Amplifier topology subjective effects - Page 27 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

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 12th December 2003, 08:34 AM   #261
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Default More sims

I've simulated the following circuit :

Click here for schematic

This has all the previously mentioned VAS'es, all loaded by a current source which looks like their twin.

They are fed from a current source of 70K impedance and are studied open-loop. This means I had first to close the loop, note the base current, then add a bias source on the base so the output of the VAS is about 0V open-loop.

Thus we can simulate VAS open-loop response, something which is utterly impossible to do outside of a simulator.

Curves coming in shortly.
  Reply With Quote
Old 12th December 2003, 08:45 AM   #262
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
And here are the VAS transimpedances (Re=100) :
These are d(VAS collector voltage)/d(VAS base current).

normal 4.7 MOhm
cascode 48 MOhm
hawksford 1420 MOhm
baxandall 76 MOhm
bax+hawk 28200 MOhm

Now I'll plot normalized output voltages with all VAS'es swinging +- 20V (by dividing the respective input currents by these factors) and study the open loop output distortion.
  Reply With Quote
Old 12th December 2003, 09:20 AM   #263
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Default Distortion

I have posted too many graphs already, so I will just talk about the distortion measurements...

In OPEN LOOP, the surprise is that all VAS'es perform about the same. ie When they put out a +-16V output sine, they all add the same amout of distortion (well, not exactly the same, but it's close).

Here I give the distortion signal in Vpp :

Normal : 1.4V
Normal cascoded : 1.5V
Hawksford : 1V
Hawksford+Baxandall : 1.2
Baxandall : 0.5

... and I added a simple Darlington which performs as well as the Baxandall in this respect !

Note that all these stages distort a lot ! The normal one has distortion at -26 dB from the main signal (at 1k).

So, the large difference in distortion using the Baxandall does not come from it behaving a LOT better, it simply comes from it having more gain... and when you study it closed-loop, feedback masks everything. I wonder if it behaves better than a simple Darlington.
  Reply With Quote
Old 12th December 2003, 10:23 AM   #264
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Default Back to the start

Now let's go back to the subject of this post, namely the Aussie amp. Why does it sound good, and why did the designer use such a complicated (and symmetrical) VAS ?

I do think that fully symmetrical amps are a bad idea. With two input stages, you never know where your operatng point is, and both have to be very well matched. So I do agree on the single input stage.

But why the symmetrical VAS ?

He could have done something like that and save on complexity :

Click the image to open in full size.

Why did he not do it ? Why did he prefer the symmetrical VAS'es and why does it sound so good according to you ?

By the way, I realized this schematic is more or less the Otala amp (see schematic here). But the otala amp has reduces OL gain because of resistors going from VAS collector to ground, which was supposed to fight TIM, which does not exist, and just crippled the sound.

As for using two VAS stages, this may be to save on power supplies. If the amp is to output voltages close to the rail, the VAS must be able to go close to the rail. Hence the necessity for some voltage shifting, which is done here by the two VAS'es. But if we used two power supplies, we could use only 1 VAS stage (skipping the middle one in my schematic) which gives us more or less the Kaneda amp :

Click the image to open in full size.

And, note also that using a symmetrical VAS like the Australian Amp does needs two stages as the currents must be shifted from the upper to the lower rail.
  Reply With Quote
Old 12th December 2003, 11:16 AM   #265
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Default Simulations

From my simulations, the last diff stage could be like it is in the Aussie amp, or a simple diff stage feeding a current mirror, and the distortion would be about the same.

Besides, the distortion profiles look about exactly the same.

Thus either the diff+current mirror VAS sounds as good as the symmetric diff, or we have to investigate beyond THD (guess..)
  Reply With Quote
Old 12th December 2003, 11:23 AM   #266
AKSA is offline AKSA  Australia
diyAudio Member
 
Join Date: Sep 2001
Location: Melbourne, Australia
Peufeu,

Brilliant, inspired work!! Congratulations; you are reaching into the soul of amplifier design.

I would be interested in your thoughts on the use of lag compensation on these VAS. While the higher Zout of some of the designs gives higher gain, and thus more feedback factor, lag compensation, all other things being equal, will have more crippling effects.

It seems whatever you do, you come back to a similar outcome.

I mention this because in all my tuning of power amps I have noticed that the voltage amplifier characteristics of great importance are the speed of the device, the rail voltage, and the lag compensation. I have tried conventional, cascode and Rush cascode, driven differentially from collector resistors. All these critical parameters interact, and more lag compensation progressively cripples the sound. Less is best, and to my reckoning, the design which can tolerate the least lag compensation seems to sound the best.

Merci beaucoup!!

Cheers,

Hugh
__________________
Aspen Amplifiers P/L (Australia)
www.aksaonline.com
  Reply With Quote
Old 12th December 2003, 12:03 PM   #267
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Default distortion

I'm finding the Aussie amp's VAS (and Otala, too) distort more than the good old Lin (and symmetrical Lin) when using a VAS collector impedance of 1Meg.

