j-FET / Triode Phono front-end

What people are working on at the moment
Laurence
I am not a user, I am a free man
Posts: 73
Joined: Wed Oct 05, 2011 6:21 pm

#211

Post by Laurence »

JamesD wrote:Just had a thought - Laurence might be onto something...

If the Fet/pentode transfer curves curvature doesn't cancel for higher order harmonics then maybe a FET-U/L triode cascode can be tweaked to cancel higher order harmonics? And the compound stage might sound better than the alternatives....

Must be worth a shot...

James
Funny you should say that. I was thinking on these lines when reading the write ups and saw that the cascode bottom half could possibly be better replaced with the Jfet with a triode top half (current allowing), for a better combination than the restricted 6dj8 family.

But my attention was drawn to Steve's opinion (also familiar with pentodes' and a man who does build currently) and then listening to Mark's gripes about trying to get the solid state element to work to his satisfaction. I did actually think just maybe the 6dj8 family is worth a go as a whole in this position instead of the jfet.

But I defer to your experience of pentodes in this position.

something will possibly come of the combinations we have discussed anyway.
jc morrison
User
Posts: 56
Joined: Thu Jan 19, 2012 11:06 pm

#212

Post by jc morrison »

sorry i've been awol so long. life takes over...

the new phono pre is working much better. the problem was in the buffer stages. that's what you get when the fancy pants stuff leads the way. we also discovered that we need a better negative 12 volt supply. all part of the job.

almost at the point i can post the whole thing without embarrassment.

i want to say one thing about the spectra of triode vs. pentode vs. fet... it is most likely that the fet (especially if you are using 2SK146-7 or a few more 2SK170s) has considerably less high order crap than the pentode. and the folded fet-triode cascode probably only has a peak of 3rd in the .01 % range open loop at the signal levels we are talking about. the partition noise is of the pentode is the culprit. the pentode is still "sweeter" and more forgiving than the hybrid. but the resolution is superior in the hybrid. the work becomes focused not on noise or distortion so much but much more on the analog character of the materials and components.

that is controversial. but that is where it end up. as you know, i opt to use a transformer together with the folded cascode, and my reasons are entirely based on aesthetics... but it is also practical: not only can the noise floor push the limits of what is possible, but the transformer can do a bandpass thing for the top and bottom that is infinitely more musical than a rumble filter and a tweaky riaa (like many high end phono pres employ) that favors the top (12K is the "air" frequency for some designers), or the bottom (50Hz bump over a 18Hz brickwall). yuck.

anyway, i just want to say that the hybrid approach is absolutely an option, and if you are having difficulty with it, don't give up. get rid of the coupling cap following the first stage and directly couple to the second stage. it's not so hard to rearrange it. that will get rid of a great deal of brightness.

jc
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#213

Post by IslandPink »

Thanks for an update JC .
Now, what was it about the buffer ? ... was it the feedback ones with the jFET / Philips circuit or the Mosfet, or both ?

I think I'm going to have to hand over the baton to the others for a few months on this, after toiling with the 2SK369 inputs & its variables.
However, some of your efforts have certainly rubbed-off, and I now know what the bass performance of a 103R or plain 103 REALLY sounds like , rather than what it sounds like through a step-up transformer , which is useful education.
My plans right now have moved into a rebuild of the current phono in a more 'classical' style to get the LCR units in there ( finally ) . Your circuits with the buffering into the LCR got me thinking about how to do the front end differently. Right now, the new top-plate has been made with the following circuit in mind :
(i) Pieter's 1:10 step-ups ( for now )
(ii) E83F pentode / 6N6P Aikido -
The E83F has about the right Trans. to get me 125 gain when loaded with about 15K , putting the anode ( importantly ) at around 150V to feed into the 6N6P pair. They will run around 23-24mA for a reasonable dissipation.
Hopefully this will be enough to teach the LCR chokes a lesson, but if not I have sockets there to double-up on the current drive ( or at least 30-35mA ) .
I'm interested to see how well the noise cancelling of the aikido connection works at this low level . If it works very well, I may be able to ditch the decoupling/smoothing stage from the VR tube supply that's required to feed st.1 at the moment .
(iii) ... into 1.5k LCR chokes. I will try the 1.5k input ( via series resistance ) and the 1.5k termination, which will require a bigger input cap .
(iv) Into C3g(triode) final stage, as I have less requirement for gain at the end , now .

I have a few 2SJ74 ( BL ) I can loan out if anyone wants to try the folded cascode some time soon .

