Fine, i will post a circuit that shows the principle later today.
It has a balanced input with BJts. I use 4 plus 4 in parallel so a lot of standard parts can be used, also BC550, BC560. Then there is a current mirror and a transimpedance RIAA, driven from both sides. The RIIA has only 40dB of amplification. That lowers distortion and does produce less problems with offset. The output is an INA that makes the rest of the amplification ( 20dB ) and has a servo.
It has a balanced input with BJts. I use 4 plus 4 in parallel so a lot of standard parts can be used, also BC550, BC560. Then there is a current mirror and a transimpedance RIAA, driven from both sides. The RIIA has only 40dB of amplification. That lowers distortion and does produce less problems with offset. The output is an INA that makes the rest of the amplification ( 20dB ) and has a servo.
Last edited:
Here is the basic circuit. At the input i use a Faulkner circuit with 4 plus 4 transistors. That did not fit on the diagram so i put it in a separate file. The BJTs are degenerated. That linearizes them and helps in current matching. No selection should be ok. With BC550,560 we get a noise impedance of 40 Ohm and with the japanese transistors it is 15 Ohm so the BCs are a bit more noisy, on par with an AD797 still. I am not going for absolute minimum noise here but i try a good combination of low noise and low distortion. Each transistor runs on 5.5mA, that is a good value for low distortion. More current would not bring a big benefit because this are no ultra low Rbe`devices. Due to the balanced nature we loose about 3dB of S/N for the same amount of input devices but we gain in hum suppression and suppression of common mode distortion, so this may be more important. Some values are not set yet, simulation will bring out the correct dimensioning. i would like to setup the RIAA for 40dB gain, that would bring 60dB in total. The gain set resistor at the INA could be substituted with a 500 Ohm resistor in series with a 1,5kOhm trimmer for example and then we can vary gain over a wide range from 54dB to 66dB aproximately. I would not run the input transistors on more then 10V because of avalanche noise that can set in at higher voltage. I have an idea for a PSU too that i already tested out with good result. The input transistors are supplied with low noise current sources and are shunted to ground with big electrolytics.
I use 10.000uF, 6.3V Panasonic FC ore two 4700uF Panasonic FM. The Ina is an INA217, setup for 20dB gain. It is a current feedback structure with distortion cancelation so speed is not affected much by gain and distortion has a flat spectrum up to over 20kHz.
It cost something over 5,-€ so this is the most expensive part but it contains 3 high quality Opamps and precision trimmed resistors so i see no reason to use a discrete or hybrid circuit. It will be more complex, more expensive and likely not better in performance. The INA217 comes in a convenient DIL8 package, so no fumbling around with SMD.
I use 10.000uF, 6.3V Panasonic FC ore two 4700uF Panasonic FM. The Ina is an INA217, setup for 20dB gain. It is a current feedback structure with distortion cancelation so speed is not affected much by gain and distortion has a flat spectrum up to over 20kHz.
It cost something over 5,-€ so this is the most expensive part but it contains 3 high quality Opamps and precision trimmed resistors so i see no reason to use a discrete or hybrid circuit. It will be more complex, more expensive and likely not better in performance. The INA217 comes in a convenient DIL8 package, so no fumbling around with SMD.
Sorry boys to bother you again but i think i have now an open loop transimpedance line stage that should work. After a lot of thought i think an emitter follower, though not standard, is the key if we do not want that the output stage is heavy into class a. There are of cause non switching arrangements possible but i am not an expert on that.
Available at UK Radiospares....[more than 500 available next day] delivery.
Texas......OPA134PAG4 ......£2.13 each
" ......OPA134PA...........£1.87 each
Here is a first attempt to dimension the circuit plus the PSU. I would like to make that double mono so two pre regulators in a separate box and each channel has it´s own on board regulator. I found that double mono sounds better. The PSU can be modified to supply both channels when funds are extremely limited.
Hi Joachim,
why no fets? Unobtainability?
I like the way the riaa is implemented, and cascoded 'current output' that isolates the current mirror Q's from R15's voltage swing.
I've tested shunts with LM317 CS and TL431 reg. as well and found them substantially worse than e.g. Salas discrete shunts both audibly and whats detectable on the scope.
Maybe someone likes to play with a dedicated regulator for the cascodes biasing (probably as shown here), I found it sounds better.
Maybe the quiet days find me at the bench again?
Rüdiger
why no fets? Unobtainability?
I like the way the riaa is implemented, and cascoded 'current output' that isolates the current mirror Q's from R15's voltage swing.
I've tested shunts with LM317 CS and TL431 reg. as well and found them substantially worse than e.g. Salas discrete shunts both audibly and whats detectable on the scope.
Maybe someone likes to play with a dedicated regulator for the cascodes biasing (probably as shown here), I found it sounds better.
Maybe the quiet days find me at the bench again?
Rüdiger