folded-cascode FET/opamp MC phono

[ThorstenL]

Konnichiwa,

Originally posted by Pjotr
Since you are coupling the cascode output AC why use an inverted anyway? Just for PSU voltage reasons?

Actually, on the principle "where does current flow" this folded cascode is very neat.

In a nomral cascode the signal output is effectively referenced to to positive supply rail but the following stage commonly amplifies the differences between some usually ill-defined "ground" and the input terminal. Under ideal conditions the supply rail and ground are the same in AC terms, in reality they rarely are.

Now the folded cascode as Werner shows:

1) The BJT's base is referenced (AC) to the positive rail, as is it's emitter and it's input signal, namely the current in the J-Fet.

2) The collector current of the BJT is controlled by the J-Fet current and pretty much little else, it, just like the J-Fet operats as variable current source.

3) A resistor (be it a direct collector resistor as Werners original shows or a virtual resistor as in my Vendetta inspired idea) then converts this AC current into voltage which is referenced to the so-called ground and which is fairly well independent from the supply rail noise and modulation (with ideal BJT's, J-Fet's and capacitors it would completely unperturbed by anything on the supply rail).

Now I call that very smart and tricky....

It might be fun to do a folded cascode "Pacific"!

Sayonara

[Pjotr]

Originally posted by Kuei Yang Wang
Konnichiwa,



Actually, on the principle "where does current flow" this folded cascode is very neat.

In a nomral cascode the signal output is effectively referenced to to positive supply rail but the following stage commonly amplifies the differences between some usually ill-defined "ground" and the input terminal. Under ideal conditions the supply rail and ground are the same in AC terms, in reality they rarely are.

(...)

Sayonara

Hi Sayonara,

Ahhh… I see and R8 does not inject a bunch of PSU related noise into the emitter? Actually combined with R1 there is a gain of app. 2 concerning PSU related noise. Therefore C2 is needed to relax that, but it works only at higher frequencies.

Maybe replacing R8 by a current source will do better, but that needs to be very low moise then.

Cheers

[ThorstenL]

Konnichiwa,

Originally posted by Pjotr
Ahhh… I see and R8 does not inject a bunch of PSU related noise into the emitter?

No, as the controlling voltage appears between base and emitter and thus C2 clamps this...

Originally posted by Pjotr
Actually combined with R1 there is a gain of app. 2 concerning PSU related noise. Therefore C2 is needed to relax that, but it works only at higher frequencies.

Well, the turnover for this RC combo is 1.6Hz, so at 16Hz noise is clamped down to 20db less and so on. Increeasing the capacitor to 1,000uF is trivial and increases the supply rejection for any given frequency by 20db and increasing nthe resistor values in the divider for the base by a factor of 10 gives another 20db, so 60db supply line noise suppression at 16Hz above your case of -6db should be non too difficult to attain....

Sayonara

[Werner]

Originally posted by Kuei Yang Wang


Now I call that very smart and tricky....

Domo arigato. But as said, the credit goes to JC and that Kirchoff guy.

Too bad that the present prototype is housed together with its
multi-transformer PSU in one box. Baaad situation. I must rectify that. Which is kinda hard with my present corporate-life and daddy-life combined.

Originally posted by Pjotr

I see and R8 does not inject a bunch of PSU related noise into the emitter?

In fact the other part of the trick is that the original folded cascode allows the use of a simple emitter-follower regulator as supply (*), one per stage. This is of very low noise, and unperturbed by the actual signal as the folded cascode's current consumption is purely DC.
So the clamping capacitor is not even strictly required. Just nice to have.


[QUOTE]Originally posted by Pjotr
Maybe replacing R8 by a current source will do better, but that needs to be very low noise then.
[/QUOTE

The rationale for this front-end is the minimal use of active devices, so I shied away from folded cascodes for a while. I really needed JC to show me that it could be done with just two transistors, and then a resistor as 'current source'. So again, arigato gozaimasu, John-sama!





(* Fed from its very own secondary and LCRC filter, of course ;-)

[a333bt]

Originally posted by Kuei Yang Wang
Konnichiwa,


Edit, attached a quick job on Werners schematic, showing instead of the folded cascode a current mirror with an option to increase the current gain of the circuit above that of the J-Fet.... The input CTRL on the second FET should adjust the current for 0V on the output (servo).

The ratio between the resistor in the current mirror should probably be around 5, not 10 as shown, this would give around 150mA/V effective transconductance which with 750Ohm load gives around 41db gain, the OPA627 stage will then give 19db gain for a total gain of 60db overall.

Switching the resistors in the Current Mirror and in the source of the second (current source) FET would accomodate a gain adjustment for use with MM Pickups as well....

For those who has less experiences:

how to adjust CTRL - second FET for 0V?


regards, Bostjan

[Onvinyl]

Hi!
While waiting for parts for another project to arrive,
I finished a protoboard of this litte phono pre I had in mind since it was posted.
I found it has some remarkable features, both sonically and practically.

At first, I build it as shown in the first posting, and then with a servo. This made a substantial sonic improvement, allowing me to omit C3. It can reveal a lot of 'microresolution' on my vinyl now. I clearly like what I hear. Schematic as shown.

1) Technical discussion.
What might not be obvious at a first glance: this circuit falls nicely together on the protoboard, a nice layout is easily made.
It is very quiet and totally relaxed when it comes to hum. I placed it directly on top of my line stage, above a not so small transformer. Even well made boards fail this test and hum badly. Not with this circuit!

It is a headscratcher, though, where Q1 gets his current from, with the servo working, because the non-inverting input of U1 is at 0V now. The sim indicates, that U2 serves as current source for Q1. I freely admit that I didn't really get it. (It was a mistake, I forgot to tie Q1 to the negative supply) There is probably room for improvement.

Since I will probably rewire my tonearm for balanced signalrouting, I have a very simple mod for the input stage on the drawing board which allows balanced input with just one additional FET which simulates great as well.

I have the vague feeling, though, that Werner's simple input stage deserves something better (simpler?) than the currently following opamp stage.

2) Sound
It sounds good.


Lots of microdetail, maybe more than the two other pre's I currently have in the listening room. It's the typical FET-sound of relaxedness and resolution.

However, it lacks the final firmness in the bass region, which the other pre's have, more or less. This relative weakness was already present with C3 in place. It is not unpossible, though, that another servo arrangement would improve things here. A bigger C5 might be all whats needed.

Thanks for listening,
Rüdiger

[Onvinyl]

I reworked the servo. The trade now is, it affects the biasing of Q1. Ideas for the servo are welcomed.

I also changed C5, for a lower -3db point in the bass department

All in all, a nice sounding circuit!

Rüdiger

[Onvinyl]

Forget the servo arrangement. With this, the whole immuneness against hum is gone. This is no big surprise, because the current loop of the input stage is not small anymore (through R5 and back to the input), but goes right through the negative supply and back through the entire ground bus to the input stage.

But with the intermediate coupling cap, the sound isn't as transparent as it could be.
Back to complimentary design?

Rüdiger

[ThorstenL]

Hi,

Originally Posted by Onvinyl

Forget the servo arrangement. With this, the whole immuneness against hum is gone.

Why not use U1 inverting, with a feedback network of ten times the impedance and then use a 1M Resistor plus a 1uF (Wima?) cap to the non-inverting input, to cancel DC.

I used a similar circuit in a phono stage with a similar frontend and a valve section following and it works very well there.

Plus in your phono Stage it retains correctly the polarity, unlike now where the phono inverts.

Ciao T

[Onvinyl]

Originally Posted by ThorstenL

Hi,



Why not use U1 inverting, with a feedback network of ten times the impedance and then use a 1M Resistor plus a 1uF (Wima?) cap to the non-inverting input, to cancel DC.

I used a similar circuit in a phono stage with a similar frontend and a valve section following and it works very well there.

Plus in your phono Stage it retains correctly the polarity, unlike now where the phono inverts.

Ciao T

Hi Thorsten,
yes, this is sensible. Here is Werners last incarnation.
I will try it.
I don't get where the other end of the 1Meg resistor goes. One of the supplies?

Rüdiger