"What's your reasoning?" and not "What's your belief?".

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
AX tech editor
Joined 2002
Paid Member
Re: Re: Hacking away at belief (cont.)

Elso Kwak said:

Hi Jan,
With reference to the article you posted:
Did you "evaluate" your partner blindfolded? Or is this the reason for the now popular "blind date" phenomenon among youngsters?
:clown:
Elso/too old for this nonsense

Elso,

ALL dates are, in a sense, blind. When you meet a nice girl/guy, you don't know her/his character, body (unless you meet on a nudist beach), intelligence, etc. You find that out AFTER you decide to go for it, isn't it?:D

Jan Didden

PS Your last statement contains two errors:
- you are NEVER to old for it;
- it IS NOT nonsense!

PPS You going to Veldhoven show this weekend?
 
Yes, that's true for thd, but not for imd. Don't forget that unlinearity
on multitonesignals creates harmonics BELOW originfreq !

Yes, that was exactly what I was saying. IMD of 14 kHz and 15 kHz will create fq components at 13 kHz, 16 kHz, 1 kHz, 2 kHz and so on which are within the passband whereas THD will not.

I think we need to work on our communication...:rolleyes: ;)

/Magnus
 
Swedish Chef said:


Yes, that was exactly what I was saying. IMD of 14 kHz and 15 kHz will create fq components at 13 kHz, 16 kHz, 1 kHz, 2 kHz and so on which are within the passband whereas THD will not.

I think we need to work on our communication...:rolleyes: ;)

/Magnus

Hmm, our communication is distorted ? :D
Okay, the mistake was (i think), i use thd to look at the unlinearity
itself. I'm convinced that IMD is a "result" of THD, if you just look at
the unlinearity. Means, if you have null THD over the whole audioband,
you can't have IMD.

Mike
 
Re: Re: Re: Hacking away at belief (cont.)

janneman said:


Elso,

ALL dates are, in a sense, blind. When you meet a nice girl/guy, you don't know her/his character, body (unless you meet on a nudist beach), intelligence, etc. You find that out AFTER you decide to go for it, isn't it?:D

Jan Didden

PS Your last statement contains two errors:
- you are NEVER to old for it;
- it IS NOT nonsense!

PPS You going to Veldhoven show this weekend?

Hi Jan, this may explain my preference for a nudist beach...
You have strange ideas how a relation develops.....
What is going to happen in Veldhoven of all places?

:clown:
 
Hmm, our communication is distorted ?

Probably just a slight phase shift... ;)

Means, if you have null THD over the whole audioband,
you can't have IMD.

Korrektomundo! :bullseye:
And what Mikek, Andy_C, myself and a few others here have tried to prove is that any nonlinearity whether being PIM, TIM, KIM or JIM and his dirty friends will show up as THD. Assuming the bandwidth is there and the guy turning the knobs on the meter knows the limits of the system and what he is doing.

Someone just have to make a cartoon out of Tim & Pim. Somehow they kind of remind me of those little Canadian characters Terrence & Phillip... ;)
 
Are KIM & JIM married ? ;)
I think, the point is, that THD is just a bad way to describe the
quality of an amp. But it might be possible that distortions don't
show up with a single sinewave, but produce harmonics with
multitone. Of course ANY nonlinearity shows up as THD. But how
evil a nonlinearity is, is much better measured with IMD.
That's what i showed with my checkings on musicsignals.

THD != THD

Unless you specify every single harmonic including it's phaseshift.

Is our communication in phase again ?

Mike
 
AX tech editor
Joined 2002
Paid Member
Re: Re: Re: Re: Hacking away at belief (cont.)

Elso Kwak said:
You have strange ideas how a relation develops.....

Do you think that may be the reason I can't seem to get it right...?

Elso Kwak said:
What is going to happen in Veldhoven of all places?
:clown:


?? VAD Hi-End show. If you want to press flesh with A-J van den Hul, Menno van der Veen, and janneman, etc ;)

http://www.vad-show.nl/home.html

I'll go Sunday with a couple of buddies.

Jan Didden
 
MikeB said:
Okay, i did my program applying distortions to a wave. (i was already
familiar with the basical mathematic for simple unlinear distortions)
The results are very interesting and explain why 3rd harmonic is EVIL !

I took a wavefile ripped from cd, and applied 4 different distortions,
all with about 0.6%thd for a single sinewave.
Here the results:

2nd harmonic only: THD: 0.62% -> 0.03%
3rd harmonic only: THD: 0.62% -> 2%
2nd+3rd harmonic: THD: 0.65% -> 0.3%
tanh distortion: THD: 0.6% -> 0.6%

for 2nd harmonic i use a simple y = x*(1-n) + x^2*n
for 2nd+3rd i use y = x*(1-n) + x^3*n
for 3rd only, too complex, but based on 2nd+3rd, 2nd cancelled out,
very identical to diffamp-behaviour.
And tanh speaks for itself...
All thd values were calculated with FFT. Of course these values vary with different music.

I think the first 2 numbers speak for themself, not surprising, but
explains some things !
Hi MikeB,

I'm confused too about the details of your experiment. I'll say what I think you did and then maybe you can correct that. It looks like you've implemented in the digital domain a nonlinearity with a reference distortion level of about 0.6%. I assume x above is the input word and y is the output word of the simulated nonlinearity. At what reference level was the 0.6% distortion taken? Full scale? Then you ran a WAV file through it and got a new WAV file or sequence. This WAV file was music I assume? Then you showed some distortion figures I'll repeat below
2nd harmonic only: THD: 0.62% -> 0.03%
3rd harmonic only: THD: 0.62% -> 2%
2nd+3rd harmonic: THD: 0.65% -> 0.3%
tanh distortion: THD: 0.6% -> 0.6%
I think the distortion on the left side of the -> is the reference distortion that you designed the nonlinearity for, correct? Then what is the distortion on the right side, and how was it computed?
 
hi andy !

