Hi,
Yes, I find the OPA1632 very interesting as well.
I just wish TI would have given us four more pins, two for the collectors of the input stage (so we can add a LSK389 instead of what is on-board) and two for the transconductance nodes (so we can put an external load and use it zero feedback)...
For "Audio" I normally prefer the OPA637, which does not really have a good alternative.
In the Linestage with tone-control, due to the variable gain and the intention to put a Volume control upfront, we cannot use the usual tricks to stabilise the OPA637 at low gains, so in this case the unity gain stable OPA627 is needed...
Ciao T
Yes, I find the OPA1632 very interesting as well.
I just wish TI would have given us four more pins, two for the collectors of the input stage (so we can add a LSK389 instead of what is on-board) and two for the transconductance nodes (so we can put an external load and use it zero feedback)...
For "Audio" I normally prefer the OPA637, which does not really have a good alternative.
In the Linestage with tone-control, due to the variable gain and the intention to put a Volume control upfront, we cannot use the usual tricks to stabilise the OPA637 at low gains, so in this case the unity gain stable OPA627 is needed...
Ciao T
Folks,
I am afraid I made a rather grievous error in the way I calculated the HF load onto the OPA627 in my Tone Control example when Bonsai asked me. And happy that my mistake seemed to support his view that "my" design was "bad" he failed to fact-check, but proceeded to bash what I had posted. Alas, as it so happens, without good reason, it would appear...
It appears I cannot read circuits either, or rather, slave to convention I unquestioningly expected boost to be with the controls UP, not DOWN, as it is in reality (this tone control is in the feedback loop and hence "turns the other way").
The real load with the treble fully boosted is around 2KOhm @ 20KHz, which will approximately lead to 0.015% THD(20KHz), not 0.05% THD(20KHz), for a 7V signal.
Sorry for the confusion.
Ciao T
I am afraid I made a rather grievous error in the way I calculated the HF load onto the OPA627 in my Tone Control example when Bonsai asked me. And happy that my mistake seemed to support his view that "my" design was "bad" he failed to fact-check, but proceeded to bash what I had posted. Alas, as it so happens, without good reason, it would appear...
It appears I cannot read circuits either, or rather, slave to convention I unquestioningly expected boost to be with the controls UP, not DOWN, as it is in reality (this tone control is in the feedback loop and hence "turns the other way").
The real load with the treble fully boosted is around 2KOhm @ 20KHz, which will approximately lead to 0.015% THD(20KHz), not 0.05% THD(20KHz), for a 7V signal.
Sorry for the confusion.
Ciao T
Last edited:
A very fair comment. The Baxandall configuration is not well-suited to this. The problem is that to do it at the LF end requires enormous (polypropylene) capacitors. At the HF end it brings down the input & feedback impedances to very low levels and it is hard to drive.
It's less of an issue with 1K pots, but the primary aim was ultra-low noise, with reasonable complexity.
The 5532 question is fully answered in my Active Crossover book, Briefly the answer is that there are significant differences in distortion performance, and the TIs are worse than Signetics.3. Finally, one of those NE5532 questions. Are all widely available examples (thinking manufacturers here) similar in performance ? There's always the usual "web chatter" about for example TI ones being inferior and the old Signetics being the best and so on. Is there any truth lurking there or not ?
You finished back 1996 asking whether that was the quietest preamp ever yet built. Perhaps that statement should be moved to the 2012 design
As for the noise- well, let's see if someone can beat it.
Folklore has the old Raytheon ones as best. Being a classic all diffused untrimmed part the bias levels can be all over the place. In the bad old days some LF356's were so "hot" that they were not unity gain stable anymore.
I think Groners comments are succinctly put and fact based and therefore a pleasure to read. I learnt something - as I did reading DS.
As for the noise- well, let's see if someone can beat it.
Am I missing something, Friis' formula etc., the SNR better be set at the phono front end or the designer is not doing their job?
Last edited:
Friis formulas for noise - Wikipedia, the free encyclopedia
In an active MC phono stage noise is frozen not friisen to 0.3nV/qHz not counting 1/f noise. To beat that can be done with passive technology.
In an active MC phono stage noise is frozen not friisen to 0.3nV/qHz not counting 1/f noise. To beat that can be done with passive technology.
Well sir, I don't mean any disrespect but I do have two questions.
