybe it's 1kHz? Let's just assume 1kHz.
Feed the 339 oscillator directly into the QA400 (watch the level -- use about 600mVRMS output from the 339A osc, monitor it with the 339A's meter) and see what the level is shown by the QA400. This is your reference level.
Now insert your filter. To see the insertion loss, set the osc. to 10kHz and read it's level on the QA400 -- it should be within a couple dB of the reference level a 1kHz. If it's much lower, then there's a problem with the filter.
If all is well, retune the osc. to 1kHz. With your filter adjustments and the fine tuning of the oscillator frequency, you can tune the fundamental notch by looking at the level of the 1kHz signal in the QA400 screen. Now you tune your filter to get the lowest level you can of the 1kHz "spike" with the QA400. You may be able to drop it 60dB, but probably not much more -- with luck and patience and a steady hand and very good parts, you may get 80dB of attenuation.
You are not trying to tune the 2nd and 3rd harmonic levels, just the 1kHz fundamental. Whatever the 2nd and 3rd levels are, that is the amount of distortion that your system has as its default performance, or baseline level. Now you can try testng an amplifier or pre-amp, etc and see how much the distortion products increase (or not). In this thread, we are trying to see just how low we can get the distortion products by playing with the oscillator design, the filters used, the analyzer capabilities, etc -- this is basically R&D, so much of this doesn't really pertain to actual testing of other equipment at the practical level of basic performance. Hope this helps.
First, Set the 339 osc. for the filter design center frequency -- maybe it's 1kHz? Let's just assume 1kHz.
Feed the 339 oscillator directly into the QA400 (watch the level -- use about 600mVRMS output from the 339A osc, monitor it with the 339A's meter) and see what the level is shown by the QA400. This is your reference level.
Now insert your filter. To see the insertion loss, set the osc. to 10kHz and read it's level on the QA400 -- it should be within a couple dB of the reference level a 1kHz. If it's much lower, then there's a problem with the filter.
If all is well, retune the osc. to 1kHz. With your filter adjustments and the fine tuning of the oscillator frequency, you can tune the fundamental notch by looking at the level of the 1kHz signal in the QA400 screen. Now you tune your filter to get the lowest level you can of the 1kHz "spike" with the QA400. You may be able to drop it 60dB, but probably not much more -- with luck and patience and a steady hand and very good parts, you may get 80dB of attenuation.
You are not trying to tune the 2nd and 3rd harmonic levels, just the 1kHz fundamental. Whatever the 2nd and 3rd levels are, that is the amount of distortion that your system has as its default performance, or baseline level. Now you can try testng an amplifier or pre-amp, etc and see how much the distortion products increase (or not). In this thread, we are trying to see just how low we can get the distortion products by playing with the oscillator design, the filters used, the analyzer capabilities, etc -- this is basically R&D, so much of this doesn't really pertain to actual testing of other equipment at the practical level of basic performance. Hope this helps.
Try this. Get a good Jfet op amp, set it up as a buffer and run it off 4 - 9V batteries.
Dick's TT uses the opa134 and it work fine. There other that will work but more expensive.
OPA827 is one.
What value of resistor and caps are you using?