Yes, I use also multi waves with optimized wave phase shifts (to get a low crest factor) for my FFR Test Bench (Fast Freq Response) as 4k..256k & to get the required resolution even for notch filters
Hp
I was sure this would work but still it is nice to see real measurements.
You have very precise balance, how much offset was needed?
Best wishes
David
Last edited:
The midpoint was approximately 5% from the half. The FETs were matched only for the similar Vgs off, but with unknown difference in the transconductance.
Thank you for the clarification.
Overall all are correct. First and second schematic is not balanced type, and the 1Mohm resistor at the gate isn't a big problem.
Both sides of the balanced schematic must have equal impedances. My oscillator has 600ohm output impedance - you can see this in the schematic. The second 600 ohm resistor imitates this impedance on the other side of the schematic (I have no 85,7ohm resistor
). I measured only voltages across the d-s of the FETs and the 100ohm resistors instead of the one FET in the two first schematics, and these voltages were app. 60mV rms.Also you can check this balanced schematic in a simulator for to see if it is correct.
Last edited:
Now, having finished the assembly of the enclosure for Victor's LDO during the weekend, I was able to conduct some measurements today.
I've added a BNC-output and included a passive twin-t notch filter and a post-notch amplifier with voltage gain of +40dB built around a OPA134 (my THT opamp stock is very limited). Everything crammed in a diecast aluminum hammond enclosure, supplied by external LM317 regulated linear PSU. PSU-cabling is unshielded (2mm banane type). All signal cabling via 50 Ohms coax.
The Notch has a depth of approx -47.6dB, the 2H is attenuated by -9.2dB and the 3H is attenuated by -5.3dB.
1) left channel of QA401
Osc. straight into to pos. Input:
With notch and post-amplifier:
2) For Reference the same setup with swapped channels. (see attachment)
I need a deeper notch... practically no harmonics visible. In other words, this seems to be the dist. limit of the QA401 only (1Vrms!).
I've added a BNC-output and included a passive twin-t notch filter and a post-notch amplifier with voltage gain of +40dB built around a OPA134 (my THT opamp stock is very limited). Everything crammed in a diecast aluminum hammond enclosure, supplied by external LM317 regulated linear PSU. PSU-cabling is unshielded (2mm banane type). All signal cabling via 50 Ohms coax.
The Notch has a depth of approx -47.6dB, the 2H is attenuated by -9.2dB and the 3H is attenuated by -5.3dB.
1) left channel of QA401
Osc. straight into to pos. Input:
With notch and post-amplifier:
2) For Reference the same setup with swapped channels. (see attachment)
I need a deeper notch... practically no harmonics visible
. In other words, this seems to be the dist. limit of the QA401 only (1Vrms!).Attachments
We maybe we can take the word of the group at Audio-Precision? They know a thing or two about measuring.
"A multitoned signal is much more realistic than a single sinewave, with both spectral and dynamic characteristics (crest factor) similar to music and voice."
-AP User's Manual.
I am going thru the test setup and getting familiar with Multitone and synchronous FFT concepts before actually doing some T&M.
Why doesn't someone try to design such DIY test gear -- a synchronous multitone generator and software to analyze it may already exist.
