After a long sabatical from electronics design i am back on the drawing bord and have designed some phonostages, pre-pre´s, a preamp and will start to design a poweramp too.Being more of an analog guy i have help from good designers of digital gear so da converters and dsp may follow.
my comercial designs are made to have very low distortion, low noise and high speed so using voltage feedback (some call it NFB) is the order of the day. this can be done in many ways and results can be excellent also soundwise nevertheless it orcured to me that i would like to build a phonostage for my own purpose based on other priorities. having worked with Malcolm Hawsford has pursuaded me that maybe the tiny details so important for a feeling of reality (some may call it microdetail or ambience) can be lost if a portion of the output is fedback to the input to cancel distortion thereby also canceling the minuscel details. Douglas Self may not agree and he has certainly proven that a heavily fedback design with even the evil electrolytic capacitor in series with the cardridje can be made with very low measured distortion and noise. This contadictory standpoints stand in the way of marketing an expensive device that does not have extremely low levels of distortion and noise. Getting tired of arguing i desided to do this design for my own pleasure. If there is any interest i plan to make some by hand. Bu be prepared that it will be quite expensive and i can make only a vew in my rare sparetime. Actually today i finished a first prototype and it worked to my satisfaction although it it not fully optimised in any way. During this thread i will give more details. Let me say so much that it is a balanced parallel symmetric transimpedance design. I have seen such a circuit on the web but can not remember who posted it. it is an elaboration on prior art by Leach, Hiraga, Rambeyrolles and Rositer. I developped it a little further and will use very good components like MAT02, MAT03, teflon caps etc. construction is done in classic air dialectric fassion with the components directly soldered point to point. i will also try to take care of mechanical resonances, shielding, grounding and powersupply. i will also make a compasison to other designs, for example cascoded fet stages like in the pass ono. i will publish measurements and parts of the circuit diagrams. any input is wellcome. P.S. MPP stands for my phono pre
Today i will post a first schematic
components do not have the right values yet (i used the default values in TINA) but it shows the basic principles.
i am new to TINA so there can be some nodes missing but for the purpose of this post it is sufficient.
today i got also permission from Hugh of Aksa-Aspen to show some comparative measurements of his Paris Pre-Pre.
during this thread i will also show measurements of other Phono Stages and some of my commercial designs, this will include frequency and phase response, distortion and noise measurements
i will try to do a subjective appraisal too.
this is how the circuit works in my words :
the first stage is a folded cascode, balanced, parallel symmetric, common base transimpedance stage.
the 75 üsec time constant is done by feeding the constant current sources to ground. the output resistor of the first stage defines the dc gain and the ac amplification in the lower reaches.
in the second stage i got a little lazy and build a semi discreet intrumentation amp. this stage can be build totally discreet too and i will later post a sugestion. the differential stage does the 318 üsec and 3180 üsec RIAA time constants in shunt feedback fashion and partly passive.
the passive section is fed into the output of the inverting servo opamp, working as an artificial ground. first i had simply shunted the components to ground but that caused around 1.8 V DC at the output because the DC amplification of the total circuit is quite high. the first transimpedance stage is not that bad with only around 10mV (slightly fluctuating so i plan to thermocouple the input stage) but the instumentation amp has high DC Gain.
one interesting thing is that this arrangement lowered the noise floor under 1 kHz by a usefull 6-10dB the servo is very slow and does not worsen noise and distortion in any way.
so we have it: a dc coupled mc phonostage with no voltage feedback in the inputstage but series, shunt feedback and passive techniques in the ouputstage so there should be something for everybody (just kidding)
The schematic in my last post contained an error.
T8 was upside down. Thank you AG for alerting me on this mistake.
today i post the corrected schematic with two additions :
one is the possibility to design the current mirrors in the input stage as Wilson types and second, one could shorten the bases of the input transistors ac wise to ground with big capacitors (say 2200 üF) reducing the noise by 3dB.
I did not try the Wilson current mirrors but i know from experience that the capacitors give an improvement in noise suppression.
Next time i will publish some measurements of this circuit compared to some others.
again the schematic does not show the actual values but shows the topology in principle.
It seems that i am talking to myself here.
anyway, i do not like to have stupid errors creeping into my blurb.
the current mirror that i described in my last post is not a Wilson Type.
i am surprised that nobody hit a verbal hammer on me.