Bass, Drum, L-Voice, H-Voice, Rasp, Air
I couldn't think up better descriptions at the moment but it is three harmonic tilts with individualized levels. I believe that such a thing is suited to a fine preamp, since its impossible to please everyone with just 2 knobs and it is also unnecessary to supplant a 24 band equalizer in a device is supposed to be a preamp that is quick to operate (but not at the expense of effectiveness). Somewhere in-between is suitable and that's my view. Thereby, I finally got to the question: What upgrade would you do for the baxandall section to enhance the usefulness?
A brand new design with NOS parts, oh my. So, if you had to buy them new, what, specifically, would you buy?
While fantastic for the Andrews Sisters on LP (which I DO like), there are a considerable amount of music types and recording media where that baxandall is not helpful but is there instead of something helpful. Considering the elaborate nature of the preamp, may I respectfully suggest these controls:
Bass, Drum, L-Voice, H-Voice, Rasp, Air
I couldn't think up better descriptions at the moment but it is three harmonic tilts with individualized levels. I believe that such a thing is suited to a fine preamp, since its impossible to please everyone with just 2 knobs and it is also unnecessary to supplant a 24 band equalizer in a device is supposed to be a preamp that is quick to operate (but not at the expense of effectiveness). Somewhere in-between is suitable and that's my view. Thereby, I finally got to the question: What upgrade would you do for the baxandall section to enhance the usefulness?
Last edited:
A very fair comment. The Baxandall configuration is not well-suited to this. The problem is that to do it at the LF end requires enormous (polypropylene) capacitors. At the HF end it brings down the input & feedback impedances to very low levels and it is hard to drive.
It's less of an issue with 1K pots, but the primary aim was ultra-low noise, with reasonable complexity.
The 5532 question is fully answered in my Active Crossover book, Briefly the answer is that there are significant differences in distortion performance, and the TIs are worse than Signetics.
As for the noise- well, let's see if someone can beat it.
Thanks for the explanations and the rationale behind it all.
Looking forward to the next installments now
Hi,
Actually, in the tone control thread elsewhere in this venue, where my controversial "One Op-Amp does it all" line stage with tone controls originated we have also discussed ME options for multiband EQ.
If we want something like the EQ found in the Cello Palette it can be in principle done by just adding more bands to the Baxandall/James stack. Given that the midrange controls have a much narrower range of control than the outermost bands this does not even compromise impedance much.
So, if you want a line stage with 6-Way Cello Palette style tone control (defeatable), physiologically corrected volume control and balance you need precisely one op-amp per channel plus a buffer if you are concerned about loading the OPA too much.
If anyone wants to discuss this level of tone control I suggest to move the discussion into the original "High End Tone Control" thread.
http://www.diyaudio.com/forums/analog-line-level/194119-high-end-tone-control.html
Ciao T
Another dozen opamps to solder?!!!
Actually, in the tone control thread elsewhere in this venue, where my controversial "One Op-Amp does it all" line stage with tone controls originated we have also discussed ME options for multiband EQ.
If we want something like the EQ found in the Cello Palette it can be in principle done by just adding more bands to the Baxandall/James stack. Given that the midrange controls have a much narrower range of control than the outermost bands this does not even compromise impedance much.
So, if you want a line stage with 6-Way Cello Palette style tone control (defeatable), physiologically corrected volume control and balance you need precisely one op-amp per channel plus a buffer if you are concerned about loading the OPA too much.
If anyone wants to discuss this level of tone control I suggest to move the discussion into the original "High End Tone Control" thread.
http://www.diyaudio.com/forums/analog-line-level/194119-high-end-tone-control.html
Ciao T
I got the french April paper issue of Elektor this morning. As usually from what comes from D. Self, this innovative design uses unusual circuits using usual parts. Apart from paralleling ICs for which we have been accustomed with the three last Douglas's books, it's the first time I see a buffer for the bass section of the Baxandall control. And albeit the configuration around Ic9b or Ic18a was not unknown to me, it's the first time I see this kind of buffering (may I call it "current dumping" ?) used in a small signal audio circuit.
Hi,
My German issue also turned up. As I meant to express and expected in a proir post, the distortion data for the tone control in extreme positions is not given (my previous note missed a NOT in there).
It's been in Burr Brown datasheets for decades, meanwhile parallel Op-Amp's have been in LT datasheets for similar ages, at a minimal exercise in memory...
What I find amusing is that the positive polarity of the output is driven by four parallel OPA's, yet the other polarity is driven by only one and the inputs and outputs are pseudo balanced, instead of true balanced connections (this design of course is not alone in this, it seems par for the course in "HiFi").
