Open loop gain (High or low) ?

[JensRasmussen]

I have been thinking about this...

When a NFB amp is designed, is the idear then to have as much open loop gain (OLG) as possible to correct any errors on the output ?

The Leach amp is a low feedback amp, does this mean that the error correction is lesser in this amp (also why the THD rises with freq)? Why do this when the errors will be more clear ??

Does alot of feedback "kill" the sound of an amp (making it sound boring) ?

At the moment I'm trying to adjust my design to have OLG at about 100-120 dB until 10 kHz, the faling at 20 dB / dec. making it a High feedback system.

I'm thinking af the amp like a "power opamp"

Is this the way to go ?

Thanks for any input

\Jens

[leo529]

sorry for my poor english, hope u can understand what i am saying...

if your open loop bandwidth is high enough to reduce the TIM, u can set the gain of your amp as high as u can...

i just read the circuit of leach amp, i think it has quite low open loop distortion, since u can see the cascode circuit in the 1st stage and some local feedback networt in its design..(R7.8,9,10 in the 1st stage, and R30, 31 in the sec stage )

Deep feedback will not "kill" the sound, if u can reduce the TIM

i am not sure your OLG idea can work or not, but i think u need to use at least 3 gain stage and keep the open loop bandwidth of each stage to over 20kHz, u can see this design in Mark Levinson's No.33X series....

[traderbam]

Hi Jens,
What a question you ask...the answer is not simple. Interestingly, the Naim amps are designed like simple power op-amps and they work pretty well. Many that are designed this way do not sound good.

Hundreds of textbooks have been written about how to use negative feedback to improve the performance of systems. And your assumption that more feedback gives less error is generally true provided that stability is maintained. But the ear is not so bothered about the TOTAL error but rather the dynamic, spectral content of the error. This is profoundly important and profoundly complex to measure and correlate with perceived sound quality. Anyhow, this is the reason why sometimes adding feedback makes a circuit measure better and sound worse. The main reason why feedback messes up the spectral content is non-linearities in the amp circuit.

So you have two key things to consider: the signal tracking performance of the NFB system AND the spectral effects of the NFB system. The first can be learned from textbooks - this is where you will find whether having a 20dB/dec roll-off at unity gain is ok or not. The second is a lot more challenging but I suggest you reduce the non-linearities as much as possible and then choose the optimum amount of feedback for your circuit by listening to it.

BAM

[Nelson Pass]

Not to put too fine a point on it, the complexity of circuit
that gets you a high open loop gain and the large amount
of feedback it can give you are both killers of sonic quality.

If you want the best sound, stick with simple gain paths,
maximize the linearity, minimize the feedback.

[jcarr]

Jens: Either local feedback or global feedback can reduce distortions and improve sound, but in my experience, local feedback is the easier technique to implement; with comparatively little chance of impairing the sound quality. Global feedback can have a greater effect on the total performance and sound, but depending on the circuit and your abilities, this may be a good thing or a bad thing - global feedback can also degrade stability and introduce side-effects that may prove sonically troublesome.

If your heart is set on trying deep global feedback, from experience I find it easier to do with circuits that are relatively simple (functionally simple, not simple as in few parts), and are designed to drive simple resistive loads. Circuits that have multiple internal stages and a number of dominant poles are usually much trickier, as are those that have to deal with capacitive or reactive loads. Also from experience, circuits with deep global feedback are more picky about their circuit board layouts and constructions.

regards, jonathan carr

[JensRasmussen]

Well thanks for the comments. Now I'm really confused.

I understand the comments, but I'm unsure what to choose.

Mr. Pass says keep it simple, while others say that simplicity is less important.

I agree that the ear is the final judge over the sound of the amp, but is an amp supposed to contribute to the sound, or schould it be completly neutral.

I think the amp is supposed to be neutral, and not "colour" the soundimage. Therefor I think measurements are important too. If an amp is more than just a gainfactor, it gets more diffecult to acess the quality of the signal path and thereby the hole system.

I rather like the idear of not having global feedback, since this demands better individual stages. The only problem is that I can't really find any books on this.

I suppose I could just start by designing one stage at a time until I have a complete no feedback amp. LC audio of denmark has a nice kit called the end, but it's hard to find out what transistors they use. Besides I would like to try on my ovn (and with the help of all you guys and gals) since I'm a EE student I'd like to learn the art of amps, and it's alot better way to learn this way.

thanks again

\Jens

If anyone is interrested lets get our heads together and make the best amp ever

[UrSv]

The End according to schema/build info (i.e. publically available)

As far as I can see among others.

2SC2922/2SA1216 O/P
2SA1606/2SC4159 Driver
SMBTA42/SMBTA92/BFN18/BC847A/BC818-16/BF822

Old listing in http://www.audiotuning.de/pdf/millenium.pdf

If nowhere else for marking on SMD:
http://www.geocities.com/CapeCanaver..._trans_it.html

It is all there on their web page so I can't really see how it could be difficult to find when they give you the schematics...

/UrSv

[traderbam]

As an EE student you will want to measure something and then adjust your circuit until the measurement goals are met. That's fine but it won't give you a great amp.

Forgive my presumption but the reason is that you don't know what to measure, beyond trivial stuff that is. If you can deduce a set of test conditions, measurements and measurement goals that are comprehensive to perceived "neutrality" then you're fine and you will soon be a millionaire. The Nelson Pass' of this world will (probably) not reveal their measurement processes because they are absolutely critical to competitiveness. So the amateur is faced with having to both develop a circuit and the measurement criteria concurrently.

In other words you may need to use your ears to correlate circuit behaviour and circuit changes to changes in the sound quality. Then work out the electrical theory and then improve the circuit. The true engineer does this thoroughly and believes in the principles of physics rather than giving up and inventing spurious "black art" reasons for things. The more you identify what causes amps to sound the way they do the more you realise how simple some of the electrical effects really are.

[JensRasmussen]

UrSv:

Thanks for the SMD info, just what I needed !

I still want to try on my ovn though

traderbam:

Thanks for the reply....You are 100 % right, The problem is finding out how a measuremet sounds !

If anyone has any tips please let med know.

Thanks again

\Jens

[hitsware]

here's my 2 cents towards the simple low feedback approach