Hybrid with Nelson topology

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

We all know Nelson Pass patented output topology, the so-called active current source.
It works wonderfully, I've already built few Pass clones (prototypes, actually) and every one is amazing.
I have to add I also love tubes; so, why not to add Nelson art with some tube "signature"?

My idea has been to "invert the output topology, and realize a voltage follower, driven by a tube input stage:

THIS

As you can see the active curernt source has been moved in the lower part of the output stage, but its behaviour should be the same.
The upper mosfet works as a voltage follower; the DC operating point is controlled by the OP07 chip, while the audio signal comes from the tube long tail pair in the input.
I have not yet made any calculation, so I do not know if the circuit works in the current version and if the components values are accurate.
I even don't know if the choosen tubes (the only one I have) have enough gain to drive the output stage to full power.
I show you the schematic just to know if this could become a working amp.

Ciao,
Giovanni
 
Official Court Jester
Joined 2003
Paid Member
cviller said:
Cool! :D

But the aleph output stage usually have quite a lot of gain, so you will be in trouble if you wire it as a follower. If you want n type frontend, the standard way to go is to use p type output fets.


in any way ......... double dutchie ........ (speaking about jaccovity ;) )

with follower config - input tube must fight with input capacitance of big IRF

with standard Aleph config ( output stage with gain ) problem is that output mosfet's input is referenced to PSU's plus ....... and having same
pSU rail for both tube and mosfet is tricky ......... and having different rails ...... even more

OK .........

tube input stage , with dedicated PSU + rail ....... neat cascoded current mirror/voltage shifter stage ...... regular Aleph output stage .....

but - is it worth ?

no , at least to my xperience .

jfets on input are exactly adequate tubes :rofl: for Aleph .

hehe

it's wise to save tubes for roles where sissy jfets are ..... sissy .

as always ..... YMMustV

:)
 
Even simpler: THIS version uses only one triode; I'm not sure if the feedback can actually work this way, never seen before.
I know the source could have some problem driving the input stage, but I don't think my CD should.

But the aleph output stage usually have quite a lot of gain, so you will be in trouble if you wire it as a follower. If you want n type frontend, the standard way to go is to use p type output fets.
Actually a voltage (source) follower stage has a gain between 0.6 and 1, so no problem at all with gain.
If you invert the position of the two mosfet in the Nelson topology, the output stage gain will drop below 1, according to the source follower phisics.

with follower config - input tube must fight with input capacitance of big IRF
Not exactly. Being a source follower (voltage follower) the driver has a much lighter load to drive; once loaded the gate does not need to be continuously biased (I hope this is clear).
The important thing is to DC bias the gate, thing done with the OP07.
Theoretically the input impedance of a voltage follower/source follower/common drain stage is infinite:
http://en.wikipedia.org/wiki/Common_drain

with standard Aleph config ( output stage with gain ) problem is that output mosfet's input is referenced to PSU's plus ....... and having same pSU rail for both tube and mosfet is tricky ......... and having different rails ...... even more
Right. This is the reason I added two diodes (D2, D3) in the signal path, to limit voltage spikes during startup.
Even more, I know I will have to add some timed relay to realize a precise startup sequence:
1. Power the input stage filament
2. Power the output stage, and wait until the OP07 DC bias it.
3. Power the input stage and wait
4. Connect the output to the loudspeakers

but - is it worth ?
Yes! Everything can give us some fun is worth the try, in our hobby...
And if it can be done, let's do it! :D

it's wise to save tubes for roles where sissy jfets are ..... sissy .
Well, consider a 6N6P tube is really cheaper than jfets; I got 10 of them for less than 1 Euro each, while jfet cost more.
And, moreover, I LOOOOVE tubes, so why not to put them in a so sweet amp?
 
I'm looking at the current sink in the upper right corner.
Honestly, I don't think anything but your mosfet is active.
Below the set limit, the transistor might as well not even
be there.

Yes, if the current draw tries to rise above the "constant
current" limit it shunts gate to the output. I'm not sure
what ever pulls the gate back up again? I see no pullup
circuitry. And the cap C4 can't conduct DC...

Perhaps you are relying on the opamp to recharge C4
on the opposing downswing? Or maybe I don't get it...

You don't need all that useless current regulating crud
on Q7, as it seems to me that Q7's job is enforcing the
output voltage as a pure follower. And Q8's to set the
quiescent DC current, and perhaps swing an opposing
AC current in Aleph style paraphase.
 
kenpeter said:
I'm looking at the current sink in the upper right corner.
Honestly, I don't think anything but your mosfet is active.
Below the set limit, the transistor might as well not even
be there.

