John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Jump to Latest
Steve Eddy

Steve Eddy

Member
Joined 2002
theaudioguild.com
bcarso said:
Thanks to my and others figuring out how to get them to work, the TDA2007A class AB amp from STM was used in the millions and to a fairly good performance level given the insanely low price. I got concerned that the supply might cease and asked a ST person about it. He said Don't worry some idiots are still using it in megaquantities :)
Click to expand...

Mouser has no stock, no availability and lists it as an obsolete part. :confused:

se
 
dadod

dadod

Member
Joined 2006
wahab said:
Well, if some people like amps with 10dB gain margin why not...
Indeed there was, and still is, a tendency to beliefs that CFAs are immune from such limitations..
Click to expand...

The gain margin of my CFA is 30 dB (Phase margin is about 80 degree) and for VFA a gain margin is no less then 15 dB (Phase margin no less then 60 degree).
I don't see problem with today standard semiconductors here, in CFA I use vertical mosfets with fast drivers for OPS and easily can get an ULGF between 3 nad 4 MHz with very good PM and GM ,and, yes, almost 80 db of GNFB at 20 kHz.
 
J

jcx

Member
Joined 2003
Bode Maximal Feedback criteria

BJ Lurie does give some feedback design rules that I hadn't seen collected in one place before, the QA appendix has some simplified verbage that more here may follow
Appendix 11 Discussions

The function of the "Discussions" sections is to anticipate questions and objections, and also to address certain persistent misconceptions. In most cases, these are condensed transcripts of conversations which took place with the authors' colleagues and students, to whom we extend our sincere gratitude.
Click to expand...
Appendix 11, Classical Feedback Control


chapter 4 is also online, relevant to the question of how much global feedback is possible and how to get there

Chapter 4

SHAPING THE LOOP FREQUENCY RESPONSE

The problem of optimal loop shaping encompasses two fairly independent parts that can be solved sequentially (thus making the design structural):

The first part is feedback bandwidth maximization which is solved by appropriately shaping the feedback loop response at higher frequencies (in the region of crossover frequency and higher).

The second part is of distribution of the available feedback over the functional feedback band.

The feedback bandwidth is limited by the sensor noise effect at the system output, the sensor noise effect at the actuator input, plant tolerances (including structural modes), and nonminimum phase lag (analog and digital) in the feedback loop. The optimal shape of the loop gain response at higher frequencies, subject to all these limitations except the first, includes a Bode step.

The Bode step is presented in detail as a loop shaping tool for maximizing the feedback bandwidth. The problem of optimal loop shaping is further described and the formulas are presented for calculation of the maximum available feedback over the specified bandwidth.

The above solution is then generalized by application of a Bode integral to reshaping the loop gain response over the functional bandwidth (i.e., for solving the second part of the loop shaping problem). It is shown that the feedback is larger and the disturbance rejection improved in Nyquist-stable systems.

Loop shaping is described for plants with flexible modes, for collocated and non-collocated control, and for the loops where the plant is unstable. The effect of resonance mode coupling on the loop shaping in MIMO systems is considered.

It is described how to shape the responses of parallel feedback channels to avoid nonminimum phase lag while providing good frequency selection between the channels.
Click to expand...
Classical Feedback Control, Chapter 4: Shaping the Loop Frequency Response
 
Last edited:
scott wurcer

scott wurcer

R.I.P
Joined 2004
Steve Eddy said:
Yeah. Neil Young is no Steve Jobs. Can't think of an example of more overblown hype.

se
Click to expand...

Our own CH contributed some no feedback circuitry, I guess it didn't help. Please don't get me wrong, an open-loop I/V at 70dB THD is probably transparent in any DBT, just that it IS transparent and sounds the same as anything else through ear-buds.
 
Last edited:
B

bcarso

diyAudio Member RIP
Joined 2005
If I felt diabolical enough I would suggest to Neil Young that incorporating extra batteries and using EUVL-style current conveyors with floating supplies would enhance the sound quality still further.


Southern California Brings Me Down

I need someone to live with me
To keep my bed warm
And keep my shorts clean
I need a maid to give for free
And sew patches on my jeans

I dreamed I saw my cowgirl housewife
I was drivin' in my pickup through L.A.
I wanna love you while I can, babe
Before I become an old man

Southern California brings me down
Southern California brings me down
Southern California brings me down

Ooh, I need some place to go
Oh, north Ontario
It's safer than Alabama
It's safer than Ohio, ohh...

