Sound Quality Vs. Measurements

Status
Not open for further replies.
Ed Cherry's JAES articles are a good source on these issues
without a input buffering ef the added degeneration of a current mirror driven "VAS" Q is largley a wash - the increased open loop input Z increases loop gain with the current mirror drive by about the same amount as the degen reduces the gm
Cherry does claim some dgen aids high frequency stability in his analysis of the standard RCA/Lin audio amp topology
 
It is local feedback, but only at low frequencies where Cdom does not do anything. How much feedback it applies at LF depends on how the stage is driven. The usual current-mirror in the LTP collectors will provide a high impedance drive so the resistor won't do much except to raise the VAS input impedance, which could raise LTP distortion by a tiny amount. A simpler resistor drive from the LTP will be affected more, but these are not seen so often nowadays.

If the 16R doesn't do anything then you won't hear any change! The bare transconductance of the VAS BJT will drop from about 0.3A/V to 0.05A/V, which will affect things even when Cdom is active. Careful simulation may show exactly what happens.


" Doesn't do anything " is a statement about my measuring equipment . The nicest compliment I had , I took to be an insult . An amplifier I built for a lady friend , she had measured by another friend in Austria ( already furious , I was gutted ) . I had no idea he had an Audio Precision measuring station . His comment was in German " what planet is Nigel from " . I was so upset and I don't mind saying that question has been part of my life especially when at school . After about 4 years I asked exactly what he meant by it ? He said something like " I have studied all my life and built many things , I never built one like this " Apparently he just couldn't find a flaw in it and was very depressed about it . When he asked I said . I only use measurements to keep me on coarse . Mostly I listen . Sometimes it takes me 20 years to know what I did . I will never accept bad is good . If bad does sound good I ensure good sounds just as good . Recently for valve designs I have allowed good to be 1% distortion . My amplifier I made for the lady would probably in 1970 be taken to be a 0% distortion amplifier . It sounds sweet and not unlike a 211 amp .

My problem with measurements and simulators is it is " painting by numbers " . However I don't object to a camera obscurer being used ( you choose what that is ) . BTW DF 96 , you are 100 % right , however I do not have enough days in my life for measuring equipment to catch up with my ears . Not least a test station I could buy . Most of my test gear I built . J L Hood comes closest to my hero if asking .

A Paul Kemble web page - working with sound, some tips.
 
Fine, no problemo, to each his own.

Obviously, I am sitting in the other camp.

4hz is fine . I bet your amp can be made into 4 Hz . Just sling those emitter resistors away .

Seriously . As long as the closed loop works why worry ?

I was watching about dust mites today and thinking about the ones that live on cheese . I rationalized this by saying the mites are made of cheese . The weirdest bit is I eat them with only kind thoughts , we both love cheese . It' s a choice and is my way of coping with the reality that science brings into my life .

Like amplifiers etc . The more nasty a cheese is sometimes the better it is ( for nasty read French ) . If not our hi fi will be like processed cheese . Doubtless the best cheese in the world is processed , the measurements I suspect prove it ? I rather like Philadelphia so not complaining . I never mistake it for cheese and I don't mean the state of if asking .
 
One other thing . I had this argument with my boss years ago about measurements . He had trained to the highest levels as an engineer . 7 years of study which was ruined by the war . Mostly he was a paper engineer except one famous connector called Hypertac ( still in production ). He did concede one point . He said even the smallest tyre maker would have a man who would know exactly what all the others know in the other companies . For all that tyres were an art . His statement to me was that empirical engineering can be a fast track to success . The mystery to him was how the empirical side seemed to be dominant , however the production side used science to make it repeatable . I thought it a very fair way of seeing it . I forgot to say " Easydrive " eventually bought by Toyota was one of his . The worst problem he had was being a dreamer and being stopped by people supposedly his superiors from running his bit of the company ( Smiths Industries ) . When I went to work for him he asked me to build him an organ . We never got past the asking .