Yes, if the current draw tries to rise above the "constant
current" limit it shunts gate to the output. I'm not sure
what ever pulls the gate back up again? I see no pullup
circuitry. And the cap C4 can't conduct DC...

Perhaps you are relying on the opamp to recharge C4
on the opposing downswing? Or maybe I don't get it...

You don't need all that useless current regulating crud
on Q7, as it seems to me that Q7's job is enforcing the
output voltage as a pure follower. And Q8's to set the
quiescent DC current, and perhaps swing an opposing
AC current in Aleph style paraphase.


Absolutely right!
I simply cut'd and paste my previous schematic from Aleph Mini...
The current limitng circuit on top is useless.

THIS should be correct.

The input is the second variation.
Thank you!
 
Bbootstrap in that schematic may not be connected optimally.
He's got his music output shunted thru resistor R7 to PSU+.
Referring to moore2355, I havn't gotten to the other yet.

If he'd tied the lower end of cap C3 to Q2's source, any
bootstrap and PSU noise would remain hidden above &
behind the Aleph regulated current as seen thru R5...

-----------------------------------------

And that article has to be the third time today I've read
some nonsense how a follower hides gate capacitance.
Its starting to become my pet peeve.

Granted, it doesn't multiply it by swinging an opposing
signal. But you still have to charge or discharge a gate,
or nothing of interest happens in the channel.

Simple case of cause before effect.
 
kenpeter said:
-------------------------------------

And that article has to be the third time today I've read
some nonsense how a follower hides gate capacitance.

The formula for C has a deltaV component right? How much delta in V is there when in follower topology? And, How much deltaV is there in common source and common gate? :xeye:
 
flg said:
Love the doggy thing...

/OT

Do you mean my avatar?
She is my beloved dog, Nayma; My sister-in-law found her in Barcellona (Spain), when she was a puppy, alone, hungry and scared (some other stray dog was attcking her, and byte her in few points).
Some not-so-good person abandoned her.
I then went there to meet my sister-in-law (I'm Italian...) and brought her back at home with me.
She looks really close to a Rodhesian Ridgeback, but she isn't.

OT/

Ciao
 
flg said:


The formula for C has a deltaV component right? How much delta in V is there when in follower topology? And, How much deltaV is there in common source and common gate? :xeye:

Depends wether you are driving only a CCS, or a CCS in parallel
with an actual load. Doesn't take much change in a gate charge
or channel current to do nothing, but swing volts against a CCS.

But to swing any Voltage against an 2~16 ohm load is something
that will require a change in channel current. Thus a change in
gate charge is always required, or nothing happens.

I believe the change in gate charge for a change in channel
current is the same, wether its an anode follower or source
follower. Only differences belonging to Miller effect as seen
at the Drain. Gate to source capacitance has to be charged
one way or another, no way around it.

---------------------------------

As for the most recent Croc effort, you can simplify the coupling
between the drain of Q4, the load, and Aleph, to two resistors.
And totally get rid of the coupling cap in the process.

Direct couple Q4's drain to the emitter of Q3.

Replace R16 (0.47R) current sensing resistor
with a series pair of 0.22R, basically splitting
R16 in two equal halves...

Connect the load in the middle of the totem.
Right between the eyes. (The series resistor
pair, I mean.)

Now ditch R3, R5, R10, C2.
Don't need em anymore.

See thread: "Possible improvements for Aleph?"
further details and explanation of this topology.

I also have a Hybrid Zen + Aleph schematic
posted there. Needing only a DC servo like
yours to be complete. I may need some help
with that part.
 
Babowana said:
Hi Giovanni,

I'm looking at your hybrid in the post#12, with interest. I could understand how it works. But not fully. About 80% . . . ? And, I'm :scratch: how Q4 gets the bias voltage. How . . . ?

Tks.

>:)<


The op-amp is there for that purpose.
The OP07 is a DC servo, which will bias Q4 in manner to maintain (talking about DC condition) 0V at the output.
Please refer to
THIS article from the TubeCAD.

Ciao
 
Noticed you where into the modulated-CCS thing. Maybe the NFB is needed then.

Two simplified versions of the driver:
 

Attachments

  • drive.png
    drive.png
    73.2 KB · Views: 900
Formerly "jh6you". R.I.P.
Joined 2006
croccodillo said:

Please refer to
THIS article from the TubeCAD.



Giovanni, thanks for the reference. It gives me a clear picture.

I was wondering why R13 was there. It seems that now I understand the reason. It is to keep very low dc voltage at the output of the servo, and by doing so, to make the dc offset at the amp's output "drift vertually at 0V", isn't it?

Ciao,

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