Please note that, to get a good CMRR, you need 0.1% (or better) resistors (for the 2 100 resistors).
Maybe this is of interest for the academics :https://www-docs.tu-cottbus.de/kommunikationstechnik/public/rohde2011ULR_MUNICH_PRESENTATION.pdf
No, no, I am taking it as an compliment
To be more clear, the simple single opamp differential amplifier of my riaa, that is used as an input buffer (in effect a pre pre), was widely criticized and even dismissed. To see it popup, here, is (for me) recognition, and validation, of the qualities of the original design.
Joachim, I cannot see where this is going, so I try again.
I think I was just saying the same thing (as you), and at the same time I tried to convey a message to the 'opamp'-skeptics.
My point was that the mere fact that I was using an opamp was criticized (by many, not you, you stated that you liked the design (as others did)). As I said in the thread, I think the time is there, opamps are ready for the BIG jobs, and that is why I designed my RIAA using only opamps as active devices (in the signal chain).
On the same time I am not dismissing that you can make 'a better' amplifier using discrete devices, but then we have to recognize that if you can do it with discrete than it should be possible to do it with opamps (maybe not now, but soon).
We have now arrived (after 30 years of opamp development) where we need to select devices (opamps and/or discrete devices) by looking at the application, and we are free to choose the one or the other, we no longer have the 'luxury' to select discrete devices all the time, just because ‘they are better’.
Yes, my design sounds great (anyway that is what the people at Friclefest told me
) and it is not the final answer, especially in relation to noise. But the tiny 8 legged monsters have come a long way, for many the message should be; these (the opamps) are no longer the monsters we thought them to be; they have grown up and can hold their own ground, some are growing into real treasures.I hope this clears it up
My point was that the mere fact that I was using an opamp was criticized (by many, not you, you stated that you liked the design (as others did)). As I said in the thread, I think the time is there, opamps are ready for the BIG jobs, and that is why I designed my RIAA using only opamps as active devices (in the signal chain).
Yes, my design sounds great (anyway that is what the people at Friclefest told me
I hope this clears it up
Hi Frans
I had not much time to read properly the last few posts loads of work at day job and weater is so nice that one get home and sit in the garden with chilled botle of Sol and lime.
I realy started to folow this tread (see my old post about my INA LT1028 MC phono) as I am learning and growing tanks to you all here.
I like the 8 leged critters and I will certanly folow that up so please keep posting, at the same time I like the Paradise as it offers me a real opportunity to learn a lot (by time I did ) + if it sound as good as "it is just there" or "when I feel like taking it out of the system" is all a big plus.
So I am going to build both and a few more like Douglas Self MC pre pre (Just need a suply) and the lot from Sound of Silence.
Yes I could go to the shoop and buy the last magazine-rave-in-a box for less.
What sodisfaction one get from ownership escapes me.
Al I want to say is just KEEP POSTING I am learning, making nice music, keeping busy doing no harm to no one and I tank YOU for that.
Who cares about the rest?
Thanks, Frans, for me it is clear anyway. I design with Opamps, discrete and even tubes.
I also said many times that it is hard to design discrete circuits that are technically better then Opamps.
Two domains are left and this is ultra low noise input stages and high power applications like output transistors .
Ultra low noise input stages ( lover then 0.9nV/qHz ) can be made with Opamps too by paralleling them, just like we do parallel BJT´s in the Paradise.
I have shown a circuit called "Dolfin" here and Glen Kleinschmidt showed a 0.4nV/qHz design with 8 Opamps in parallel on another thread.
Even Power amps can be made by paralleling chips, see D.Selfs Grid Amp for Elektor.
It is just a matter of creativity.
The reasons we designed the Paradise discrete are three fold :
1 : It is more fun.
2 : It is a bigger challenge.
3 : It behaves technically different then a global feedback design, Opamp or not.
Let me explain point 3. ones again:
The Paradise has the structure:
Transconductance input - Transimpedance RIIA - Open Loop Buffer.
It has no global feedback but you could view the Transimpedance RIAA as frequency dependent current feedback or even simpler call it degeneration.
That has two consequences : distortion goes down in level monotonically ( the less the voltage swing the less distortion ) and distortion goes down in the treble.
This is much as the ear that resolves low level-high frequency signals very well, especially to "measure" and localize any aggressive impact, the "Caveman syndrome ".
I was rationalizing that this feature of the Paradise gives very good low level resolution and a very 3 dimensional sound.
