One could also not use that resistor at all. The cartridge would then pass the minuscule gate current to the sources and the Fets would then conduct.
I could be missing something, but doesn't R12 provide the cartridge load? Would it make sense to retain R12 but place it so that it is connected to the gates and to the node between R112 and R155?
Yes, that is what i just wanted to post ! This is then a floating load. Without the resistor the cartridge is its own floating load, terminated by the low impedance of the negative input, weird. Anyway, a floating resistor would make sure that the Fets conduct, even when the cart is not connected.
This looks good Joachim, but I'm afraid I already tried this arrangement and if my memory didn't froze over the day from this weather we are having here there was something fundamentally wrong with it. If I remember correctly fets sources were just swinging in phase with a source without any reference to ground, but I may be confusing something here. Let me put it to sim and get back to you. Or even better, I can try it on real proto.
Ok, fine, i think it works as strange as it is. We can reference both input to ground though like in an opamp based differential stage.
ok, it doesn't work without cascodes in proto (LF oscilation modulated by input signal with gain on fet's sources, same thing as I tried some time ago), but if I'm getting this right now, cascodes IS the key. Time to build some. I'm assuming some low current K170 instead of "High Impedance Current Source" in your schematics would do for experiments ?
Sure, they can be casdoded with J310 and then we are in the several tens of megaohms range. The sources of the Fets have to drive the bases and the current source. The higher the impedance of the current source the more drive goes into the bases. For testing one low current 2SK170 is fine.
Mhm, now I got it, cascodes with symmetrical ccs'es are making reference for signal, it should work without any resistors to ground. They are kinda in signal path, but that can be taken care of by making extremely good ccs'es. Or even better, we can put them inside mirrors and replicate that MiiB's auto biasing part (that I really like). Then servoing it and voila - dc coupling. Damn, group thinking is such a powerful thing 🙂
One thing though, J74... i paid 40Eu for my last batch of non fake matched octet, and supply's are running out fast 
Ok, it works like a charm ! But I don't see any common mode cancellation 😕 Attached fft of 2 matched channels, red - balanced, light blue - se. Have to make some balanced measurement setup later, so this is only same cart channel connected to aether input. So much for today, I'm out 🙂
Attachments
Michael, i really would like to use Fets at the input this time. They have a certain charm that is really helpful here. Balanced with Fets will give an alternative in sound and performance to the Paradise. I have access to enough P-Channels and can organize a group buy if this design makes it on a PCB. This will be the last complete phono stage design i will support. I am simply running out of time. I think i presented enough food for thought and will keep this thread going. There are always new ideas and a lot of old problems can be discussed and solved. What i would like to do in the future besides posting here as often as my schedule allows is making a PSU for a turntable motor. I think it is best to open a new thread for this maybe in three month from now.
Good morning to you all.
I got a batch of the BC 32740 and BC33740 direct from Fairchild Direct
Farnell does not have the 40s and
Onsemy does 16 and 25
Multycomp is I believe a sourche from multiple alternative manufacturer copies
Frans posted a realy decent link to his suplier and I would suggest you look there first.
Any way
I will start matching soon.
The DCA55 Atlas component tester is ok apart from chewing bateries and having to wait and scroll buttons
I was wondering if using a set up as in the first picture could speed up things and be reliable for a preliminary sort of the batch.
This is the link of the web page.
Transistor matching
I got a batch of the BC 32740 and BC33740 direct from Fairchild Direct
Farnell does not have the 40s and
Onsemy does 16 and 25
Multycomp is I believe a sourche from multiple alternative manufacturer copies
Frans posted a realy decent link to his suplier and I would suggest you look there first.
Any way
I will start matching soon.
The DCA55 Atlas component tester is ok apart from chewing bateries and having to wait and scroll buttons
I was wondering if using a set up as in the first picture could speed up things and be reliable for a preliminary sort of the batch.
This is the link of the web page.
Transistor matching
