DigitalSignalProcessing -> future? - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves

Tubes / Valves All about our sweet vacuum tubes :) Threads about Musical Instrument Amps of all kinds should be in the Instruments & Amps forum

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 6th June 2004, 12:37 AM   #1
diyAudio Member
 
Join Date: Jun 2004
Location: Cluj-Napoca
Default DigitalSignalProcessing -> future?

Many of you all tube addicted folks, I don't want to be missunderstud from the begining, but, I wonder if u're "tube addicted" or "tube's sound like addicted". If u answer this question with the second answer, that means u can stick to this thread for a while.
As many of you already now we're in 21 century and the technology is now more than ever "at full power". So I heard about DigitalSignalProcessing (DSP). For ones who never heard of it maybe you'l find this usefull. With DSP u can do almost everything whit a sound signal - compress, expand, filter, boost, delay, almost everything that it can happen with the sound in all the diffrent kinds of environements. One of these environments it's the vaccum tube. The vaccum tube has its particular effect on sound signals (and not only). Recentlly I managed to make a DSP plugin on my computer, after stuying some data algorithms about the "tube effect". I have an old "Sound Blaster Live" equiped with KXproject drivers ( www.kxproject.com ). The DSP plugin that I was cappable to make sound verry warmthfull. It basiclly contains 2 ways of processing the sound. One is applyed for frequencyes bellow 100Hz, and the other above 100Hz. The lower frquency branch contains a Dynamics processor (with particular settings of fast attack @5ms, 500ms of release time) and the other branch has too its dynamics processor with higher time attack @30ms, loger release time @3s, and lower threshold @-30dB. Next on this branch is a so called "tube"effect simulation who ditorts the high frequencyes (>2000Hz) giving some excellent "wet" high freq. Next follows a summ device for getting back together the two ways. Afterwars is a 10 band EQ, again with a particular curve type (16k very higt (+12-+16dB).
I can tell you when I look at those monster amplifyers, an huge prices for them, I listen a bit, and I'm going forward (not back) to my DSP. I really worked hard to make it function as it is now, but it sounds very very much tube like. So you never mind about the power amplificator. It can be one Si based but be carefull, 'cause it has to be a very linear one (the signal has already been processed and all it has to do it's amplifing it).


ps:I have no tubes anywere in my acoustical chain, the final amplifier its an silicon based one.
ps1:Long storry huh? I bet 60% of u din't read all of it, but youl'd better. For those of u I promise minimal rows in next posts.
  Reply With Quote
Old 6th June 2004, 11:53 AM   #2
benny is offline benny  Australia
diyAudio Member
 
benny's Avatar
 
Join Date: May 2004
Location: Melbourne, Australia
Send a message via MSN to benny
ok, now i hate to sound really skeptical, but no, you can't really model a tube amp using digital. i mean, in all fairness, you could.. but you would need to spend huge sums of money, and employ many people to research for many years to make an acurate model... and even then you would still be missing out on the tube sound, no matter how good your models are... i will come back to this point later. you need to remember there are literally HUNDREDS of factors in the tube sound... take your digital model into a hifi store, and if you ask nicely enough, they will probably let you compare it to some of their tube amps... you will see... play yours through a set of speakers, using the best SS power amp you can find, then listen to a tube amp throught the same speakers... do a blind test, and i bet that if your have sensitive enough ears you will hear quite a difference. which one's better is of course subjective. although you might be able to guess what way my subjectivity makes me lean towards, i won't base this on opinion... so here is a couple of facts.

just to list a few points, focusing more on the power amp and it's distorion...

