Hi.....took some freedom here....
I changed the current-source to be mirrored inside the cascodes and made it floating so the current s are always the same up an down.. I also used this mirrored current source as the point of entry for the servo, thus really shielding the servo by the impedance of the mirror +500Kohms from the signal-path.
I do believe that a THAT340 could be a very good choice for the input/+ the two basis/emitter drops in the floating current source, this way thermal drifts will be at a minimum. and a second THAT 340 for the additional input transistors. The THAT is ultra low noise at the level of the best fets (Voltage Noise 0.75 nV/ √Hz (PNP)0.8 nV/ √Hz (NPN), and they can be run on low current here 2.5 mA/each, for reduction of thermal/current noise...
http://www.thatcorp.com/datashts/THAT_300-Series_Datasheet.pdf
I have unbalanced the mirror to get more current on the backside for maintaining the Gain
I know I have deviated from the concept of fully balanced, but this was 1. take and can be built into fully balanced as well.
Distortion is very low, and I do think that it looks quite promising...
I changed the current-source to be mirrored inside the cascodes and made it floating so the current s are always the same up an down.. I also used this mirrored current source as the point of entry for the servo, thus really shielding the servo by the impedance of the mirror +500Kohms from the signal-path.
I do believe that a THAT340 could be a very good choice for the input/+ the two basis/emitter drops in the floating current source, this way thermal drifts will be at a minimum. and a second THAT 340 for the additional input transistors. The THAT is ultra low noise at the level of the best fets (Voltage Noise 0.75 nV/ √Hz (PNP)0.8 nV/ √Hz (NPN), and they can be run on low current here 2.5 mA/each, for reduction of thermal/current noise...
http://www.thatcorp.com/datashts/THAT_300-Series_Datasheet.pdf
I have unbalanced the mirror to get more current on the backside for maintaining the Gain
I know I have deviated from the concept of fully balanced, but this was 1. take and can be built into fully balanced as well.
Distortion is very low, and I do think that it looks quite promising...
Attachments
MiiB, cool circuit. The distortion profile is amazing. Have you made noise calculations with the BCs ? Lets take that as the unbalanced version. I would prefer a BJT buffer though. Just because there should not be any Fets in there. I have an idea for a diamond buffer with high input impedance. I worry a bit about the quite low and dissimilar Hfe of the Thats chip but honestly i have not tried them so you are the expert there.
How would a balanced version look ?
How would a balanced version look ?
No here its simulated with the 550/560C transitors, so the gain is high...60 dB...i have not calsulated the possible voltage swing in the base region, may run out of current headroom there...app 20 mA through the backside of the mirror..
quite sure the zeners could be substituted with something better, maybe even an internal shunt supply
quite sure the zeners could be substituted with something better, maybe even an internal shunt supply
Last edited:
Here is where the BJts beat the Fets. They have around 10x more transconductance against Fets at the same bias current. We end up with less distortion. Also they like to be degeneted to aprox. 2/3 gain when they show less distortion. A degenerated Fet will have a complex spectrum though. What do you think about a Tl431 instead of the zener. That would be cheep and simple.
I think it's vital to get the noise down...impedance is not so important as the main focus is to maintain an absolute steady DC at +-15 V,
so we need do increase degeneration and maybe increase the current at the backside, i think that noise will be low, especially if we select some low noise transistors (where is my stash of R-ohm) or could also be the 2240/970 pair, but then we migrate into more costly BOM
I really like this exploration of One-shot no feedback current riaa's..There's now quite a collection, from yours sen/cen inspired over this to my funny-folded.. guess you may be the one with the best overview to the sonic qualities... Not only can tha distortion be low, but the way disotrtion envolves over tha frequencies is also unigue....Can't see that there's any stronger concept...
so we need do increase degeneration and maybe increase the current at the backside, i think that noise will be low, especially if we select some low noise transistors (where is my stash of R-ohm) or could also be the 2240/970 pair, but then we migrate into more costly BOM
I really like this exploration of One-shot no feedback current riaa's..There's now quite a collection, from yours sen/cen inspired over this to my funny-folded.. guess you may be the one with the best overview to the sonic qualities...
Not only can tha distortion be low, but the way disotrtion envolves over tha frequencies is also unigue....Can't see that there's any stronger concept...
Last edited:
Yes, it works very well AND the signal path is short. I recommend the 2SA1015, 2SC1815 combo. They cost 4 cent a piece at Reichelt and seem abundant ( big secret, but i already published that on my Nobrainer thread ). They are superbly quiet although the Rbe`is 40 Ohm. They have low flicker and 1/F noise two. Just a pair in the Nobrainer is so quiet that the guys at our pre Frickelfest meeting jumped up the seats and went to the speakers after we had switched from an tube MC stage. They thought that the music was coming from nowhere and the phono stage was off. Ask Sam or Benedetto, they where there. OK, the Nobariner is common base so the base current is shunted to ground but in theory common base and common emitter are similar in noise in first order approximation.
Here is also my first shot at a high impedance BJt buffer.
MiiB, do you have an idea for a better voltage stabilizer then a Zener ?
Here is also my first shot at a high impedance BJt buffer.
