Konnichiwa,
Well, Holger Barske mentioned this little critter so I had a look. Hey, this looks right on tasty.
There we have a fully differential Op-Amp which not only has spectaculary low distortion, but also manages a fairly low noise (1.3nV|/Hz @ 10KHz, though sadly up to 3nV|/Hz @ 100Hz) and best (but in some ways also problematic) does not get low distortion by tons of negative feedback, having a DC gain of only 66db minimum and 78db typhical. Even better, unlike most such parts which seem limited +/-5V supplies this one does +/-15V supplies.
The unity gain bandwidth is 180MHz and scales well with gain, so at 20db gain we have 18MHz, at 40db Gain 1.8MHz, at 60db Gain 180KHz, it seems that around there is the open loop gain corner, it is not too clearly specified, but we know with a DC gain of 78db we will have most definitly an open loop bandwidth of somewhere between 18KHz and 180KHz, which is quite frankly smashing, as most Op-Amp's have open loop bandwidth closer to the 10...100Hz region!!!
So, here is a "Op-Amp" that seems predestined for a Phonostage with passive EQ, because it's open loop bandwidth is actually reasonably sensible. And thanks to the differential output it can feed a "floating balanced" RIAA Circuit with nearly 20V RMS when running on 15V supplies, which is a good few db more than common Op-Amp circuits.
Really interresting is the simplified internal structure:
(from:http://focus.ti.com/lit/an/sloa054d/sloa054d.pdf)
So, we have a simple folded cascode and a pair of output buffers (likely diamond transistors) and a DC servo (Candeias LEF anyone?).
The only annoying thing (which in light of the the intended use is understandable) is that BB have not (seemingly) used emitter degeration in the input stage and equally made the output nodes of the folded cascode available, which would have allowed the use of the OPA1632 with adjusted gain open loop.
Still, what we have here looks like a seriosuly tasty piece of analogue silicon.
If we have the output stage working at 30db gain, preceeded by the RIAA (passive) we need the input stage to have another 30db gain at Moving Magnet levels and at around 50db for normal output moving coil cartridges (60db gain at 1KHz), with an option to bump up the gain of the output stage to also as much as 50db for an 80db @1KHz gain for extremely low output MC's.
The input stage can be operated single ended in like a normal Op-Amp with an extra opposite polarity output:
(from:http://focus.ti.com/lit/an/sloa054d/sloa054d.pdf)
This allows us a wide range of gain and easy compatibility with normal tonearms and the option to use moving magnet cartridges (the input current noise is low enough) and the ability to kick gain up a gear for MC's.
In addition we can actually make the feedback loop balanced and connect our low output MC directly, balanced and in "current mode" very much like the AQVOX Phonostage. This could be easily switchable as well....
Another little problem is how to make the input impedance of the second stage, which would have balanced feedback, high enough (we do have to watch noise there though) to not require completely ridiculous values of RIAA Capacitors. Using a balanced T-Network can help:
If we make the kind of RIAA Krishu.de shows in the spreadsheet here:
http://old.krishu.de/down/riaa2003.xls
but re-arranged for a floating balanced arrangement where the termination resistor R4 is split into two identical resistors that form the input resistors in the circuit for the second stage above and equally the input series resistor is split we end up with this:
If we make C1 = 220nF then C2 is "close enough" with 3 X 220nF in parallel (theoretically 642nF), R3 becomes theoretically 495R, 510R is closest and if we make R2 = 2 X 2k2 then R4 becomes 2 X 7K5 with the second pair of resistors in the feedback network also 7K5.
Then we have in effect a 7K5 differential impedance for noise purposes on the input of our second stage giving an input noise of 1.5uV over a 20KHz bandwidth which is slightly more than what we get from the conversion of the current noise into voltage noise in the impedance.
With 30db 2nd stage gain and 400mV nominal output we get around 50uV output noise which is 78db below the 400mV nominal output (in other words quiet enough).
So, we get something like this:
This should be quite cute and interesting. I'll have to stick this job into TI's TINA simulator and see what it ends up as and then build it....
Sayonara
Well, Holger Barske mentioned this little critter so I had a look. Hey, this looks right on tasty.
There we have a fully differential Op-Amp which not only has spectaculary low distortion, but also manages a fairly low noise (1.3nV|/Hz @ 10KHz, though sadly up to 3nV|/Hz @ 100Hz) and best (but in some ways also problematic) does not get low distortion by tons of negative feedback, having a DC gain of only 66db minimum and 78db typhical. Even better, unlike most such parts which seem limited +/-5V supplies this one does +/-15V supplies.