Quote:
Next on this branch is a so called "tube"effect simulation who ditorts the high frequencyes (>2000Hz) giving some excellent "wet" high freq.
firstly, what kind of distoriton is this, and what is being distorted more... if it is just straight "i'm gunna make this thing clip above a certain threshold" type, then it is no where near tube like distortion.

for a very start, you must understand that tube distortion is different in nature to just straight clipping of the wave form.. it is a softer type of clipping that rolls into distortion instead of cutting straight in, and there are many things in a tube amp that cause distortion, and all play a big part in the sound.

firstly, what configuration (SE, PP) are you simulating?
both of these configurations give very different distorion spectrums. you might be able to say "right, i want this amp to have a THD of say 0.001%. THD = [(H2)^2 + (H3)^2 + (H4)^2 + ............ )^1/2... you can have a THD value, but as you can see, a THD is the sum of all the harmonic distorions.. and you could increase say seccond harmonic distoriton, and decrease third order, and still have the same THD, but both sound different. Also, remember THD is also the total distortion of ALL the components in the system, and each component distorts differently and by different amounts depending on many things... to look solely at one point in the power amp distortion... in a PP configuration, you get more odd order distortion due to cancelation of the even order when the two inverse waves are summed together... and PP distortion has more higher order harmonic distortion content that SE. this all is due to SE output has no cancelation of even order distortion, and also accentuates it even more due to asymetrical clipping, and it clips lower order more... this has to do with the nature of tube distortion being softer, and not so imediate, so it compresses before it distorts.

also, take into consideration recovery time... put very simply, when a tube clips, it causes a sag which therefore reduces clipping threshold. the signal must drop below this already lowered threshold long enough for it to recover back to the original... this all happens very quickly, probably too quick for a computer to model, although i'm not sure as i'm not really knowledgable in computers. so it's a very variable amount of distortion too.

and even if you considered configuration, what class of power amp are you modeling?? (A, AB, B... there are also further divisions within this, like AB1, AB2, etc used to describe where it's sitting, as there is no set 'this point gives class a... rather refer to it as something like high class AB or whatever) ? did you take into account crossover distortion, which is more and more as the bias gets lower.

now, crosover distortion only effects PP amps running any lower than absolutey in class A, which is pretty unlikely. so you think, ok, it's SE im medeling, so i'm safe on this last bit... wrong!

let's think what comes after the power amp... a transformer... and this also adds to distortion, or if you like, saturation. also, SE and PP trannys are different, and saturate differently... PP tranys have no air gap as they rely on having ballanced DC's in their primary canceling out each other, and SE have an air gap because of one way DC. also, magnetic flux will only ever be one way at any given time in a SE tranny, whereas in a PP there is flux in two direction at any time... just one point to consider. i am not sure how these effect distoriton charachteristics, but for a start to get you thinking, the SE tranny generally will saturate less proportionately because of the air gap.

we've gone this far... let's say you made a perfect model of a tube amp... so in theory, your model is exactly the same... so to go back to my point earlier in the start i said no matter how perfect your amp model, you will never ever, no matter how advanced, be able to model tube sound... so what could be left??? oh yeah, speakers for a start! guess what, tube amps are more reactive to the speakers than SS, and their variable impedance reflected back changes distortion and distortion spectum too... depending on what impedance an amp is driving, it will distort different amounts.. as speakers are a reactive load, they will load the power amp differently at different frequencies... this has many implications, and we could go all the way through all the previous points discussing them with respect to speaker reactance... but for now as this post is starting to get kind of long, despite my efforts to keep it brief, i will cut this off... moving onto volume... this also effects the system... everything from acutal tube distoriton to speaker reactance, tranny saturation etc... can you turn up or down your models volume???

so, even if you could model that, you are still left with the problem of amplifying it from a digital signal to an audible signal... remember THD is the sum of all the components in your chain... well you model is more like the preamp, so moving out of the preamp we need a DAC, and now we need to amplify the analog signal, that has just been coloured by the DAC, and had a bit of distortion added... and guess what, no matter how linear you make the real life amp, you are still altering sound when the power amp reacts with the speaker...

talking about the speakers and how they work conected to a valve amp further, one further point i want to bring up with respect to the power amp config (SE/PP) i should have bought up earlier... a speaker also distorts, and colours your sound. and because the speaker works based on magnetic induction, and when we induce an EMF in a magnet, it will be out of phase with the source field. what happens, is speaker distortion also tends to be lower order, and more even order... now, because the speaker distortion, related to movement is out of phase with the force driving it, you get different cancelations/additions of distortion dependant on the makeup of the distortion coming out of the amp... PP alows the speaker to distort more... SE cancels out some of the speaker distortion...