MiiB, do you have an idea for a better voltage stabilizer then a Zener ?
I'm not so sure if that is a good idea, since the fc-Q of opposite polarity acting as current sources to each other, and we would introduce some current fighting, don't we?
Rüdiger
Sziklai-Paar ? Wikipedia
honestly, i do not know but i have seen buffers with that input stage.
honestly, i do not know but i have seen buffers with that input stage.
No not really a significant better, could be a string of red leds will be better..., but with the current source and and the fixed rail only attached to the cascodes...I'am not sure how significant it is to get really lower in noise..than the 15V zener+caps
Nice with the low-noise transistors, are they still in production..??
I'll model the buffer in.. is that the one you use behind the toaster..??
Nice with the low-noise transistors, are they still in production..??
I'll model the buffer in.. is that the one you use behind the toaster..??
Something like that, still in development but sounds already great ( the secret buffer ), although the version i use is not optimized for very high input impedance. Maybe we even do not need MOHms because the RIAA is pleasent low in resistance.
No, sorry, that transitors are also Toshiba and not made any more. Still they are much more abundant then the Fets. I have prepared a list of substitutes, just in case. I really looked deep into that glass.
We could RC filter the zeners but i think Rüdiger does not like that.
No, sorry, that transitors are also Toshiba and not made any more. Still they are much more abundant then the Fets. I have prepared a list of substitutes, just in case. I really looked deep into that glass.
We could RC filter the zeners but i think Rüdiger does not like that.
for my simulation i have used 100K input impedance.. I think it's useful for the buffer...
I do like this bridge type of circuits, but I don't like balanced cable, as the sending end needs two independent transmitters.. you double the problems...
Could be that the input from the coil can be run in balanced mode, but there you also run into problem as the MC-does not have center tap....so It'll have to be floating with the shield and arm grounded
I do like this bridge type of circuits, but I don't like balanced cable, as the sending end needs two independent transmitters.. you double the problems...
Could be that the input from the coil can be run in balanced mode, but there you also run into problem as the MC-does not have center tap....so It'll have to be floating with the shield and arm grounded
The champions, really low Rbe`, around 2 to 6 Ohm :
Hitachi 2SC2547, 2SA1085
Toshiba 2SC3329, 2SA1316
Rohm 2SD786, 2SB737
Sanyo 2SC3504, 2SA1433
Sanyo 2SC3601, 2SA1407
The middle ground, Rbe` 10 - 40 Ohm:
Sanyo 2SC3503, 2SA 1381 Fairchild KSC3503, KSA1381
Toshiba 2SA1015, 2SC1815
Toshiba 2SA970, 2SC2240
2N4401, 2N4403
Zetex ZXT790, ZXT690
On Semi MJE340, MJE350
On Semi MJE15030, MJE15031
JLH BD437, BD438
Beis BF469, BF471
Beis BF471, BF472
Various MAT, SSM and Thats arrays
Usefull if paralleled :
BC550,560 BC850, 860
Oddballs
BC140/160 can be selected for surprising low noise ( Stan Curtis, Grundig )
Secret Weapon when the world breaks down:
Zetex FZT653, FZT753 medium power SMD, easy to solder, in production, ultra low noise although not specified that way
there is hope
Hitachi 2SC2547, 2SA1085
Toshiba 2SC3329, 2SA1316
Rohm 2SD786, 2SB737
Sanyo 2SC3504, 2SA1433
Sanyo 2SC3601, 2SA1407
The middle ground, Rbe` 10 - 40 Ohm:
Sanyo 2SC3503, 2SA 1381 Fairchild KSC3503, KSA1381
Toshiba 2SA1015, 2SC1815
Toshiba 2SA970, 2SC2240
2N4401, 2N4403
Zetex ZXT790, ZXT690
On Semi MJE340, MJE350
On Semi MJE15030, MJE15031
JLH BD437, BD438
Beis BF469, BF471
Beis BF471, BF472
Various MAT, SSM and Thats arrays
Usefull if paralleled :
BC550,560 BC850, 860
Oddballs
BC140/160 can be selected for surprising low noise ( Stan Curtis, Grundig )
Secret Weapon when the world breaks down:
Zetex FZT653, FZT753 medium power SMD, easy to solder, in production, ultra low noise although not specified that way
there is hope
MiiB, i have a lot of experience with the Sueeskind Rauschfrei based on the FPS. It sounds better in balanced mode ( more open, transparent, dynamic, you name it) and there is absolutely no hum, even in rough environment.
That may be an artifact of that circuit. OK, we skip the balanced. Maybe Rüdiger comes up with something balanced anyway. Lets focus on what we have. When its satisfactory done lets put in on a PCB.
That may be an artifact of that circuit. OK, we skip the balanced. Maybe Rüdiger comes up with something balanced anyway. Lets focus on what we have. When its satisfactory done lets put in on a PCB.
I used to have a PREPRE with the Rohm (2SD786, 2SB737) transistors, that was compered to most extremely see through. and also very expansive and dynamic, compared to stuff like the Borberly and Threshold JFET products, sure if we find the right transistors this will both be very quiet and also very musical and expressively dynamic...