phono stage building blocks...

[piano3]

Yes DF96, I expressed myself badly; I meant that you cannot start from a desired frequency response-RIAA Bode plot say, with the 3 dB points-and simply write some lines of algebra which then spit out some component values for you.

 

[rayma]

you cannot start from a desired frequency response-RIAA Bode plot say, with the 3 dB points-
and simply write some lines of algebra which then spit out some component values for you.

That is indeed how an RIAA circuit is designed, starting with the defined poles (3180uS and 75uS)
and zero (318uS), and the gain required. The form of the equations depends on the circuit topology.
Limited loop gain can alter the circuit's response from the ideal in a nfb circuit. A passive circuit
does not have this problem, but instead must take into account the source and load impedances
for the passive RIAA network.

 

[sreten]

sreten: thanks for making the example for me. as I can be a little thick some days, I want
to look at this as a group of building blocks and your last post helps. As always quite succinct.

Hi,

Yes there is lots of information missing, but nothing you can't work out.
Like the voltage gain you need for each active stage and the current
needed to drive the chosen impedance of the passive RIAA sections.

rgds, sreten.

 

[piano3]

rayma, what I am saying is that if I were to ask someone to design an arbitrary filter for a new type of gramophone record with 3 dB points at 25 Hz, 300 Hz and 5000 Hz then they couldn't simply dust off their engineering mathematics textbook and work it out with pencil and paper.

 

[rayma]

rayma, what I am saying is that if I were to ask someone to design an arbitrary filter for a new type
of gramophone record with 3 dB points at 25 Hz, 300 Hz and 5000 Hz then they couldn't simply dust
off their engineering mathematics textbook and work it out with pencil and paper.

Yes, that is the design process. Once the circuit topology has been chosen,
the RIAA components can be calculated from the circuit equations.
Of course, there are many other details to be considered as well.
The calculation must be done exactly (symbolically) before doing the arithmetic.
Good references on RIAA calculation: http://waynestegall.com/audio/riaa.htm
http://www.pearl-hifi.com/06_Lit_Ar...s/Lipschitz_Stanley/Lipshitz_on_RIAA_JAES.pdf
http://andyc.diy-audio-engineering.org/phono_preamp.xhtml

 

[DF96]

It is a while since I looked at RIAA or similar circuits, but I would guess that to get component values you would not need to do anything worse than solve a quadratic equation. Even a cubic equation has an analytic solution, although very complicated!

 

[piano3]

rayma, thank you for the links which are easy and interesting to follow.

I first got interested in filters when I needed to build a 78 preamp more than 10 years ago. I scoured textbooks and couldn't find anything at all to help; lots about network analysis in both frequency and time domain but nothing about the process in reverse. I then made the assumption that the design process was an iterative one; i.e. tweaking a network until it produced the desired response.

 

[DF96]

Sadly, that is how some people do RIAA. Build it, then tweak it until it sounds 'right'. The result is inevitably wrong, but some people will pay a lot of money for such a phono preamp. Others do use a calculator, but are unaware of the fact that there is not always a trivial relationship between individual CR time constants and circuit time constants; they interact, in ways which depend on circuit details.

 

[rayma]

lots about network analysis in both frequency and time domain but nothing about the process
in reverse. I then made the assumption that the design process was an iterative one.

Yes, network synthesis is a more advanced topic than analysis. Tweaking usually isn't practical, since in most
circuit topologies the time constants interact. The RIAA calculations are fairly simple in the s-domain.
http://homepages.cae.wisc.edu/~ece902/LectureNotes/Simulation_1up/lec3c.pdf

 

[piano3]

I ended up by splitting the equalisation over two stages since I was confident that I could get that right but also this gave me a wider choice since there are quite a few different curves for 78s and early LPs as well.

Really DF96, do some circuit designers try to work out components by ear? That seems incredible......

 

[DF96]

They call it 'tuning'; basically a fixed tone control which matches the audio system and preferences of the 'designer'. I have seen people on this forum argue for this method.

 

[piano3]

I suppose I shouldn't be surprised. I know some "connoisseurs" who prefer to listen to the Vienna Konzerthaus Quartet on the earliest American Westminster records (time constants 1590, 400 and 100 microseconds) or the conductor Furtwangler on the earliest HMV ALP records ( Blumlein 500 Hz, i.e. only one clearly defined time constant of 318 microseconds), all on a modern super-expensive pre-amp with RIAA EQ only. This is claimed to be a superior musical experience.

 

[piano3]

Nice set of lecture notes rayma, thank you.

 

[rayma]

Nice set of lecture notes rayma, thank you.

The s-domain is like magic, time domain differential equations become simple algebraic ones,
that are easily solved.

 

[Nanook]

invoking Einstein....

rayma:

Invoking Einstein will not help: general relativity is given to us by the universe but we had to discover it; RIAA equalisation is something we decided to do

True, all I was suggesting was a single "unified" equation for the RIAA curve. And the idea of a curve of best fit to solve the equalization curve.

 

[DF96]

I said that, not rayma.

I don't know what you are asking for. The RIAA curve is described by a single rational function. In what sense is this not the "unified equation" you seek? Why use curve fitting when analytic solutions are known?

 

[TomWh]

Hey Nanook

Get yourself Morgan Jones 2nd or 3rd edition of valve amplifiers. He goes over the phono stage like no other. That being said when you are done with all the math you will need to check the curve in the circuit with some test equipment and switch out some caps/resistors to dail it in.

Thanks for your thread on diy tonearms it got me making unipivot tonearms. OK so I cheated, because I have a small machine shop, and used jewel bearings and lots of billet alum. I did use a alum arrow on one arm and carbon fiber on another.

Enjoy the ride Tom

 

[rayma]

The RIAA curve is described by a single rational function.
In what sense is this not the "unified equation" you seek?
Why use curve fitting when analytic solutions are known?

I think that Nanook lacks the math background to understand this,
and also that this unique rational function in fact DEFINES the RIAA curve.

 

[DF96]

Yes, it can be difficult to answer a question satisfactorily if the questioner doesn't understand the question and so cannot understand any correct answer.

 

[SY]

Get yourself Morgan Jones 2nd or 3rd edition of valve amplifiers.

VA3 is terrific, yes, but VA4 is significantly more comprehensive. Likewise Merlin Blencowe's latest book on preamps.