now, that was only me being brief, and only covering a few of the points with restpect to power amp distortion... now, dont forget power amp frequency response, dynamic range (yeah, distortio realtes to this, but it is more than just distoriton), etc... and then lets look at earlier sections, you preamp...

now, if i can crap on for this much over only some of the many points, trying to be brief and non technical on all of them, imagine how much you have to consider... and there is already, in what i said, enough to keep you ocupied for a long time.

i hope that this shows that simply aplying a global distortion/compression and equalization between a couple of frequencie points is far too simple...

Quote:
I really worked hard to make it function as it is now
now if it was hard for you to do what you've done so far, this might give some insight into the imposibility of the taks of creating a true digital model.

just remember, i am not discrediting your model, and i'm sure it does make the sound much nicer... but it's not necesarily a tube amp sound... no way

cheers
  Reply With Quote
Old 6th June 2004, 12:32 PM   #3
benny is offline benny  Australia
diyAudio Member
 
benny's Avatar
 
Join Date: May 2004
Location: Melbourne, Australia
Send a message via MSN to benny
Quote:
now, crosover distortion only effects PP amps running any lower than absolutey in class A, which is pretty unlikely. so you think, ok, it's SE im medeling, so i'm safe on this last bit... wrong!
this was worded funny... what i meant it it is unlikely that a PP tube amp will be running 100% calss A... it can come close, but it's not likely you will find an amp at 100% class A operation.
  Reply With Quote
Old 6th June 2004, 01:03 PM   #4
diyAudio Member
 
Circlotron's Avatar
 
Join Date: Jun 2002
Location: Melbourne, Australia
Default The crack of willow on leather. Brass and polished stone.

One of the nice parts of having a tube amp is that when you are listening you can look at the box of bottles over there in the corner and say "Mmm.... tube amp...." The other part is the sound of course. You may indeed be able to get the sound of a tube amp with DSP, but you will not get the smell of warm transformer shellac or the magic blue glow of electrons in a vacuum.

There is more to sitting in front of the hifi than listening to music accurately reproduced to the 100th decimal place.
__________________
Best-ever T/S parameter spreadsheet.
http://www.diyaudio.com/forums/multi...tml#post353269
  Reply With Quote
Old 6th June 2004, 01:25 PM   #5
benny is offline benny  Australia
diyAudio Member
 
benny's Avatar
 
Join Date: May 2004
Location: Melbourne, Australia
Send a message via MSN to benny
Circlotron said:
Quote:
There is more to sitting in front of the hifi than listening to music accurately reproduced to the 100th decimal place.
i absolutley agree.... one of the charms of a valve amp is that they're not perfect. it is an amps imperfections that gives it charachter.
  Reply With Quote
Old 6th June 2004, 01:32 PM   #6
diyAudio Member
 
Join Date: Jan 2003
Location: Sweden
I see one big problem with your scheme, you assume that a tube amplifier distort in some way that is more pleasant than the distorsion that comes from SS amps, some people call this euphonic distorsion.

However what happens if you basic assumption is wrong? Maybe it is so that tube amps maybe give more measureable distorsion than a SS amp but less of hearable distorsion or distorsion that in general that has a negative impact of the sound reproduction?

Compare for instance how distorsion varies with power level in SS and tube amps, or how the harmonic signature look like in a SS amp an a tube amp, (does a tube amp have more or less of say 9th order distorsion than a SS amp?). I think you make a big mistake assuming that tube amp sound depends on that they have hearable distorsion, maybe it is the other way around?

Regards Hans
  Reply With Quote
Old 6th June 2004, 02:40 PM   #7
diyAudio Member
 
Join Date: Jun 2004
Location: Cluj-Napoca
quote: "you must understand that tube distortion is different in nature to just straight clipping of the wave form.. it is a softer type of clipping that rolls into distortion instead of cutting straight in"

-Yes I know that... and already made my point there.