His favourite question to me about an amplifier in his deliberately broken English ( he could speak 21 languages , 7 fluently including Russian ) " Nigel does it have the discerning ability " . I loved that and it sort of makes sense . For a true scientist he could bend a little . He worked on the JET project for nothing as a translator . It probably kept it in the UK as the Germans had no idea we were so advanced in fusion ( his opinion , I can beleive it ) . Even though the Germans team spoke English Emeric could speak technical German . That work came from him visiting Harwell one day trying to sell them something ( high speed video ) . He came away with an unpaid job . He understood Fusion at the level of the engineers , he bored me to tears about it ( just kidding ) . When I got married he wrote the formulea for fusion in my wedding photo album saying here is your future ( tritium and helium ) . He died the night Hallie's comet passed our planet ( I was flying over Beograd Dvv at the time , honest ) . I fancy it came for him.
 
I know . I even caught broken English from him ( truly ) . He tried to teach me Latin without success . If I got it wrong it was because I am " English " . If we annoyed him he would refer to his time in intelligence and silent killing . He said it to one of my friends who instantly pinned him to the floor and said prove it . That sort of sums up the paper engineer , it's all talk .
 
Nige,

From our personal correspondence, we both know we have similar views regarding engineering by numbers and by ear. We agree measurement is an excellent way to keep us on track and limit our flights of fancy.

Unlike DF96, I do care what my open loop bandwidth is and it has to be at least 50% wider than our hearing bandwidth, i.e. it must not be less than 30 kHz. Given my age, I guess I could be happy with say 24 kHz. More is better, but I won't go head over heels to make it ultra wide at any price. If I can do say 100 kHz, I will gladly trade it off for simple, plain, old fashoned stability. If it's not reliable, I don't want it.

I do not believe in extremes regarding global NFB, I neither like zero, nor 60 or 80 dB of it. I believe in moderation. I do not claim to be the Buddah of it all, that's just my view, no more. A rule I choose to live by, but by the same token, others will choose their own rules.

My reason is simple - I believe this approach has a better chance of producing a good sounding amp than either of the extremes, but I agree that maximum global NFB can produce excellent results here and there. Zero global NFB has not, to the best of my knowledge, produced any memorable (for me) results yet

I understand why say Thorsten loves FETs so much, it's just that I am not convinced that FETs by default sound better. Surely they can here and there, but so can BJTs as well, and tubes, etc.

Only my hearing is "by deafult", everything else varies. As I grow older, my deault will tend to decrease, and there's nothing I can do about it except wash my ears regularly.
 
I think it was King George VI Equerry who received a letter saying " Did the King know it was bad taste to have brown shoes with a blue pinstripe suit " ? The Equerry wrote back " He does , therefore it isn't " .

That's the point . We know and we choose to do differently .

My other point is I often sit at my modest test gear and play " what if " . I know for a fact Formula One guys do this . No one can say they lack facilitates or brains ( how many here would be able to live in that world , engineer or not ? ) . I looked at a Renault on test day . I said your problem is the back end is too light and you are not getting the power down . We all knew it was the regulations and not wanting extra weight . I suggested hairspray on the tyres . We laughed as we knew it would be banned . However two races might be the time to get banned . The idea came from Speedway racing , it was banned . Doubtless it wouldn't work as the speedway was thought to be dirt sticking to the tyres giving 1 second a lap . But who knows without trying ? The scientists of 1880 must have thought they knew most things . Most never guessed what was around the corner ? Listening to live Beethoven the other day told me we got nearer in 1938 than now . The recording engineers can not be 100 % of the problem . Vitally no recording I hear sounds like Beethoven . To sum up , real Beethoven sounds like direct cut 78's , it's magical . Maybe something that comes from another dimension can not be trapped so easily . Like us it has a soul whatever that might be ? Some of those 78's reached 32 kHz on the stampers if pressed on Vinyl ( post war FFRR , 7 kHz on the disk and only for one play ) ! Bob Stuart was saying this in the 1970's " Causes of distortion in audio amplifiers due to loss of information " i.e the soul .

TAS 194: Meridian Audio's Bob Stuart Talks with Robert Harley | The Absolute Sound

I took the chief engineer of Thoren's with me to FI test day . He only went to prevent me asking his wife ? She was furious as she is a journalist and she would have had access to the German drivers ( Silverstone test day ) . Kurt did say to Renault that the Red Bull were faster with the same engine as the car sounded different into corners ( rough ) . The engine the same and the mapping the same . It was the " Blown Diffuser " . They smiled at Kurt politely . He was right . Kurt is as penniless as me , that would have been a nice little number for him . The crumbs from the table of F1 would feed us nicely .