I am sure there are good reasons for this, but it is disappointing, given that there are already so many parts in use, that to add the remaining few to get true balanced in's and out's would have been a trifle (and may have exposed some interesting design techniques at the same time).
Well, at least noise is low, compared to options with more conventional arrangements. The single op-amp circuit based on an early 1980's japanese design I showed earlier, when used at -20db overall gain with a 10K volume controls and 10K tone controls is going to be -104dBu output noise. Meanwhile DS's design offers -115dB, almost precisely the difference in noise between 10K vs. 1K...
My best noise "fun" line stage had a 4nV|/Hz noise Op-Amp and switched shunt output attenuator/lower feedback resistor. The Op-Amp was a LM6181 plus a J-Fet input buffer. At -20dB gain it's noise was basically that of a resistor of around 3 Ohm.
But that was a bit extreme and lacked tone controls... The noise was incidental to the design and not a goal BTW. As some may have seen in my comments, I find extremely low noise overrated.
What is the point of noise levels 30-40dB below the audibility threshold, except the satisfaction of having a very nice looking number? It would be nice to see those who promote such noiselevels to present actual evidence that they are of use.
Ciao T
My German issue also turned up. As I meant to express and expected in a proir post, the distortion data for the tone control in extreme positions is not given (my previous note missed a NOT in there).
It's been in Burr Brown datasheets for decades, meanwhile parallel Op-Amp's have been in LT datasheets for similar ages, at a minimal exercise in memory...
What I find amusing is that the positive polarity of the output is driven by four parallel OPA's, yet the other polarity is driven by only one and the inputs and outputs are pseudo balanced, instead of true balanced connections (this design of course is not alone in this, it seems par for the course in "HiFi").
I am sure there are good reasons for this, but it is disappointing, given that there are already so many parts in use, that to add the remaining few to get true balanced in's and out's would have been a trifle (and may have exposed some interesting design techniques at the same time).
Well, at least noise is low, compared to options with more conventional arrangements. The single op-amp circuit based on an early 1980's japanese design I showed earlier, when used at -20db overall gain with a 10K volume controls and 10K tone controls is going to be -104dBu output noise. Meanwhile DS's design offers -115dB, almost precisely the difference in noise between 10K vs. 1K...
My best noise "fun" line stage had a 4nV|/Hz noise Op-Amp and switched shunt output attenuator/lower feedback resistor. The Op-Amp was a LM6181 plus a J-Fet input buffer. At -20dB gain it's noise was basically that of a resistor of around 3 Ohm.
But that was a bit extreme and lacked tone controls... The noise was incidental to the design and not a goal BTW. As some may have seen in my comments, I find extremely low noise overrated.
What is the point of noise levels 30-40dB below the audibility threshold, except the satisfaction of having a very nice looking number? It would be nice to see those who promote such noiselevels to present actual evidence that they are of use.
Ciao T
I agree the noise thing like THD becomes the typical pee-peeing contest. What JFET has .22nV noise, or maybe I'm not understanding the configuration?
But almost never seen in full detailed projects as this one.
The purpose of these four OPAs is to correctly drive the volume pot which presents a load with a much lower impedance than almost any following device.
Hot and cold outputs having equal impedance, here of 47 Ohm, typify the balanced connection.
According to you, what circuits present "true" balanced outputs ?
Hi,
At -20dB gain the OPA was operating unity gain, fet buffer (2SK170/2SJ74) on non-inverting input 20dB attenuation after the active circuit.
Maybe not 0.22nV|/Hz, but very low (did a rough estimation), calculating in more detail it is around 0.67nV|/Hz for input short, so around 30 Ohm equivalent noise, not 3. Looks like I lost a zero somewhere... Still, it would be -138dB in theory.
The trick is to start out with low noise and to then attenuate that with the signal...
Ciao T
At -20dB gain the OPA was operating unity gain, fet buffer (2SK170/2SJ74) on non-inverting input 20dB attenuation after the active circuit.
Maybe not 0.22nV|/Hz, but very low (did a rough estimation), calculating in more detail it is around 0.67nV|/Hz for input short, so around 30 Ohm equivalent noise, not 3. Looks like I lost a zero somewhere... Still, it would be -138dB in theory.
The trick is to start out with low noise and to then attenuate that with the signal...