Instead it is a circuit that JLH in his book "The Art of Linear Electronics" calls
"Improved Current Mirror" (2006 Edition, Chapter 6, Page 93)
The German word for it would translate into something like "augmented 3 transistor current mirror" talk about babilonian times.
the advantage of that circuit is that it removes the base current of both input transistors from the reference current..
i will add a PDF to this post to show the differences. it shows the ICM and also the Wilson and the cascoded version. there is also a 4 transistor version of the Wilson that i did not include. by the way there is no boundary for even more sofisticated bias arrangements but i do not include them here because that will likely put an additional semiconductor in the forward path and reduce voltage swing somewhat, something i can not afford because i drive the circuit on a quite low +-9V supply to reduce noise. Maybe i try +-12V later. If you are interested in what can be done look no further then Per Anders Superbuffer. He seems to be the unsung master of the Biassing Art at least in discreet circuitry. IC designers of cause know all about it out of various reasons.
My next Post is maybe more interesting because i will add some measurements of the MPP prototype compared to the ASPEN-Aksa Paris and my soon to be released commercial Pre-Pre named Quantummusic Gold Standart
P.S. the current mirror with the J-Fet is just an idea, i did not try it sofar.
will it work ? maybe
I'm almost completely lost when it comes to solid state electronics, but as a speaker designer i recognized your name and would like to welcome you to the forum.
Hi Dave !
Thank you for the invitation.
I designed more speakers then i can remember and when i was younger that included horns and TML`s too. With the modern measurement and simulation tools like Klippel, Praxis and LSP CAD the work got much more easy.
Sometimes a bit too easy for my taste so i took a detour into electronics.
Actually one of my first commercial products in 1983 was a low noise, high speed pre-pre so i am no stranger to this art.
I got also interested in active speakers so maybe i can contribute something for you speakerbuilders in another thread too.
strangely enough with my age of 52 i see a kind of new spring comming and my own creativity really surprises me.
one of the reasons may be that i do not travel as much worldwide as i did and also took a step back from dayly operations like invoicing, selling etc.
you may know that i left Audio Physic some 5 years ago and design now for Sonics by Joachim Gerhard. I also work for Quantummusic created by Joachim Gerhard so you may see some interesting things comming in the future.
We haven't met before, but our mutual friend Malcolm Hawksford had very great things to say about you when I interviewed him for AudioXpress a few months ago. (Interview part I published in the Nov 09 issue of AX).
I am looking forward to read and talk about your designs!
As promissed i will post some measurements i made on the MPP prototype, the Aksa-Aspen Paris and the Quantummusic Goldstandart.
First i whould like to show you a loop measurement of my measuring equipment. i use the DAAS Pro 96kHz/24Bit measurement suite with modifications to be able to measure at low level. the modifications i made are quite similiar to the sytem that Douglas Self proposed to measure low level distortion artifacts. you can see that arangement on his website
The Douglas Self Site under "Science and Subjectivism"
by the way very funny reading. i love that english pragnatism and he drives a hard bargain
the difference to his solution is that i made the curcuitry works in pseudo balanced with 5 Ohm Johnson resistance to mimic my reference cardridge, the Lyra Titan i i also use the Olypos, the Helicon and an older Clearaudio Signature. for reality check i have a Goldring MM that came with their turntable by the way i am able to measure the resistance and inductance of cardridges without destroying them but sofar at least the low impedance designs have such low inductance that you can drive them into more or less any load without loosing much treble. i did not measure the higher impedance Clearaudio though and whould love to get my hands on a Benz LP.
It is promissed to me from the German importer as soon as they have some spare.
you can see that the resolution is around -150dBV if you ignore some intermodulation producs from the powersupply and other radiating equipment on my workbench. i dressed the cables as good as posible but this is the limit. as you can see the distortion of the chain is at ca. -140dB.
i find that adequate for my purpose.
Hello Jan !
i really whould like to meet you some day. maybe i just drive by if you care.
by the way i started to buiild your Pax anplifier and may need some advice from you.
How could i buy this issue of AudioXPress ?
Yes Malcolm is the most creative design genius i ever met and we did a lot of exiting stuff together. unfortunately he is far away and i am not so flexible any more then i was 15 years ago. i hate that driving on the left side ! i am a poor driver anyway.
|All times are GMT. The time now is 11:22 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio