Some time ago HiFi Daddy published a little circuit that i found fascinating. It is a balanced version of the famous Hiarga Le Pre-Pre with an added output transistor. immediately i had some ideas how i could improve it ( hopefully ). One thing is to add an RC filter after the transdiode. That should lower noise by 3dB. The other idea is to cascode the output transistor because in the original design it amplifies the positive terminal in common base and the negative terminal in common emitter. That should result in different bandwidth for the positive and negative side. See the original circuit and my redesign ( not dimensioned yet ).
Attachments
Yes, Rüdiger, this is an interesting circuit and got virtually no attention. Sorry that i posted my version in this rudimentary form. I said " not dimensioned yet " but it is good that AS and you take care that i do not post too much nonsense. I whould be interested how your version of it whould look.
Making the resistors different is necessary anyway, because
R3, R6 have to supply current to T4, T10 etc. also.
Hi Joachim,
Thanks for your proposal. I found interesting aspects in your proposed circuit, but I also found details going into a direction not what I am aiming for:
(1) the mirroring might result in drifting idle current currents, while Vbe drifts (thermal distortion, Early voltage)
I know from your moving coil article in the Jan Didden book, that you prefer to have simple circuits mirrored. From my experience bass tightness and PRAT are negatively affected by drifting operating points, so I strictly exclude mirrored circuits from my design principles.
(2) the collector voltages of the input transistors are still different, with one transistor feeding the emitter, the other feeding the base of the second stage transistor. So through Early voltage, even with matched transistors like MAT or THAT there are is about 5..6mV Vbe difference between plus und minus inputs, but in your variant this would cancel out through mirroring, just translating into a current difference.
(3) with a second stage transistor driven from low impedances, I cannot see any benefit having a cascode in the output. In the 90ies, when I tried out cascodes, I learned they create new problems, and even when handled with care they are a mixed bag, soundwise.
In the meantime, I have put some other bipolars (THAT 300, MAT02) in the original circuit. I will post details later in my original thread. I could spend weeks testing transistors, but I have the feeling that I have to develop the circuit sooner or later, also.
kind regards,
Hartmut
By now you may know that i like to design parallel symmetric. It cancels second harmonic and can be made low noise. Most of my recent designs use J-Fets. Unfortunately P-channel low noise Fets got rare and expensive because they are not made any more. I seem to prefer degeneration in the input stage instead of NFB and single ended N-channel stages simpy have too much distortion for we when no feedback is used. J-Fets have other disadvantages. They have a rather high 1/f frequency and an unlinear input capacitance. Degeneration lowers distortion but gives a rather complex spectrum in J-Fets. Bipolars are an alternative with lower 1/f frequency and higher Gm. Degenerating them lowers distortion without added complexity of the spectrum. The major problem with bipolars is input offset. I have concentrated here on bipolar stages with common base input and that creates even a bigger offset problem in parallel symmetric stages because Ube is different in NPN and PNP transistors when used with the same bias current and the current flowing in the emitters is rather high. With the Nobrainer stage that will be published in Holger Barskes Klang & Ton i have shown a way to adjust the input offset with a trimmer. An alternative is to use the bases as input. The input offset current is reduced by the Hfe that way although there is the same offset voltage problem then in the comon base stage. This can be configured as a common emitter or emitter follower stage. I have shown a variety of the Essex Battery Stage here that uses common emitter input plus offset cancelation with a trimmer. The complication with the Essex stage is that it is based on several floating supplies and an even more complex battery management system.
So for a while i thought how a bipolar stage could be made that is simple and works well enough. In fact this time instead of doing something complex i made a stage that is so simple that i think it would not work the way i want it when i take anything away. The only sofistication is the cascoding that reduces distortion and makes the circuit faster. In fact it is rather fast. -3dB is at 6MHz with a gain of 26dB. The output clipps at 3V P-P and noise is 0.7nV/qHz. No world record this time but 3dB better then the usual suspect low noise opamps. Driven from my 50 Ohm Generator input offset is 0.3mV and 50uA. This will not fry the cartridge but in the proceedings i will show a way to trim that to minimum. There will also be a DC coupled version and noise can be brought down by 6dB with more BJT´s in parallel.
So for a while i thought how a bipolar stage could be made that is simple and works well enough. In fact this time instead of doing something complex i made a stage that is so simple that i think it would not work the way i want it when i take anything away. The only sofistication is the cascoding that reduces distortion and makes the circuit faster. In fact it is rather fast. -3dB is at 6MHz with a gain of 26dB. The output clipps at 3V P-P and noise is 0.7nV/qHz. No world record this time but 3dB better then the usual suspect low noise opamps. Driven from my 50 Ohm Generator input offset is 0.3mV and 50uA. This will not fry the cartridge but in the proceedings i will show a way to trim that to minimum. There will also be a DC coupled version and noise can be brought down by 6dB with more BJT´s in parallel.