THx-RNMarsh
Last edited:
you need to read the words in the full context
not cherry pick phrases to present as 'sound bites' that mislead in isolation
AP doesn't say that those features of multitone test signals contribute to improved resolution of analog electronic device nonlinearities
FastTest, HST multitones are used for Fast Production Testing, not SOTA low distortion electronics development
Multitone Analysis with APx500 - Audio PrecisionAudio Precision
single, few tones allow higher signal levels, higher distortion product levels, better resolution in noise of the informative product frequencies
Using Multitones in Audio Test - Audio PrecisionAudio Precision
even generously reinterpreting the non sequitur
doesn't really pan out in the 'real world' with even some better prosumer PC soundcards where I keep pointing to mulitone plots by PMA, Scott, others that show no noise floor modulation at all
Manuals, App Notes, White Papers can be excellent sources of engineering info, they can assume their audience has a high level of relevant technical expertise, allowing them to drill down to detailed levels, present, summarize quickly things that may need a dozen IEEE papers to fully develop
superficial 'syntactic matching' by less expert readers searching for the afor mentioned 'sound bites' to justify preconceptions does seem to be a internet thing too
not cherry pick phrases to present as 'sound bites' that mislead in isolation
AP doesn't say that those features of multitone test signals contribute to improved resolution of analog electronic device nonlinearities
FastTest, HST multitones are used for Fast Production Testing, not SOTA low distortion electronics development
Multitone Analysis with APx500 - Audio PrecisionAudio Precision
single, few tones allow higher signal levels, higher distortion product levels, better resolution in noise of the informative product frequencies
Using Multitones in Audio Test - Audio PrecisionAudio Precision
even generously reinterpreting the non sequitur
doesn't really pan out in the 'real world' with even some better prosumer PC soundcards where I keep pointing to mulitone plots by PMA, Scott, others that show no noise floor modulation at all
Manuals, App Notes, White Papers can be excellent sources of engineering info, they can assume their audience has a high level of relevant technical expertise, allowing them to drill down to detailed levels, present, summarize quickly things that may need a dozen IEEE papers to fully develop
superficial 'syntactic matching' by less expert readers searching for the afor mentioned 'sound bites' to justify preconceptions does seem to be a internet thing too
Last edited:
Dick, when I said 'Not that I agree with that reasoning...' I was referring to the reasoning that 'music is not sinewaves etc'. I do agree with all what AP says about multitones approaching music wrt signal complexity.
I should have been more clear.
Don't let us hold you back ;-)
Jan
@Black Chicken,
I'm confused as to your measurment.
are you trying to measure at -40dB? Then why does the
op amp have +40dB?
Or, are you trying to set the oscillator at 0dB, then set the
notch depth to -40db, to feed the opamp and see the result
at what gain?
@Richard,
can you do it with a few square waves?
I'm thinking that running 60 frequencies at once through
something wouldn't correlate to any music that I am aware of.
But, if you are hellbent on that process, what could correlate to that,
that you could actually use that won't sound like Scheisse?
Here might be some existing things that already exist:
Find some recordings of symphony orchestrae tuning up their instruments.
A. Ask someone who has a keyboard or piano to record the playing of the following:
B. Get a piece of wood the length of the white keys and strike them all at once.
C.Do the same for the black keys (can reuse the wood).
D, E Then redo B&C using the sustain pedal.
Someone should be able to accomplish that with a good digital piano
and make a high res recording of it and put to disc.
Those should take care of the large groups.
Then you can always find some 3-5 piece combo
and record them one evening at a jam. What you
want to find is the "trainwreck".
That is especially effective if in the same combo
that someone actually trys to sing a song.
For example, in the decade where I went to jams
there was one particularly crappy four piece who'd
get up to do their three songs. They did it for about
a year when I was discussing how bad they were and
how crappy their tone was. Finally I asked a fellow
guitarist about it and said "why the f* don't they just
do something simple that works like the T-Bone Shuffle?"
I continued, "that's pretty hard to f* up." I mean everyone
left the bar to drink on the porch, or for those that do, partoked.
The owner hated it because many folks wouldn't come back in
and drink more.
So, here is the unbelievable part of the matter. The guitarist
buddy told me then and there challenging me something awfuller,
"they are". I replied, "they are what?"
"Listen up" he replied, "that's what they're playing,
the 'T-Bone Shuffle'".
"Heilige Scheisse", I blurted after I heard the singer
say..."nothing that a T-Bone Shuffle won't cure."
"You mean they've been playing that for a whole
year now and it sounds that bad? I mean, it is so
bad, they screwed it up that much and I never
even recognized the song", I said in disbelief.
"Yeah," he said.
Okay, then I bought the next round and then we walked
out to the front porch to watch the hookers and the bikers
cruise by.
It really was that bad.
Cheers,
Post Script--
Now I see a few others replied while I post quick reply,
but rather lengthy at that. Hope I don't bore y'all with the detail.
I'm confused as to your measurment.
are you trying to measure at -40dB? Then why does the
op amp have +40dB?
Or, are you trying to set the oscillator at 0dB, then set the
notch depth to -40db, to feed the opamp and see the result
at what gain?
@Richard,
can you do it with a few square waves?
I'm thinking that running 60 frequencies at once through
something wouldn't correlate to any music that I am aware of.