JC, does the transformer you use with the folded cascode have a wider frequency band than a typical step-up ? - that would be good. I don't know what the primary inductance of my step-ups is, but it must be an issue with the 103 , as they sound a lot better in the low bass with lower-impedance cartridges ... unfortunately those sorts don't give me the tone and musicality I need .
"Once you find out ... the Circumstances ; then you can go out"
jc morrison
User
Posts: 56
Joined: Thu Jan 19, 2012 11:06 pm

#214

Post by jc morrison »

the transformer we use in that preamp is a silver litz 1:10 made by christof kraus (silvercore). it has a cobalt/iron amorphous core. this is the owner's fave and it is fabulous. and expensive.

i like the hashimoto, which is a permalloy 1:10, as well as the tamura permalloy model... i am terrible with model numbers usually. they only make one now. it used to be called the SAT-5. it has another name now but i think it is the same. i have heard really good things about sowter's step up from people i respect. slagle's is also supposed to be lovely...

i am not totally fixated on bass. but timing is big deal... i can deal with some roll off down there as long as the transient edge hangs with the bass. keeping a preamp mainly dc coupled helps maintain that, and one good trannie in the path does not usually screw things up, but caps are more of a compromise, in my opinion. i have heard some excellent pres with one cap and one trans, but few with more than that.

but phono preamps make more of an impact on the sound of a system than amps do. next to speakers, of course! small changes are much more noticeable... it should come as no surprise that taking out or changing step up trannies is fundamental.
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8862
Joined: Mon May 21, 2007 9:42 pm

#215

Post by Paul Barker »

Thank you JC. It looks like a goal with a high price tag and a long time to achieve but an adventure worth taking steps towards.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
jc morrison
User
Posts: 56
Joined: Thu Jan 19, 2012 11:06 pm

#216

Post by jc morrison »

the tamura step up was SAT-1000. my memory sucks! it is called something different now but i believe it's the same. they haven't really changed it. new is expensive. mine were used and very reasonable. if you ever see them come up on ebay, grab em. the ortophon was also good and comes up reasonable from time to time... i haven't seen any though. i do look.

i do think hi-perm trannies are better than M6, for small signal stuff. nickel or amorphous... the one has better small level stuff and the other has better bass and highs. but both better than M6, which is kind of bland and missing detail with copper. with silver, it's another matter. but, silver and amorphous is an amazing combination. wide range and subtle nuance. it can be too much with nickel cores, unless the electronics are soft. a balancing act for sure.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#217

Post by IslandPink »

Interesting ...
http://www.hifiheroin.blogspot.co.uk/20 ... tions.html

Direct-coupled ... pentode first stage ...
"Once you find out ... the Circumstances ; then you can go out"
User avatar
Joe Roberts
User
Posts: 16
Joined: Fri Jan 20, 2012 10:02 pm

#218

Post by Joe Roberts »

I heard this Emia preamp in a semi-final state. It is really good. Nice weighty musical sound. Lots of meat on the bone.

Unlike jc, these guys believe in old school LC filters and tube rectifiers...none of that double regulated capacitance multiplier stuff.

Surely this classicist PSU approach brings certain qualities which are increasingly rare in the modern phono amp world where most units have three tubes and 15 MOSFETS.

The version I heard did, in fact, use a DN2540 current source for V2, but I trust Jackson and Slagle to avoid robotic SS artifacts at all costs.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#219

Post by JamesD »

I've been working on the inverted cascode idea on and off for the last year but Marks reports of its sound as being slightly poor on tone compared to a SUT and JCs use of a SUT input have rather put me off building it (plus not having had the time until this autumn...) so I've been thinking about what to build - a FET/valve cascode a la Nick (and others!) a PNP bipolar inverted cascode or something different???

Back when I had an audio amplifier company we made a mc head amp that used common-base input circuitry and it sounded extremely good - in my head its the reference sound for mc head amps or SUTs and has never been equaled (but long term acoustic memory is unreliable so take this with a pinch of salt). Anyway it made me look at common base input phono amps and, of course, the famous Leach design... This inspired me to think of how I would build something like this now... Well I have some nice low noise JFETs which should be better than using bipolars so how about a JFET version - biasing would be easier if I used selected matched Idss complimentary pairs... Nice so I came up with a design for the front end as attached...

It uses two 9V batteries for power. The interesting thing is that it looks like the signal reaches the output through the two lots of capacitors... and whilst these caps are in the current loop of the signal path they experience no voltage modulation! The batteries are completely floating and the top and bottom signal lines move up and down in step with each other leaving no voltage modulation of the caps :) Since it is voltage modulation of capacitors that causes distortion in capacitors we essentially get a free ride through the caps :-) (I'm sure there are some very low level effects due to the very small current modulation of the cap but all the normal distortion generating mechanisms for caps are voltage driven)

The two resistors from the top and bottom rails are very high value and serve only to define the midpoint between the two rails - signal wise the caps completely bypass them - the Gm of the FETs works with the output resistor to develop the output signal voltage and that drives the next stage. Input impedance looks like about 20-25 ohms for the FETS I have (LSK170 and LSJ74) Idss looks like 6mA or so.... For more gain and lower noise one could add parallel comp. pairs to the amp - the price being current consumption and therefore battery life...