The unity gain bandwidth is 180MHz and scales well with gain, so at 20db gain we have 18MHz, at 40db Gain 1.8MHz, at 60db Gain 180KHz, it seems that around there is the open loop gain corner, it is not too clearly specified, but we know with a DC gain of 78db we will have most definitly an open loop bandwidth of somewhere between 18KHz and 180KHz, which is quite frankly smashing, as most Op-Amp's have open loop bandwidth closer to the 10...100Hz region!!!
So, here is a "Op-Amp" that seems predestined for a Phonostage with passive EQ, because it's open loop bandwidth is actually reasonably sensible. And thanks to the differential output it can feed a "floating balanced" RIAA Circuit with nearly 20V RMS when running on 15V supplies, which is a good few db more than common Op-Amp circuits.
Really interresting is the simplified internal structure:
An externally hosted image should be here but it was not working when we last tested it.
(from:http://focus.ti.com/lit/an/sloa054d/sloa054d.pdf)
So, we have a simple folded cascode and a pair of output buffers (likely diamond transistors) and a DC servo (Candeias LEF anyone?).
The only annoying thing (which in light of the the intended use is understandable) is that BB have not (seemingly) used emitter degeration in the input stage and equally made the output nodes of the folded cascode available, which would have allowed the use of the OPA1632 with adjusted gain open loop.
Still, what we have here looks like a seriosuly tasty piece of analogue silicon.
If we have the output stage working at 30db gain, preceeded by the RIAA (passive) we need the input stage to have another 30db gain at Moving Magnet levels and at around 50db for normal output moving coil cartridges (60db gain at 1KHz), with an option to bump up the gain of the output stage to also as much as 50db for an 80db @1KHz gain for extremely low output MC's.
The input stage can be operated single ended in like a normal Op-Amp with an extra opposite polarity output:
An externally hosted image should be here but it was not working when we last tested it.
(from:http://focus.ti.com/lit/an/sloa054d/sloa054d.pdf)
This allows us a wide range of gain and easy compatibility with normal tonearms and the option to use moving magnet cartridges (the input current noise is low enough) and the ability to kick gain up a gear for MC's.
In addition we can actually make the feedback loop balanced and connect our low output MC directly, balanced and in "current mode" very much like the AQVOX Phonostage. This could be easily switchable as well....
Another little problem is how to make the input impedance of the second stage, which would have balanced feedback, high enough (we do have to watch noise there though) to not require completely ridiculous values of RIAA Capacitors. Using a balanced T-Network can help:
An externally hosted image should be here but it was not working when we last tested it.
If we make the kind of RIAA Krishu.de shows in the spreadsheet here:
http://old.krishu.de/down/riaa2003.xls
but re-arranged for a floating balanced arrangement where the termination resistor R4 is split into two identical resistors that form the input resistors in the circuit for the second stage above and equally the input series resistor is split we end up with this:
An externally hosted image should be here but it was not working when we last tested it.
If we make C1 = 220nF then C2 is "close enough" with 3 X 220nF in parallel (theoretically 642nF), R3 becomes theoretically 495R, 510R is closest and if we make R2 = 2 X 2k2 then R4 becomes 2 X 7K5 with the second pair of resistors in the feedback network also 7K5.
Then we have in effect a 7K5 differential impedance for noise purposes on the input of our second stage giving an input noise of 1.5uV over a 20KHz bandwidth which is slightly more than what we get from the conversion of the current noise into voltage noise in the impedance.
With 30db 2nd stage gain and 400mV nominal output we get around 50uV output noise which is 78db below the 400mV nominal output (in other words quiet enough).
So, we get something like this:
An externally hosted image should be here but it was not working when we last tested it.
This should be quite cute and interesting. I'll have to stick this job into TI's TINA simulator and see what it ends up as and then build it....
Sayonara
Interesting amp, but I'd not recommend it as a phono preamp. The 1/f noise corner is at around 10kHz which means that noise in the audio band is going to be much higher than other ICs (i.e., op-amps) you'd want to consider.
I would also think that the input bias current is on the high side for DC coupling to a phono cartridge.
It would have a number of applications for line-level signals.
I would also think that the input bias current is on the high side for DC coupling to a phono cartridge.
It would have a number of applications for line-level signals.
Konnichiwa,
I noted this, however it is not that bad.
The corner is 1Khz, not 10KHz. At 100Hz noise is 3nV|/Hz and 20Hz is 6nV|/ Hz. That is better than the NE5534 which often enough features as MC Headheadamp in commercial gear.
It's not too bad. I was, however comming to that, for MC it's fine, again, I have seen similar situations for MM, though I would not normally like to.