But, if you are hellbent on that process, what could correlate to that,
that you could actually use that won't sound like Scheisse?
Here might be some existing things that already exist:
Find some recordings of symphony orchestrae tuning up their instruments.
A. Ask someone who has a keyboard or piano to record the playing of the following:
B. Get a piece of wood the length of the white keys and strike them all at once.
C.Do the same for the black keys (can reuse the wood).
D, E Then redo B&C using the sustain pedal.
Someone should be able to accomplish that with a good digital piano
and make a high res recording of it and put to disc.
Those should take care of the large groups.
Then you can always find some 3-5 piece combo
and record them one evening at a jam. What you
want to find is the "trainwreck".
That is especially effective if in the same combo
that someone actually trys to sing a song.
For example, in the decade where I went to jams
there was one particularly crappy four piece who'd
get up to do their three songs. They did it for about
a year when I was discussing how bad they were and
how crappy their tone was. Finally I asked a fellow
guitarist about it and said "why the f* don't they just
do something simple that works like the T-Bone Shuffle?"
I continued, "that's pretty hard to f* up." I mean everyone
left the bar to drink on the porch, or for those that do, partoked.
The owner hated it because many folks wouldn't come back in
and drink more.
So, here is the unbelievable part of the matter. The guitarist
buddy told me then and there challenging me something awfuller,
"they are". I replied, "they are what?"
"Listen up" he replied, "that's what they're playing,
the 'T-Bone Shuffle'".
"Heilige Scheisse", I blurted after I heard the singer
say..."nothing that a T-Bone Shuffle won't cure."
"You mean they've been playing that for a whole
year now and it sounds that bad? I mean, it is so
bad, they screwed it up that much and I never
even recognized the song", I said in disbelief.
"Yeah," he said.
Okay, then I bought the next round and then we walked
out to the front porch to watch the hookers and the bikers
cruise by.
It really was that bad.
Cheers,
Post Script--
Now I see a few others replied while I post quick reply,
but rather lengthy at that. Hope I don't bore y'all with the detail.
Last edited:
You might want to check out my AES paper "A Fully In-band Multitone Test for Transient Intermodulation Distortion", available on my web site cordellaudio.com. It uses 3 tones (triple-beat) to reflect odd and even-order distortion products down to around 1kHz.
Bob
Well, you make some assumptions, some true and some not true.
First of all, I don't care about what you think/know it does or does not show. I have my own use for the test. And, I have said what that test is for.
Noise modulation was brought up in a phone conversation with A-P tech staff. So, I will add it into what I AM looking for.
I think it is much better for what I am looking to find out.
THx-RNMarsh
I am not sure if I really understand your question but the setup is as follows:
LDO -> -47dB Notch @1kHz -> +40dB@0...20kHz Bandwidth OPAmp -> QA401
For sure the Q of the passive Notch is not high enough, and the depth is not sufficient for a valid characterisation of Victor's oscillator. But it is a start.
Or am I mistaken?
OK. But I don't want to measure anything very specific. I'm just looking for a large representative Total.... sort of like THD is a total number.. But including IM products and harmonics with a large number of tones. To see the total level/max.
More importantly, to Me will be what the entire system looks like... on record side and on thru play back side. Not of only one piece of gear in isolation. Rather all the distortion products produced going into another part of the chain and that total going into next gear and so on and so forth.
I suspect the total distortion level with be high enough that it will intrude into listening detection.
The same way with a single tone could be done also... and I have explored that. I suspect with Multitones, the TDL (Total Distortion Level) will let us look if it is audible level or not. Right now, we measure a single tone or two and each has distortion that is not audible in and of itself. Therefore, it is assumed, we cant hear any differences. But if Multitones and the accumulated distortion artifacts raise to a higher level, then we have a different story to tell.
THx-RNMarsh
Last edited:
Would you be willing to say what you see for a single tone, and under what conditions (say, power level, etc.)?
I ask because people often do a lot of listening at low levels. So distortion at various power levels would be of interest.
Also, there are also some people who seem to think that distortion (both linear and nonlinear) may be more noticeable when it is dynamic, changing with level, frequency, and whatever else there may be, perhaps time history for filters that may ring, or pre-ring, as the case may be.