I am therefore perplexed and will go for some biking after I eat...

Hugh, do you have any idea to explain the supposedly exceptionnal performance of that amp ?
  Reply With Quote
Old 12th December 2003, 12:27 PM   #268
peufeu is offline peufeu  France
diyAudio Member
 
Join Date: Mar 2001
Location: Lyon, France
Thank you Hugh...

> I would be interested in your thoughts on the use of lag compensation on these VAS. While the higher Zout of some of the designs gives higher gain, and thus more feedback factor, lag compensation, all other things being equal, will have more crippling effects.

Well getting unreal collector impedances in the VAS is not really useful because the output stage which comes after that will have a low impedance anyway. Maybe not that low if we use a follower, but not that high either. I'd say an output stage with follower, driver and output transistor will look like ... maybe a few megohms at DC, and above that, it'll look like crap anyway thanks to parasitic caps.

But maximizing VAS transimpedance still has its uses as it lets us learn which configurations are vulterable to the coupling through Cbc which is nonlinear and magnified by Miller effect... and it is still good to have a VAS which distorts less itself than what the output stage does to it...

All these loop systems always turn around as you say : if more gain = more distortion, feedback is useless.

> I mention this because in all my tuning of power amps I have noticed that the voltage amplifier characteristics of great importance are the speed of the device, the rail voltage, and the lag compensation. ...

What's Rush Cascode ?

> I have tried conventional, cascode and Rush cascode, driven differentially from collector resistors. All these critical parameters interact, and more lag compensation progressively cripples the sound. Less is best, and to my reckoning, the design which can tolerate the least lag compensation seems to sound the best.

Two thoughts :

Less compensation = the amp was faster to begin with, hence it'll have more feedback, less distortion.

And any parasitic effects in the compensation cap are magnified out of proportion by Miller effect :

When the operating point is defined by feedback, you only need to send very little current in the VAS base to make it swing the full voltage swing of the amp?

Say if your amp swings +- 40V, VAS collector impedance is 1Meg, current gain is 500, thus VAS transimpedance is 500 Meg (or 500 V/uA !) thus a full swing of 80V just needs 0.16 uA (not counting current going into Cdom at HF). It's even higher if your VAS has two transistors.

Now imagine your cap has a little dielectric absorption or (hell) microphonics in it, imagine if it generates only 0.16uA of spurious current this will be equivalent as a full swing... it will be hidden by feedback of course, but its contribution do distortion in the VAS and Input stages will be exactly the same as if it had been part of the music signal...

Hence the need for a good quality cap for your VAS. It's basically the part in the whole amp whose tiny imperfections are the most magnified... I usually use Styroflex...

That's what I think, what do you think about this ?
  Reply With Quote
Old 12th December 2003, 04:28 PM   #269
diyAudio Member
 
traderbam's Avatar
 
Join Date: Jan 2002
Location: Earth
Quote:
We are looking for the highest collector impedance, because this means less parasitic currents running through non-linear capacitances.
But what price is being paid to reduce this source of distortion? To tackle this the circuits use more amplifying devices, each of which is grossly non-linear in its primary signal path, let alone its parastic paths.

Typical designs use a single device and swamp the Cbc with a parallel linear cap, as you know. The cap also serves to slow the amp down and keep it stable. This works extremely well provided other problems in the system have been correctly resolved.

My question is how much distortion is being generated with the enhanced VAS circuits once they have the same lag C across them as a simple VAS?
  Reply With Quote
Old 12th December 2003, 05:16 PM   #270
jcx is offline jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
Excellent work Peufeu,

The next step after maximizing VAS gain is figure out how to use it to improve overall amplifier distortion (at least for us simulator jockeys!), “throwing away” the gain in a lag C (and sometimes a parallel R) to ground seems wrongheaded although it is seen some well regarded amps

“Miller” local feedback C around the VAS is the most common compensation technique and high gain inside the loop improves performance by reducing VAS output impedance where it often interfaces to nonlinear output driver impedance – such as directly driving the output fet gates which have very nonlinear Cin

Two less pursued options are Cherry’s moving the feedback C to enclose the output stage and 2 pole compensation (which work particularly well together in my sims)

Cherry’s compensation is rarely seen in discrete audio amps, a number of op amps use the technique; see the 100W TDA7293 data sheet for a strong "proof of principle" that the idea is viable in audio power amps, see also: http://www.ne.jp/asahi/evo/amp/J200K1529/report.htm fig 6&7 (not strictly “Miller” compensation but enclosing the output stage does improve the dist #)

Self briefly explores 2 pole compensation around the VAS stage in his book, this technique begins moves beyond the absolute gain stable feedback region but not very far, and higher order conditionally stable control loop techniques have been in use for over half a century, some threads here have tried to adress this
  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



New To Site? Need Help?

All times are GMT. The time now is 02:36 AM.


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