The Phonostage is far from finished, for now I was looking at first principles.
Sayonara
BrianL said:
I noted this, however it is not that bad.
The corner is 1Khz, not 10KHz. At 100Hz noise is 3nV|/Hz and 20Hz is 6nV|/ Hz. That is better than the NE5534 which often enough features as MC Headheadamp in commercial gear.
BrianL said:
It's not too bad. I was, however comming to that, for MC it's fine, again, I have seen similar situations for MM, though I would not normally like to.
The Phonostage is far from finished, for now I was looking at first principles.
Sayonara
Very nice Thorsten. I've been on the same route as you are, but not with the T-network for the second amp. I was on the "ridiculous cap value" side, but your approach is of course far more elegant. I'll be the first one to do a prototype when the design is getting further. (4 opa1632 already around, waiting for solder 
)
)
Konnichiwa,
Thank you. The next step of course is to use a a Steven & Billington LCR RIAA Module as EQ, then the input resistors for the second stage become 300R each and with 30db gain feedback resistors become 10K, which is still very tolerable. This gives BTW 30db gain to the differential outout, if a single ended output is desired the feedback resistors need to be 20K, no T-Network jiggery pokery needed.
For the frontend also 10K/300R for the MM input in SE (giving around 40db overall gain) and 10K/30R for the MC input in SE (giving around 60db overall gain).
BTW, if you like, get in touch with S&B, as Journalist and DIY'er you miy very well be able to get discounted sampled, also of the transformer volume control - might make a nice set of article for K+T.
For current mode inputs, if using an "average" MC cartridge we get around 0.03mV/R, or 0.3mV/10R so to get 50db out of that we need to have symmetrical feedback resistors of 16K each.
If we make them switchable to 51K/16K/5K1 we should be able to cover the range of Pickups from such as the very low output, respective to the impedance Dynavector Karat 17D (0.25mV@38R = 0.006mV/R) to the high output for the impedance Lyra Helikon(0.5mV@5R = 0.1mV/R).
Looks to me like a 6-Position 4-Pole rotary switch would do good, SE 40db; 50db; 60db and CI (Current injection) 0.01mV/R; 0.03mV/R; 0.1mV/R, then connect the SE Jack permanently, also an RCA jack connected pseudo symmetric and finally add a true XLR (or mini xlr) input for a true balanced connection.
I also have four, my ones are already soldered onto browndog adapters....
I only have to decise if I want to use up my spare of S&B RIAA Modules for this or if I only do an RC equalised version....
BTW, I suspect a servo will be also needed to null out differential offset, simply feed it into one of the inputs of the output stage via a sitable RCR filter....
Sayonara
HBarske said:
Thank you. The next step of course is to use a a Steven & Billington LCR RIAA Module as EQ, then the input resistors for the second stage become 300R each and with 30db gain feedback resistors become 10K, which is still very tolerable. This gives BTW 30db gain to the differential outout, if a single ended output is desired the feedback resistors need to be 20K, no T-Network jiggery pokery needed.
For the frontend also 10K/300R for the MM input in SE (giving around 40db overall gain) and 10K/30R for the MC input in SE (giving around 60db overall gain).
BTW, if you like, get in touch with S&B, as Journalist and DIY'er you miy very well be able to get discounted sampled, also of the transformer volume control - might make a nice set of article for K+T.
For current mode inputs, if using an "average" MC cartridge we get around 0.03mV/R, or 0.3mV/10R so to get 50db out of that we need to have symmetrical feedback resistors of 16K each.
If we make them switchable to 51K/16K/5K1 we should be able to cover the range of Pickups from such as the very low output, respective to the impedance Dynavector Karat 17D (0.25mV@38R = 0.006mV/R) to the high output for the impedance Lyra Helikon(0.5mV@5R = 0.1mV/R).
Looks to me like a 6-Position 4-Pole rotary switch would do good, SE 40db; 50db; 60db and CI (Current injection) 0.01mV/R; 0.03mV/R; 0.1mV/R, then connect the SE Jack permanently, also an RCA jack connected pseudo symmetric and finally add a true XLR (or mini xlr) input for a true balanced connection.
HBarske said:I'll be the first one to do a prototype when the design is getting further. (4 opa1632 already around, waiting for solder
I also have four, my ones are already soldered onto browndog adapters....
I only have to decise if I want to use up my spare of S&B RIAA Modules for this or if I only do an RC equalised version....
BTW, I suspect a servo will be also needed to null out differential offset, simply feed it into one of the inputs of the output stage via a sitable RCR filter....
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.