A new X preamp.Call all eXperts!!!

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
So here's the thing.I believe possible to build a new preamp for myself,using the X benefits.So here's my design options.Should be able to convert Unbal.inputs to balanced outs.I desire to use a 2 stage design:jfet 2sk389 inputs,in differential pair.Then cascoded by Jfets too_Output will definitely be mos,with high bias ( relative to preamp's function.) Gain must be variable,from 0dB to 20dB approx.Feedback will definitively be X type.

I believe this research is to get best sound possible,with low distortion,high output swing capabilities in low loads ( such 600 ohms,for example),very low noise indeed.

The power supply will be classical one,with high current bridge+capacitors,CRCRCRC pi filters and mosfet regulation.I will provide my ideas,but hell what do you think you guys of this?Would be so nice to fit a pair of XA-100...already studying.

Any feedback ? Go ahead.I kindda' like it :drink: :drink: :drink: (Thanks Nelson!)

Yeah, I've got an X preamp coming. Gotta do something to go with the Aleph-X, right?
There are several reasons I haven't put it out yet. Number one being that this has been one of the worst summers of my life, though hopefully some of the dust is settling. Two is that I intend to do a set of four variations: 6SN7, 6DJ8, JFET, and MOSFET. Builders can take their pick as to which suits their needs and/or prejudices. I've got the two tube versions pretty much ready. The solid state versions are lagging behind. I'll get there.
Reason number three is that I went and started this little thread about an amp and I wanted to kinda let that settle out before I stirred the mud any further.
Reason number four is that I got sidetracked into building a bunch of speakers. One pair of Ariels, two pairs of ME-2s, a modified pair of CS-1s (old KEF design), and three subs (maybe more before all is done...you can never have too many subs). Parts aren't all in yet, so I'm amusing myself by doing the woodworking. And believe me, Ariels take a bit of fiddle-factor to put together. ME-2s aren't that bad. CS-1s are just boxes, although I gave into the urge to piddle with the basic design. Two of the subs are transmission lines; the third is a bass reflex. Anyway, point being that I've been kinda busy.
Did I mention that I've also been writing? Oi! I need more hours in the day...
As far as topology goes, I see no need for cascodes in linestages. Yes, linearization, Miller, distortion, yackety-yack, and blah, blah, blah. Whatcha going to do with all that gain, eh? Burn it off as feedback? Not me. I've also got a neat little non-NFB distortion-buster trick that I'm using. And it's X, of course. And balanced. Etc.
Like I said, I'll get there. Just lemme get the sawdust swept up first...

nobody special said:
Why JFETS? You can (as i have) use a cascoded single-stage current-sourced differential pair of MOSFETs with X feedback.

I suspect the 2sk389s were picked for their low noise. If I recall correctly ("watakashi-no kyoku ga tashika naraba") the 2sk389 is a dual monolithic N-channel JFET, so you'll get the best match possible between your two amplifier halves.

You can adapt the X100 design work (separate thread) to work as a preamp. That would give you a "one stage" amp with relatively low output impedance.

Your jfet input will be inherently low current and thus you likely need an output buffer. Bipolar emitter follower would likely work better for you than FET follower especially if you are planning to have zero offset at balanced output ....

As far as topology goes, I see no need for cascodes in linestages. Yes, linearization, Miller, distortion, yackety-yack, and blah, blah, blah. Whatcha going to do with all that gain, eh? Burn it off as feedback? Not me

Yeah, I agree... It was more for the point of doing it. My plan is to get as many variations of the balanced zen line stage together as possible, and make PCB's that will plug into the same chassis, and test each subjectively. The planned topologies are:

1. Completely standard BLS-
2. Current sourced BLS- otherwise standard
3. Current sourced BLS with 2X+ bias
4. Current sourced BLS with X feedback
5. Current sourced / cascoded BLS
6. Current sourced / cascoded BLS with X feedback

This is to allow me to try all the "flavors". There has been so much talk here in the past as to the sound of cascodes, minimalist vs. active current sources, etc. that i wanted to try it all out for myself. I first must finish my Aleph X's. I hope to publish what I find to a website eventually. Unfortunately, I don't have access to a distortion analyzer, so my testing will have to be on a subjective basis.
Which power,V gain and bias will you run your Aleph-X ?Just to have an idea...