Gonna go home now
Where I can grow old
With the cowgirl of my dreams
I'm gonna stay stay stoned now
Just stare out my basement window and scream
Ahhhh.....

Southern California brings me down
Southern California brings me down
Southern California brings me down

Oooh, ooh, gotta get away now
Southern California brings me down
Ooh, ooh, babe
Ooh, gotta get away...

Topanga Canyon freaks, you won't see around no more

Actually I have enjoyed his music immensely over the years, and sometimes have to have my Cinnamon Girl fix when nothing else will do.
 
scott wurcer

scott wurcer

R.I.P
Joined 2004
bcarso said:
If I felt diabolical enough I would suggest to Neil Young that incorporating extra batteries and using EUVL-style current conveyors with floating supplies would enhance the sound quality still further.


Southern California Brings Me Down

I need someone to live with me
To keep my bed warm
And keep my shorts clean
I need a maid to give for free
And sew patches on my jeans

I dreamed I saw my cowgirl housewife
I was drivin' in my pickup through L.A.
I wanna love you while I can, babe
Before I become an old man

Southern California brings me down
Southern California brings me down
Southern California brings me down

Ooh, I need some place to go
Oh, north Ontario
It's safer than Alabama
It's safer than Ohio, ohh...

Gonna go home now
Where I can grow old
With the cowgirl of my dreams
I'm gonna stay stay stoned now
Just stare out my basement window and scream
Ahhhh.....

Southern California brings me down
Southern California brings me down
Southern California brings me down

Oooh, ooh, gotta get away now
Southern California brings me down
Ooh, ooh, babe
Ooh, gotta get away...

Topanga Canyon freaks, you won't see around no more

Actually I have enjoyed his music immensely over the years, and sometimes have to have my Cinnamon Girl fix when nothing else will do.
Click to expand...

I don't share your needs, I need an occasional Mark Pauline/ Survival Research Labs fix. I think Waly understands.

SRL - Survival Research Labs
 
Last edited:
Steve Eddy

Steve Eddy

Member
Joined 2002
theaudioguild.com
scott wurcer said:
Our own CH contributed some no feedback circuitry, I guess it didn't help. Please don't get me wrong, an open-loop I/V at 70dB THD is probably transparent in any DBT, just that it IS transparent and sounds the same as anything else through ear-buds.
Click to expand...

Yeah. But it was also CH's insistence on using some big through hole caps that gave the player its Toblerone shape which was a rather consistent complaint and the butt of many jokes.

And not everyone was using it with cheap earbuds. Plenty were using it with very good $1000+ headphones and in-ear monitors.

se
 
scott wurcer

scott wurcer

R.I.P
Joined 2004
Steve Eddy said:
Yeah. But it was also CH's insistence on using some big through hole caps that gave the player its Toblerone shape which was a rather consistent complaint and the butt of many jokes.

And not everyone was using it with cheap earbuds. Plenty were using it with very good $1000+ headphones and in-ear monitors.

se
Click to expand...

Soooorey, I didn't know that, I guess the unwashed masses were not wooed.
 
M

Markw4

Member
Joined 2016
Steve Eddy said:
Yeah. Neil Young is no Steve Jobs. Can't think of an example of more overblown hype.

se
Click to expand...

I think its pretty easy to see how Neil Young could have gone wrong.
My first "good" A/D/A was a Lynx 2 card. Specs here: Product Information
I tried an little experiment with it: Took a SM-57 mic, into a Grace 101 preamp, went in and out of the Lynx card (with no digital processing), thence into a Bryston 4B amp, and then to some ubiquitous NS-10 speakers. I talked into the mic while playing with sample rates and bit depths. Had a guy from a band there with me who was interested in the subject. We didn't have any test equipment at that point, so we just listened. Despite basic theory, the sound quality varied a lot from 16/44 up though 24/192. Why? I don't know. I'm sure there was some good physical reasons for it. Maybe part of it was the way Lynx divided up the clock to get different sample rates. Maybe more or less jitter in some cases? Again, I don't know, but it was quite obvious, nothing subtle. Anyway, the best sound of all was obtained by removing the Lynx card entirely and connecting the preamp directly to the power amp.