FET's measure worse . They are very obliging . Usually all their problems can be solved . Too many people read various things and it becomes " fact " . Facts + belief = mediocrity . Think about it . If the facts are the facts then only one solution is correct . That being true only price matters . As F1 proves . You can have all the facts and still loose a race . It is the unseen factors that count . That rock from space proves that .

I was looking at TDA 2030 this week . It is not far from being as good as we need . I have been driving a motor with it . I am impressed as they even at the UK's inflated prices cost $1.50 ( 5 off RS , PCB about same price ) . It might be a bit mediocre , dam well measures OK . I am so tempted to use it in feed-forward with some dumpers to crack 1R ( 10 V rms ) . In bridge 20 V rms ! Here is the irony , it might be better than a muscle amp into that load ? I dare say the 2030 can be stacked as did Douglas Self with NE5532 . After all the 2030 is just an op amp with grunt .
 
I bet they still would run rings around valve amps on that problem ? When Henry Ford grafted two model T engines together the world never was the same . Suddenly an engine so very humble was not . The Flathead Ford V8 . Model T was remarkable . With the state of our roads maybe that's what I need ? People who laugh at model T do not realize it's special design . If the going gets tough is almost a 4x4 and no complexity .

Any tips on stacking 2030's ?
 
Yeah - for audio there are better choices than TDA2030 I reckon. Not very transparent because of their RF susceptibility. I abandoned arrays of TDA2030s primarily though because sometimes on start-up they'd crowbar the supply, which was fine running from my bench supply because I could set the current limit so nothing got blown up. But I was unwilling to troubleshoot this problem as it was intermittent. Another reason for avoiding the TDA2030 was its lack of mute.

If you want to stack chips then TDA7293 would be my choice - its much easier as it has a master-slave facility built in. It also has mute. Its not quite as cheap though as TDA2030 (I was using the A version which has slightly higher max supply voltage).
 
its interesting how these low numbers of global feedback keep getting bandied about - this is still in the region where Baxandal's "harmonic multiplication" math shows some higher harmonics may be increased by the negative feedback

while >40 dB over audio bandwidth is quite practical today with RET or MOSFET output Q speed - the Linear Audio Putzeys article is unusually clear but the result that negative feedback should be maximized when used at all if you value accuracy is a standard feeback theory result

Putzeys Bruno (you have to cruise the site for the free sample of the article)

and at higher loop gains the higher order harmonic "hump" gets pushed into the noise floor - all visible harmonics are decreasing with order


DF 96 wrote about that sometime ago . As far as I can see ( under his guidance ) all that happens with increasing negative feedback is we are left with the harmonics which the failing gain of the amp can not fix ? I looked very carefully after that and found they were always there and looked the same . I had thought previously they emerged . If the amp goes unstable that is another matter . I did notice trends .These sometimes suggest an optimum feedback level has been reached . This might be , I can have 20% more wattage but it looks very ugly . I can still have that wattage and it will be 10% distortion all in perfect harmonic reduction in the harmonics . So what should I do ? Have the 10 % distortion and not the 2% . The reason being the below 80% matters much more . If designing transistor amps these trends still exist and can be manipulated . Best one I know of is let the driver amp clip first . No text books says that ? You can always have a switch to do it how you have been told to . Also makes for very quiet drivers ( and cheap , RC filter ) . Why use regulated power supplies when the output transistors are happy to do the same if not driven into ripple ? And it is so dam fast !
 
I think that as technology advances we will be seeing more and more all-in-one power amp chips.

I suspect that in say 10 years' time, discrete will be only in luxury items, while the Lo Fi and Lower Mid Fi will use power chips. Many already are.

On the other hand, I believe people will refine the sound of the power chips.

For example, here's one for Nige to ponder over - would a power chip sound better if it was buffered by a FET input op amp? I have a nagging feeling that it really might sound better, at least in this stage of their development.
 
Status
Not open for further replies.