JFET input, MOSFET VAS, LATERAL output = Perfect!!

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

We go back to the small signal MOSFET input and BJT VAS and stop changing the circuit so much to try and get it to work with a JFET input and MOSFET VAS.

I still think my original version with the zvn3310a input and bjt vas was right up there in terms of sound quality, and it had a very stable offset. Every subsequent modification, including adding the fet VAS, has been done to try to get the circuit to sound good with a jfet input. In fact I have a good mind to revert to that version.

Thanks Hugh for that circuit description. I was under the illusion that the voltage at the gate to the vas fet was more or less fixed, and that the increase in temp was turning it on harder due to its decreasing Vgs, resulting in more voltage drop across the bootstrap resistors and thus increasing offset. I need to look back even further into the circuit to see the culprit!
 
Hugh,

I really appreciate your input here, you have a wealth of experience that many of us are lacking but please don't feel obliged to check out avenues that you arn't really interested in.

I already learned a lot from your post on the mechanics of offset and I'm sure we can figure something that will suffice.

cheers

mike
 
Would be nice with a simple servo solution that will keep offset stable,, and not include and OP-amps in the signal path....

What if the servo adjustment was placed in a CCS for the input Jfet.. and that it worked by changing the reference point for the CCS...like with a current variable resistor...That way the servo could be outside the amplifier loop..
 
My idea for a servo, if it is needed, is to have a fixed i/p drain resistor with a ( perhaps optical ) variable resistor / transistor across it to trim it. This mean the bulk of the current will come only through the fixed resistor and perhaps 10% only through the servo / trimmer variable resistor.

Details to be designed :)
 
Swordfishy,

In the preamble to the JLH 80 - 100W amp article he more or less says that it's hardly worth bothering to design yet another amp unless we feel it can be significantly better than anything else that is out there at present - perhaps that is more of a challenge these days but I still feel that it worth aiming as high as is possible

In this, and indeed any design, the quality and low noise of the i/p stage is absolutely critical. Any noise here gets amplified along with the signal and cannot be removed.

For me, doing some necessary fiddling to accommodate the best, lowest noise i/p device means we have gained an advantage that simply cannot be regained if it is discarded.

So, if you forgive me saying this without having built yet, I think it's best to persevere with the Jfet.
 
Last edited:
Mike,

A circuit of this type, what I'd call elegant minimalism, is best done with two relatively high transconductance devices of opposite gender in cascade, usually a couple of bipolars. In fact, I've done exactly this with my Soraya, Maya and NAKSA power amplifiers, and I've become quite accustomed to it.

What prompted my interest was the energy of SWF, his open mind, and his approach using a brace of fets right through. I've never designed with laterals before, and it drew my interest.

I now find that the lack of transconductance of the very first device has created some problems, reduced others... I've learned a lot here. BUT, the use of a vertical FET as the MOSFET does create problems with offset which may not easily be solved without a servo. Since the pcb I've layed out has only four devices - one less than my lower power NAKSA, a creditable achievement in itself - I'm reluctant to go servo in this application.

I would prefer to see the ZVN3310A input mosfet coupled to a bipolar VAS, a technology I know very well. The other thing that puts me off is the appalling tolerances of JFETs, they are very, very difficult to work with because this bears directly on the loop gain. Another is the new Supertex DN2530, an outstanding depletion mode device available from Digikey with 200mA Idss, 0.74W Pd, and high gm of 300mS. If Greg (SWF) feels this can be done without sacrificing sonic quality, then I say we should do it - because it won't need a servo and will maintain the very low component count (and not waste all those hours I've spent laying out a great board!!).

For the VAS I recommend the 2SA1360, a Toshiba planar epitaxial pnp device rated to 2.5pF Cob, 150V Vce and 200MHz speed. These devices are video amps and will handle a 50mA inductive load with up to 5W dissipation - absolutely ideal. They are easy to drive, and with proper compensation sound terrific. A ZVN3310A front end, being a mosfet, will be a tougher front end, less prone to gate destruction, and exhibit a little less gm than the 2SK170. My only concern is the current it will draw, somewhere around 5mA, I would think.