You quoted me at a particular point (about the >2000Hz tube effect). I don't know much about that software device, but I can tell that it simulates tube' ordered harmonical distortion patern, making the HF much more pleasant.

I'm a little bit formal too, like u guys and I like too the "glow" feeling you describle, but I'm talking about efficency. I don't have "all the picture" in tube matters. I just wantet to say: "hey! what do u like more, eyes or ears?

Respect.
  Reply With Quote
Old 6th June 2004, 03:07 PM   #8
diyAudio Member
 
Join Date: Jun 2004
Location: Cluj-Napoca
Default benny

You really gave us a description about the "footprint" of tubes on sound signal. You say u don't know much about computers. I don't know mush eather, and my plugin is at essence very simple, but the sound it's very improved comparing with a normal linear response. The all the plugin does is prevent clipping above 0dB (and is very accurate) and compressing in a particular way the lower freq. so they sonud a little bit more "round" (the fast attack). It follows "the tube effect" at HF. And that's all. This way manifests all you have brought upper. All the little electronical "concert" you describled (I din't simulated it yet perfectly because no one will ever could (I'm nostalgic here ) gives just a few effects that the listener feels: compression and ordered harmonics (and maybe a few little more).
Although if my mind can and my devices (who cost 30US$), that means a professional in DSP with much more knowledge and budget can make things much better. I'm just waiting them to post...
And don't understand me wrong. I like more "tube sound" too, that's the reason I'm posting this because I want to have it anywhere even if I have not the real thing so I can see it too.
I know also that this techinque has been already implemented by the "big" musicians in their studios and it's quite for a while by now.
  Reply With Quote
Old 6th June 2004, 03:23 PM   #9
benny is offline benny  Australia
diyAudio Member
 
benny's Avatar
 
Join Date: May 2004
Location: Melbourne, Australia
Send a message via MSN to benny
[/QUOTE]
Quote:
You quoted me at a particular point (about the >2000Hz tube effect). I don't know much about that software device, but I can tell that it simulates tube' ordered harmonical distortion patern, making the HF much more pleasant.
i wasn't talking about frequencys that tubes clip at i was talking about the shape of the clipping. it is softer because the transition into clipping is more gentle. this is one of the big differences between transistor and tube amps. i think i also went on later in that post to talk about tube recovery from clipping and what not. once a tube has clipped, it takes a lower signal to keep it clipping until it recovers basically. did you model this?

you say:
Quote:
hey! what do u like more, eyes or ears?
but i said:
Quote:
just remember, i am not discrediting your model, and i'm sure it does make the sound much nicer... but it's not necesarily a tube amp sound... no way
it's not a question of eyes here. i did recognise that your amp model would make a nicer, and i will add a more tube 'like' sound, but i'm simply stating some facts about the seeming imposibility to truly model a tube amp. tubes are reactive to the signal they are presented with, and the effect a tube has on the sound is different even within the same song for example, as it reacts differently to different dynamics and frequency ranges, and all the factors are dependant. from what you described, everything you've modeled is independant of each other.

cheers.
  Reply With Quote
Old 6th June 2004, 03:29 PM   #10
diyAudio Member
 
Join Date: Jun 2004
Location: Cluj-Napoca
Default image

The's the panoramical view of my DSP software:
[IMG]d:\kxDSP.jpg[/IMG]
I hope it can be posted, if not I'm dissapointed, and sorry. I can't upload it because it has 80Kb and it's too big.
  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
The Future of CD SY Music 3 5th January 2008 12:42 AM
Future SMD artz Parts 4 9th June 2006 05:37 AM
what we could do with DigitalSignalProcessing Polimorph Digital Source 15 10th June 2004 09:39 PM
3 way for the future navin Multi-Way 0 16th November 2002 05:45 AM
Future Wish Lists.... mrfeedback Everything Else 2 15th November 2002 05:05 PM


New To Site? Need Help?

All times are GMT. The time now is 12:19 PM.


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