Having said that, I later tried the same thing with the A/D in a Cranesong HEDD going in and a DAC-1 coming back out. Didn't notice the same effects with that setup. It behaved essentially as basic theory would predict. Rarely used the Lynx after that, but it did add a little subtle edge to the sound at 24/192 that might be nice for acoustic guitar. If I wanted that effect, I might have used it over the HEDD in some limited cases.

On other occasions I also noticed that cheap sound cards as well as some "professional" outboard boxes sounded different at different sample rates and bit depth similar to the Lynx, but generally sounded much worse than the Lynx.

So, I can see how Neil might have gotten the wrong impression depending on what he was listening to, and who he was working with.
 
Steve Eddy

Steve Eddy

Member
Joined 2002
theaudioguild.com
Markw4 said:
I think its pretty easy to see how Neil Young could have gone wrong.
My first "good" A/D/A was a Lynx 2 card. Specs here: Product Information
I tried an little experiment with it: Took a SM-57 mic, into a Grace 101 preamp, went in and out of the Lynx card (with no digital processing), thence into a Bryston 4B amp, and then to some ubiquitous NS-10 speakers. I talked into the mic while playing with sample rates and bit depths. Had a guy from a band there with me who was interested in the subject. We didn't have any test equipment at that point, so we just listened. Despite basic theory, the sound quality varied a lot from 16/44 up though 24/192. Why? I don't know. I'm sure there was some good physical reasons for it. Maybe part of it was the way Lynx divided up the clock to get different sample rates. Maybe more or less jitter in some cases? Again, I don't know, but it was quite obvious, nothing subtle. Anyway, the best sound of all was obtained by removing the Lynx card entirely and connecting the preamp directly to the power amp.

Having said that, I later tried the same thing with the A/D in a Cranesong HEDD going in and a DAC-1 coming back out. Didn't notice the same effects with that setup. It behaved essentially as basic theory would predict. Rarely used the Lynx after that, but it did add a little subtle edge to the sound at 24/192 that might be nice for acoustic guitar. If I wanted that effect, I might have used it over the HEDD in some limited cases.

On other occasions I also noticed that cheap sound cards as well as some "professional" outboard boxes sounded different at different sample rates and bit depth similar to the Lynx, but generally sounded much worse than the Lynx.

So, I can see how Neil might have gotten the wrong impression depending on what he was listening to, and who he was working with.
Click to expand...

And this was all done non-blind, yes?

se
 
Bonsai

Bonsai

www.hifisonix.com
Joined 2003
hifisonix.com
Paid Member
scott wurcer said:
Yes as I said what can be achieved given a level of technology, just looking for some justification for a universal rule. In the end as much GNFB as technology allows seems to be OK. So if the ultra-high speed devices actually existed 60, 70, or even 80dB would be OK.
Click to expand...

yes agree - higher Ft output devices will allow greater loop gains at 20 kHz. I think I misunderstood your point.
 
M

Markw4

Member
Joined 2016
Steve Eddy said:
And this was all done non-blind, yes?

se
Click to expand...

Yes, and I did't say it was good science. It wasn't. Nonetheless, I was sure hoping it would not turn out the way it did. The Lynx was very expensive for me at the time, and I didn't have disk drive space to do a lot of recording at 24/192, as disks were still expensive as well. I didn't want to believe it and didn't like it, but there it was. In the end, I decided to compromise on cost and record at 24/88. I also learned to be very sparing with effects plugins when mixing. They all cause some damage to the sound in my experience.

Another interesting thing that goes against basic theory is digital EQ sample rates. In theory there is no reason to upsample, EQ, then downsample, but it does tend to sound better in my experience when done at higher sample rates. Again, I don't know why. I don't think any non-physical explanation is called for. Maybe A/D nonlinearities and clock jitter produce small mathematical errors in EQ algorithms. Or, maybe the upsampling and downsampling act like filters to smooth the rough edges of the sound. I looked to see if any research had been published on the subject, but never found anything. I expect if somebody does look into it carefully, some physical explanation will be found.

Having said all this, I know I that one possible explanation is that some or all of the differences are in my sensory perceptions more than in the equipment. But, I also think that some of this stuff hasn't been studied adequately by theorists. No serious academic I know would waste time on it because even if they did find something it wouldn't help their careers. It would be a lot of work with very little payoff, especially when there is so much other stuff to investigate.
 
Last edited:
Status
Not open for further replies.
Top Bottom