Cheers,

Hugh
 
Last edited:
I think that the creative dynamics within this thread are such that it was unlikely that we where ever going to agree on a single design, but I also have learnt a lot from Hugh and many others but particularly Swordfishy for his fast building & good listening and of of course Lineup for the fantastic idea - thanks.

I am going to build two of these amps and one of them will be whatever swordfishy says he likes the best from there my personal research project on this circuit will begin.

over & out for today

mike
 
AC versus DC coupling - affects on DC offset.

Hi All,

As I said, I had never done a proper DC offset test with the MOSFET VAS and AC coupling. Well here it is. I'm sure you will agree the results are truly excellent. This is with the IRF9610 VAS and bootstrap CCS, as per the direct coupled version.

Only a 40mV turn on "thump", settles to within 10mV within minutes and stays there. The large spikes are caused by me messing around with other things nearby and bumping things.

This circuit is exactly the same as the direct coupled version I posted in post 690. The only difference is the addition of the feedback capacitor (1000uF), and changing the jfet bias from ground to a few volts with a LED and POT array - as per my previous versions. The VAS device, VAS current, VAS temperature, feedback resistors and devices are all the same.

Hugh, I believe for the final version we can safely stick with the MOSFET VAS. Wouldn't you agree? What a good result!

If it is a choice between adding more devices to allow DC coupling or putting up with two capacitors, I pick the two capacitors.
 

Attachments

  • Image 016.jpg
    Image 016.jpg
    63.1 KB · Views: 582
Last edited:
Mikelm.

Here are the DC offset test results for the direct coupled circuit I posted. As you can see it did not stabilise within an hour, though it was better than I expected.

I must admit, this is the first thorough test I have done with the mosfet VAS, so I don't know if this drift is caused by the fet VAS or the direct coupling or a bit of both.
It is constantly drifting for an hour.
The level is not so high, though. Better than I thought.

But what would happen to the DC-offset if we play music for a while.
Idle amplifier is one thing.
Working amplifer a totally other thing.
 
All,

I just tried the direct coupled version shown below.
-----
Lineup, I think a FET ccs for the input might be a good compromise between complexity and stability.
I use BC550C current source for input.
I use BD139 current source for VAS.
I dont know why use JFET current source.
One thing maybe would be the same temperature characteristic as the input device.
 
Hi lineup,

I would like to try the recommended version of fetzilla. I have most of the parts except the IRF9610. I do have a few PMOS to try first, I will order the proper VAS MOSFET if the trial sound good.

I will not use the input cap (C2), will substitute that with a 1K resistor. I use only a 25K pot & so I will increase the R8 to 100K. Otherwise I will build as shown in the schematics.

Cheers, Stanley
Great sng001 Stanley. :)
I need more people to build.
I would be happy and much interested to see AKSA's real version
 
I think that the creative dynamics within this thread are such that it was unlikely that we where ever going to agree on a single design, but I also have learnt a lot from Hugh and many others but particularly Swordfishy for his fast building & good listening and of of course Lineup for the fantastic idea - thanks.

I am going to build two of these amps and one of them will be whatever swordfishy says he likes the best from there my personal research project on this circuit will begin.

over & out for today

mike
MiiB
I agree. We will seldom be liking the same setup.
I once said there are at least as many Fetzilla as there are people building.
But this is a part of the fun - the discussion.
Discussion make us learn, develope.

I would be happy to see you build, MiiB
Especially your own version. The one you like.

Thanks for being around. Thanks for your ideas.
 
Hi All,

As I said, I had never done a proper DC offset test with the MOSFET VAS and AC coupling. Well here it is. I'm sure you will agree the results are truly excellent. This is with the IRF9610 VAS and bootstrap CCS, as per the direct coupled version.