I mostly think about a 2 stage version now,with 2sj109V input differential,cascode I think is not necessary,but would work well with an other Jfet,so convenient to use in fact (I just link its gate to its source pin,those connected to the 2 common sources pin of the differential.I think about something like 2sk246,or equivalent if I decide to go for a N version differential.)

I intend to use a V version for the 2sk389 or 2sj109,with higher bias possible(but not at idss anyway.Damn!!!...)

Output will be a 610 or 9610,with Voltage gain.Feedback will be adjustable,to allow for a 0 to 20 dB gain setting.


So the output will be mostly like an aleph design,but with no Aleph link and resistors as the input of power amp.is quite an easy load compared to a speaker.:cool:

I remember a note from Nelson saying that eliminating resistors to ground at the input would be nicer for a better unbal.to balanced conversion.So will have to work perfectly well with unbalanced sources,mostly all mine!! And yours too I guess...

I just thik about a mini XA for line levels in fact.But with no Aleph rear topology.I guess if i want no DC offset I should load outputs to ground.But I think 100 Ohms value will not do it.Nelson,should it be OK if the load resistors are 1000 times the value of output impedance of the preamp?

We should think about this value against input impedance of the power amp following,if we don't want to see any loss bandwith nor gain...

Sorry for long post!!!Any feedback GREATLY APPRECIATED guys :wchair: :wchair:
The one and only
Joined 2001
Paid Member
Uhhhhm, yeah.


About eliminating input resistors, the very best results
occur with matching impedances, which include the
single-ended source impedance, but it's a subtle thing.

As for resistors to ground at the outputs, they aren't
essential, and they don't particularly influence the
differential DC output, rather they help stabilize the
absolute DC value.

If you're using this for a line stage and the power amp
is gonna be DC coupled, I would put some nice film
caps before the power amp, otherwise your life will
just get more complicated.
I'm not really sure what I will end up with as far as bias goes. I am going to build it with 3 pairs of outputs per side, and then see what my rail voltages come out to be, and bias so that it will deliver as much current needed for the given voltage swing into 3 ohms. I may end up actually using more pairs per side, depending on the device dissipation. My speakers are 4 ohms, (minimum impedance of 3.2 ohms) so I am designing with a fairly low rail voltage and a lot of bias to accomodate them.
Not in the right thread but...

Nobody special,

I intend to run my XA 100 on 8 ohms speakers.So with +/-24V and 4A total bias it will do it without a problem.Even on 4 ohms in fact.

But I plan also to use it on a pair of small Westlake .Seems its impedance is very near 2 ohms.So I have to overbias a little the circuit to 5A total-2.5A per side-about 30W by mosfet.I have a large stock of 240s from I.R./Digikey.I think they will do it.

For the X preamp:in fact I have under my eyes the schematics for Aleph P preamp.version 610/9610,not 240.Seems it's an X circuit...isn't it? In that case I wonder why the resistor from drain to gate of input diff.pair.It's X of course,but why not have put it between real output of circuit/after film cap and gates?The X effect should have been extended this way I presume...

Oh,please note the resistors from drain to ground too :eek:
Why use jfets?

Well,many reasons.First of all,jfets are extremely low noise(see what I plan.a jfet current source for the diff.pair.),and perfectly matched(for diff.pair.I think about k389 or jXXX,2matched on the same substrate).

For those two reasons,I believe
that jfets are perfectly suitable
for X circuits,inherently low noise
produced and perfect matching
should give advantages for X-ing.
The X effect works well,and best if
the differential error correction is just
to be corrected itself,and not to compensate
other mismatches in the circuit.

The output stage will be mosfet,
with both voltage gain and current gain.
Feedback will be adjustable to permit
adjustement of the gain of the circuit itself.

And here we roll...

I have the Ariels and ME-2s for the video system running now, and will be heading back into the electronics end of things. The preamp I've been yakking about at intervals will be high on my list of things to do, as my venerable CJ Premier 3 doesn't understand the idea of balanced signals at all, and that's something I'd like to play with for a bit.
On the other hand, I've got to strip wallpaper and paint a room in the next few weeks, so that's going to eat some time. However, I think I'll be able to squeeze in a few hours late at night to work on the solid state versions (the tube versions being more-or-less ready to go--just want to work on one more detail).
Is anyone still interested in an X preamp in combination with a non-NFB distortion cancelling trick or should I let more dust gather on the prototypes?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.