A global feedback phono would do different : static distortion is maybe lower but distortion does not fall monotonically with level and the distortion goes up in the treble and is of a more complex nature, although potentially on such a low level, so that it is not audible and the distortion in the treble rising only over few 100kHz where we can not hear anyway. You see the Global Feedback case i call "brute force" and the discrete Paradise topology i call " natural".
One advantage of the discrete is also that we have more control over parts choices, for example finding components that have a positive effect on the sound like special resistors and such. We can also tune in bias, voltage swing, Layout etc.
An argument could be setup that in a discrete design the signal travels through less parts that also have a higher quality so we potentially have higher transparency.
The problem is that the choices we can make bear the problem that we make the wrong choices so at the end we may like the sound but in a strict sense we do not reproduce the signal faithful any more. The old subjective vs. objective debate.
I also said many times that it is hard to design discrete circuits that are technically better then Opamps.
Two domains are left and this is ultra low noise input stages and high power applications like output transistors .
Ultra low noise input stages ( lover then 0.9nV/qHz ) can be made with Opamps too by paralleling them, just like we do parallel BJT´s in the Paradise.
I have shown a circuit called "Dolfin" here and Glen Kleinschmidt showed a 0.4nV/qHz design with 8 Opamps in parallel on another thread.
Even Power amps can be made by paralleling chips, see D.Selfs Grid Amp for Elektor.
It is just a matter of creativity.
The reasons we designed the Paradise discrete are three fold :
1 : It is more fun.
2 : It is a bigger challenge.
3 : It behaves technically different then a global feedback design, Opamp or not.
Let me explain point 3. ones again:
The Paradise has the structure:
Transconductance input - Transimpedance RIIA - Open Loop Buffer.
It has no global feedback but you could view the Transimpedance RIAA as frequency dependent current feedback or even simpler call it degeneration.
That has two consequences : distortion goes down in level monotonically ( the less the voltage swing the less distortion ) and distortion goes down in the treble.
This is much as the ear that resolves low level-high frequency signals very well, especially to "measure" and localize any aggressive impact, the "Caveman syndrome ".
I was rationalizing that this feature of the Paradise gives very good low level resolution and a very 3 dimensional sound.
A global feedback phono would do different : static distortion is maybe lower but distortion does not fall monotonically with level and the distortion goes up in the treble and is of a more complex nature, although potentially on such a low level, so that it is not audible and the distortion in the treble rising only over few 100kHz where we can not hear anyway. You see the Global Feedback case i call "brute force" and the discrete Paradise topology i call " natural".
One advantage of the discrete is also that we have more control over parts choices, for example finding components that have a positive effect on the sound like special resistors and such. We can also tune in bias, voltage swing, Layout etc.
An argument could be setup that in a discrete design the signal travels through less parts that also have a higher quality so we potentially have higher transparency.
The problem is that the choices we can make bear the problem that we make the wrong choices so at the end we may like the sound but in a strict sense we do not reproduce the signal faithful any more. The old subjective vs. objective debate.
Last edited:
I certainly second that ;-))) although I am not really a fan of opamps, more vacuum tubes, and here I am shoveling sand all the time....
I feel the same way since I started building my own stuff
received your boards yet? I am so much interested in hearing from your impressions!
oh, please don't get me started... ;-) last time I counted I have 6 or 7 phono stages, some of them not really presentable but who cares! Getting to a stage where it is too much for my wife ;-)
that being said, I have a box full of C3g, and some high-mu cores from Epcos, and a couple input transformers from Lundahl, so a two-stage topology with input trannies and LCR equalization between two gain stages would be in order....
that being said, I have a box full of C3g, and some high-mu cores from Epcos, and a couple input transformers from Lundahl, so a two-stage topology with input trannies and LCR equalization between two gain stages would be in order....
Sure, i have more then enough to do.
By the way i solved the problem of making a transimpedance line stage with enough dynamic range. One trick is the pot meter i already discussed with you Michael but there is also another way ( or combined ). So far we have loaded only the output but a transimedance amp has not only a positive input and output but also a negative input. That negative input can be resistively loaded too. For example loading the output with R1 ( where normally the RIAA is ) and loading the negative input with R2 = 1/2 R1 gives a gain of 2.
By the way i solved the problem of making a transimpedance line stage with enough dynamic range. One trick is the pot meter i already discussed with you Michael but there is also another way ( or combined ). So far we have loaded only the output but a transimedance amp has not only a positive input and output but also a negative input. That negative input can be resistively loaded too. For example loading the output with R1 ( where normally the RIAA is ) and loading the negative input with R2 = 1/2 R1 gives a gain of 2.