Only a 40mV turn on "thump", settles to within 10mV within minutes and stays there. .
That was an excellent stability. At a low level.
 
Official version, per now

The version I recommend, with 2 current sources at bottom.
For Input and for VAS.

I post it for new readers and potential new builders.
This version should be stable and sound good.
Which means troublefree run after construction.

Also good for the discussion to refer to this schematic.
(swordfishy, AKSA and others can post their versions, too)
 

Attachments

  • u_fetzilla_6l_nocap2_public.png
    u_fetzilla_6l_nocap2_public.png
    10.2 KB · Views: 630
2SJ313 from Toshiba

Hi All,
and in particular SWF, good to hear where's a will to go a bit wild...

And I'm good and curious to see and hear about Your more exotic findings!

Re: the distorting figure comparisons:

I don't thing there on the basis of a difference of 0.002% is a reason to prefer one device over the other?

On the other hand, then experience tells that this can sound very different indeed...:cool:

2sj313 can be acquired on eBay. Just check with seller, that they are willing to ship to Your current whatever remote area...
I know of this issue myself, since I'm currently Long term consulting behind the walls of the 'Reign in the Middle' IE. Mainland China...
Good eBay sellers are willing to deal with this 'little chestnut' though...

All the best
DocO
 
Re: the distorting figure comparisons:

I don't thing there on the basis of a difference of 0.002% is a reason to prefer one device over the other?

On the other hand, then experience tells that this can sound very different indeed...:cool:

2sj313 can be acquired on eBay. Just check with seller, that they are willing to ship to Your current whatever remote area...
I know of this issue myself, since I'm currently Long term consulting behind the walls of the 'Reign in the Middle' IE. Mainland China...
Good eBay sellers are willing to deal with this 'little chestnut' though...

All the best
DocO

Doc0.
I will try to find 2SJ313 spice model. I dont think it will be easy.
IRF9610 and ZVP3310A are good transistors.
Many, after build Nelson Pass stuff, have already IRF9610 at home.

My guess after SPICE compare is that no transistor is better
as a VAS device. They give the same result.
This is only guess, cause we can not trust SPICE always.
THe true result comes when somebody build and measure and listen.
 
Fetzilla as Preamplifier.

It is easy to use Fetzilla as preamplifier.
Just remove the output stage and take output from VAS.
One can adjust for more or less current.
In my diagram it is 11mA. This is standard for VAS.

A good project to begin with :)

Distortion is low and sound should be good.
But this remains to see/hear when somebody
builds Preamplifier Fetzilla.
.
 

Attachments

  • u_fetzilla_6l_preamp.png
    u_fetzilla_6l_preamp.png
    15.9 KB · Views: 590
VAS MosFet choice

Doc0.
I will try to find 2SJ313 spice model. I dont think it will be easy.
IRF9610 and ZVP3310A are good transistors.
Many, after build Nelson Pass stuff, have already IRF9610 at home.

My guess after SPICE compare is that no transistor is better
as a VAS device. They give the same result.
This is only guess, cause we can not trust SPICE always.
THe true result comes when somebody build and measure and listen.

Dear LineUp,
I think Your comment here more or less in line of what I proported...

I have been looking for a 2SJ313 spice model myself to no avail...
It's complementary 2SK2013 has however got one!
I will see If I can manage to adapt it suitably as an approximated 2SJ313 - but not to many firm promises here pls.:rolleyes:

What's really interesting as well to me, is that it was reported that Erno Borbely has used these Toshiba's in one of his constructions.
That should indeed vouch for something sound quality wise?
It does for me anyway, since I got a lot of time and respect for his work.

I'm currently not in a position to help with building up a working amp (would if I could), so I have to limit myself to be one of the 'fence sitters' or equally irritating 'car passenger (not very helpful) driving advisor's'.;)

Best of greetings
DocO
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.