Ok, i should have explained better. You got it all correct, the reference
THD is for full scale sinewave. The wavefile containing music has
been scaled to max volume.

The resulting distortions on the right side of "->" are the effective
harmonic distortion, a real TOTAL Harmonic Distortion, including
the TIM. (Of course for an amp with unlimited bandwidth)

I calculated these by doing a FFT on both waves, the original and
the distorted. Then i made the difference between the two
freqspectrums/harmonics and calced the ratio of difference and
total sum of all harmonics.
There's a slight issue in this algorithm, the distortions could have
changed slightly the overallvolume, so i need to compensate the
ouputlevels between both wavefiles. This might introduce a slight
error to the numbers. I will correct this and get new numbers, but
they shouldn't differ much. The phenomena with the 2nd harmonics
is too intense to be some error. I think they cancel out each other,
as they are 90° shifted by nature.

Mike
 
Post #434
quote:
Originally posted by lumanauw
Opamps have 3 pins. Like NE5532 have certain gain-bandwith product. We can make high bandwith Closed Loop with certain low gain. Is there any trick to change the internal OL bandwith from low bandwith OL to high OL bandwith using these 3 legs?


Correction: op amps have a *minimum* of 5 pins, and they *all* are inputs. See http://home.comcast.net/~walt-jung/wsb/html/view.cgi-showresources.html-TopRes-Walt-27s-20PDFs.html In particular, see the 4 audio op amp series. If you build a discrete op amp, you eliminate the thermal feedback input. You can also do this with a buffered output stage.

The power supply (PS) pins are also inputs. But few folks talk about that. Unfortunately, different op amps have differing sensitivities to PS noise, and swapping op amps using a poor PS can nullify any conclusions as to which one is "better".

The effective open loop (OL) bandwidth can also be manipulated. See above series again.

Walt Jung
 
discrete vs. IC+buffer link output stage

I have made quite enough comparison of discrete link output stages and that ones with IC + monolithic buffer (BUF634T, LT1010). The comparison were often made with renowned audio preamp products and I would be very careful to state that discrete design must be sonically "better".

I hope that serious designers like JC never speak about unbuffered OpAmp link output stage.
 
I have made quite enough comparison of discrete link output stages and that ones with IC + monolithic buffer (BUF634T, LT1010). The comparison were often made with renowned audio preamp products and I would be very careful to state that discrete design must be sonically "better".

So, this just reinforces my point. There is a whole lot more to making a good audio stage beyond whether it is IC, discrete, whatever, as errors get in from just about everywhere.

The previously cited references do illustrate how to extend the effective OL BW, even with the 5532. Please read, or re-read.

WJ
 
Walt,

I will re-read, though 5532 is not my favorite at all.
Could you tell me what do you think about AD8024?

Thanks,
Pavel Macura

Pavel:

The 5532 isn't my favorite either, among the reasons being the non-linear bipolar front end (subject of this thread). I like FET input amplifiers, which reduce the gm modulation, by virtue of the inherently better transfer curve. But one can, optionally, usefully compensate for a non-linear bipolar front end. An example was shown in: Jung, W., "Optimizing Op Amp Transfer Linearity," within Boak, J.: ‘The Headphone Connection’, The Audio Amateur, Issue 3/1982. This article also references the original Barrie Gilbert article on multiplying with bipolar transistors.

On the other point, the AD8024 looks very good on paper, but I don't have any experience with it at all. FYI, I no longer work for ADI. That said, I was/am a big fan of the gutsy current-feedback amplifier types, such as the (older) AD811, etc., and the voltage-feedback AD817, both of which deliver a clean 100mA output. Almost any amp that's rated for video applications can do better for audio, since to meet video specs, they have to be very linear.

Getting back to the subject of this thread, both of these amps minimize the input stage distortion, either by emitter-degeneration (for the AD817), or the transimpedance architecture that the AD811 uses, which is linearized by virtue of the external resistor.

Hope this is of some help.

Walt Jung
 
I agree with Walt Jung that 'thermal feedback' can be as important as open loop bandwidth.

Well John, I didn't exactly say just that, I said that there are a lot of things that are important! And certainly there are.

It is my gut feeling that linearizing the input stage, even when the open loop BW is *not* > audio BW, is still a big help. It will reduce the PIM in proportion to the non-linearity reduction.

In practice, a lot of folks like the FET input amps, for example the often cited OPA627 etc. My opinion is that at least part of this is due to the more linear front end. If you look at the OL BW of this amp, as well as other FET input types, they are still rather narrow, i.e., less than audio BW.. See OPA627 DS, p4.

So, why is it that so many prefer these FET amps?

Walt Jung
 
Hi WaltJ !

From my understanding, when OL-BW IS < audio-BW, the only chance
to reduce the TIM is to reduce openloopdistortion, means linearizing
inputstage and vas. But typically this reduces openloopgain, hence
increase openloop-BW ?

My experience with jFET-input is, they sound nicer/smoother,
but this experience is with discrete designs.
Of course it's always better if distortion does not happen at all
instead of linearizing afterwards via FB.

Mike
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.