Next step will be to use this to feed a Pentode input RIAA stage - so another Gm amp...

Oh this isn't new config - its heavily based on the Leach mc pre and the two comp FETs are used by Nelson Pass and many, many others in this sort of arrangement but it the first time I've used it for an mc frontend :D It should sound very natural and it should have very low distortion - we'll see...

more to follow as I design the rest of the phono amp.

Its not built yet so if I've dropped any clangers please shout!

J

edited to correct drawing!
Attachments
jfet mc frontend
jfet mc frontend
JFET complimentary MC head gain block.png (4.15 KiB) Viewed 11269 times
Andrew
Eternally single
Posts: 4206
Joined: Thu May 24, 2007 2:18 pm

#220

Post by Andrew »

Hi James,

Would it be possible to do a version that wasn't single ended? Seems to me that's one free lunch you might be missing out on by grounding the cold channel from the arm? Perhaps I have misunderstood? It is early in the day...

I suppose this might sensibly lead us into a discussion on balanced phono inputs....but a SUT will give you that option.

Andrew
Analogue, the lost world that lies between 0 and 1.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#221

Post by JamesD »

My initial reaction to that is that the huge difference in input impedance would be a problem ... but I'm not quite sure how.. I'll have to spice it!
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#222

Post by JamesD »

Attached is the phono amp as it stands at present.

The choice of pentode is one variable and I worked through these as options: E810F, D3a, E280F, C3g, EF184. There are many more one could choose of course but these represent a choice at price points of €60 to €6 each. They have a Gm of between 50 and 20mA/v at suitable op points and require between 40 and 20mA anode current for these Gm values. A quick table of equivalent noise resistance for these valves is interesting:

Eq. Noise R @
E810F 110 45 MHz
D3a 150 45 MHz
E280F 220 45 MHz
C3g 300 45 MHz
EF184 330 40 MHz

But what is really interesting is the frequency that the Req applies to - they ae all RF noise figures... A little digging in tube data references found that the audio Req will be between 10 and 1000 times worse depending on the valve - for the ones quotes it is between 10 and 50 times worse with probably D3a and C3g being the two best behaved at audio frequencies followed by E810F. But it explains the need for a head amp or SUT when using mc cartridges...

I have a couple of E810F handy so I designed for that rather than D3a - which is probably the most sensible one to choose from a cost effectiveness point of view...
Gain structure needs careful thought... Its easy to get so much gain before the RIAA that huge voltage headroom is required. I did one run through and ended up needing 4000V for the headroom at 20KHz!!! How does that work? Well assume 300uV output at 1KHz for a 5cm/s cut. 20KHz is 10 times higher than 1KHz so that 3mV for the same cut. I have always taken maximum cuts to be some of the DDM cuts that max'd out at 55cm/s peak cut but recently I've read pof a 95cm/s cut peak on a special 45 rpm test disk. So allow for 20 times for maximum cut - that makes 60mV. Allow for a 10 times peak on top of that for a really bad scratch and that is 600mV at the input to the amplifier. The mc head amp has a gain of 30 times so that is 18 Volts peak (this is limited by the batteries used to power it +/- 19.2Volt rails - up from two 9V). Then the E810F is run at 200 times gain and we have 3600V!!! Of course the E810F input clips before this! Its set to clip at 2V input so it really only swings 400V on its output! But it is run at 150V on the anode so that clips too...

SO gain structure is a really tricky calculation for this beast to get clean signal through... and so is choosing where it will clip and therefore how it clips on scratches... I'm still working that out and it might mean changing to a lower Gm pentode and/or dropping the parallel input FETs...

The output triode is an easy choice as it just needs a gain of 8 (probably) and low Ra to drive into the next stage oh and low distortion and low noise... 6N30pi is a simple choice for this...


Anyway this is where my thoughts are today :D

James...
Attachments
FET/Pentode RIAA amp.
FET/Pentode RIAA amp.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#223

Post by IslandPink »

So, who's going to build it and tell me what it sounds like ? :D
"Once you find out ... the Circumstances ; then you can go out"
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#224

Post by JamesD »

I've started breadboarding the mc head amp part...

Maybe I should finish that bit and send it to you to use inplace of the SUT :shock:

What gain would you like from it and what Zin? :lol: :lol: :lol:

J
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#225

Post by IslandPink »

Aaarg !
When do you want it doing ?
Like work , everyone wants me to help them out , simultaneously ...

Z-in, about 100R .
Gain , well 10 to max 20 really, to interface to existing D3a/D3a.

Can't promise to do anything before mid-Nov, other stuff I want to do before Owston .
"Once you find out ... the Circumstances ; then you can go out"
Post Reply