Ciao T
I agree that extremely low noise is over-rated. This is especially true if achieving it is possibly at the expense of other performance that may affect audibility in subtle ways. If the pursuit of low noise causes one to use much lower impedance levels, making op amp output stages work harder, that may not be a good thing. If achieving very low noise causes one to choose a BJT op amp instead of a JFET op amp where the latter may have better sonics (or the very high input impedance is helpful), that may not be a good thing.
It is much akin to THD. Really low THD is very nice, and everyone knows I like to see it, BUT if one compromizes other aspects of the design to get it below, say, 0.001%, that is just silly. The guy who uses too-lite FB compensation to get THD down by another factor of 2, and compromizes stability, has not got his priorities right.
Good engineering is a wise balancing act. Single-spec optimization is no better than single-issue voters in an election.
Cheers,
Bob
Hi,
FWIW, here roughly the schematic of this experiment, ca. 1997 or so IIRC...
This was in my "Batteries and superfast circuits" time while I did my 2nd degree at UNL and worked as repair man (they called it engineer) for london chain of 2nd HiFi/Pro-Audio/Records/Books/Clothes shop chain based mostly around Notting Hill Gate and was even poorer than I am now (and drank more Guiness than I do now).
Nat Semi had a most generous sample program and Sandy at Cricklewood used kind of like me and sell me stuff below cost...
The circuit was rather twitchy and tended to oscillate if the cap across the rails had the wrong value. I think I had a trimmer in the FET sources for DC.
It had a MC Phonostage using shunt feedback EQ and current input with AD811 and LM6181 which bypassed the J-Fet buffer and 8 pcs of 2nd hand alarm system SLC Batteries (6V/2AH or such) as supply.
The two pots was actually a single 32-position GEC-Plessy silver contact switch (which I bought for an outrageously expensive 5 Quid per deck at the Tonbridge Wells Audio Jumble) with discrete resistors (including the then fashionable vishays - I changed to Rohpoint precision wireounds soon) so that if gain was needed they switched the right resistors to ground from -in and the output was unloaded while for below unity operation the right resistors where switched to ground from the output.
It did sound quite interesting. It was kind of my active answer to the then very popular (in England anyway) Audio Synthesis "Passion" passive preamp.
All ancient history.
Attempts at building this should be limited to anyone who has a> 100MHz scope and experience in spotting oscillation and related issues.
The circuit needs to be constructed 3D with the supply regulator Op-Amp's basically above the signal Op-Amp with minimal loop areas. The cap is a low pF silver mica IIRC, the whole active circuit occupies no more volume then the size of stamp squared with MEL resistors for feedback soldered to the DIP package pins.
These days I'd probably just do something with tubes and transformers...
Ciao T
FWIW, here roughly the schematic of this experiment, ca. 1997 or so IIRC...
This was in my "Batteries and superfast circuits" time while I did my 2nd degree at UNL and worked as repair man (they called it engineer) for london chain of 2nd HiFi/Pro-Audio/Records/Books/Clothes shop chain based mostly around Notting Hill Gate and was even poorer than I am now (and drank more Guiness than I do now).
Nat Semi had a most generous sample program and Sandy at Cricklewood used kind of like me and sell me stuff below cost...
The circuit was rather twitchy and tended to oscillate if the cap across the rails had the wrong value. I think I had a trimmer in the FET sources for DC.
It had a MC Phonostage using shunt feedback EQ and current input with AD811 and LM6181 which bypassed the J-Fet buffer and 8 pcs of 2nd hand alarm system SLC Batteries (6V/2AH or such) as supply.
The two pots was actually a single 32-position GEC-Plessy silver contact switch (which I bought for an outrageously expensive 5 Quid per deck at the Tonbridge Wells Audio Jumble) with discrete resistors (including the then fashionable vishays - I changed to Rohpoint precision wireounds soon) so that if gain was needed they switched the right resistors to ground from -in and the output was unloaded while for below unity operation the right resistors where switched to ground from the output.
It did sound quite interesting. It was kind of my active answer to the then very popular (in England anyway) Audio Synthesis "Passion" passive preamp.
All ancient history.
Attempts at building this should be limited to anyone who has a> 100MHz scope and experience in spotting oscillation and related issues.
The circuit needs to be constructed 3D with the supply regulator Op-Amp's basically above the signal Op-Amp with minimal loop areas. The cap is a low pF silver mica IIRC, the whole active circuit occupies no more volume then the size of stamp squared with MEL resistors for feedback soldered to the DIP package pins.
These days I'd probably just do something with tubes and transformers...
Ciao T
