a couple of questions on the phono ONO

[Stefanoo]

Hi guys,

i am on the way of building the Aleph ONO.
These below are the modifications i am plannin on making to it:

1- PSU --> i will probably go for the shunty regulated PSU

2- Eliminate the 10uF output coupling cap by installing a variable resistor at the collector of Q1

3-Eliminate the inverter and use in case a specific output transformer to float the output (i will evaluate this in case)

4-Try to replace the led on the current source mirror's block with an appropriate voltage regulator to obtain maximum stability (i will run extensive listening tests and i will go for whatever solution sounds best)

5- I won't install jumpers for gain and cartridge regulation but i will "fit" the ONO to my specific needs by adjusting it to my pick up.

6- Try to see if i can, instead of using the the 220uF electrolytic coupling capacitor's type on the signal path, i will try to use a good 30uF polypropylene (i know that Nelson recently switched from a 10uF film cap at the output of the MC on the ONO to a 220uF on the Xono) because i think, beside film caps are bulky and pricey, they do sound better and cut frequency with a 30uF should still be enough low (around few Hz if i am not wrong), just my two cents.


if anybody wants to comment these 6 points is very welcome.



I have ordered the parts and they should get here pretty soon hopefully.
Tomorrow i will start working on the PCB design.
Today i have finished the simulation session so i have the complete model simulated on my laptop that is always helpful to debug an eventual circuit's fault.


I have few questions for you guys:

a) It seems on the simulation i have run, that there are 100ms settling time before the output waveform becomes symmetric (look at the picture attached).
Is this reasonable? I was thinking that it might be due to the RIAA equalization's network in the feedback path, may this be the reason?


I will use this stage along with my Denon DL 103 (0.35mV) but to do so i will need a gain around 70dB and i was wondering whether

b) it would be better to increase the gain of the MC stage or the MM stage (which is not very high anyways with its 40dB)?

therefore in case i would decide to increase the gain on the MC stage would it be better to:

b1) increase the current sense resistor R40 that apparently could be raise as much as i want, but what is the drawback of increasing this resistor?

b2)cascoding another 2sk170 in order to increase the current sourced to the emitter follower stage? but then wouldn't this increase the transconductance of the cascode increasing therefore distortion figure?


c)in case i would decide to increase the gain on the MM stage instead, where can i act? i was thinking of varying the values on the voltage divider on the feedback path R7/R21 but i am not sure.


Last thing:

from what i have learned so far here on this forum the more the devices are biased and the better they sound.
Therefore going of on this token i was thinking,

d)since the Push-Pull Class A output stage is biased at 10mA, has anybody tried to modify it by varying R56 and thus increasing the voltage drop at R38/27?
this operation would of course require heatsinks on the two devices but oh well.
Also is it not worth to increase a little bit the bias on the input differential stage of the dual nJFET 2SK389?
Or these bias points are what so called "sweet spot"?


Thank you guys very much for your attention.


Best,

[MRupp]

I would think that the Ono is already a pretty perfect design and cannot easily be improved. But since I cannot help tinkering and am (still) in the process of building my own version – or corruption - of the Ono, here are some ideas or comments that I have:

@ 1 : Don’t know this particular reg. but shunt reg’s are said to be good sounding and are certainly a good idea. Additionally, you could try cap. multipliers for the MM stage instead of the 1000uF caps C12 / C20. For the MC stage, I saw an alternative cap. multiplier approach in John Curl's Vendetta MC stage that might be worth a try. He is using JFets and the biasing is set via an additional selected JFET current source / resistor. You could leave the BJT for Q24 but still use a current source / resistor to set the bias, but this is creating a voltage "reference" and you need to select the current and resistor value.

@ 2 : One alternative would be a “variable resistor” in the current source allowing to adjust the bias ever so slightly.

@ 3 : Do you really need the balanced out, maybe a single ended output would be sufficient ?

@ 4 : I would think that the LED is pretty stable and low noise already. You could possibly add an RC filter to Q16. Secondly, you could cascode the current source(s) Q16 (and Q17). A BF245C or similar JFET with high pinch-off voltage might be sufficient. For thermal stability - when eliminating the output cap. – you could replace Q16&Q17 with a dual matched NPN transistor such as SSM2210 (does not cost too much), or 2SC3381 if you can still get them (I learnt they were actually conceived as cascode devices for the 2SK389?)

@ 5 My plan too, If gain setting is not required you could throw out R66, R81, and combine R41 and R80 into one single resistor. Btw., I read somewhere that C16 has been omitted in the XONO.

@ 6 : Personally I would leave the electrolytic caps.

[Nelson Pass]

I think you can try all those things without damage.

[Stefanoo]

Nelson,

you are right, I will try them out and see how it goes.

[Stefanoo]

MRupp,

thank you very much for your answer.
I will reply to you in a short time.

[Stefanoo]

MRupp,

i can finally give you a detailed answer.

with regard to the balance output: no i don't really need it now since i don't want to go through the hassle of stray noise pick up by the cartridge due to the fact that one of the cold signal is not common grounded (if i want a real balance line from the wiring of the tonearm).

Led are stable and low noise, but of course not a much as this baby here :

MAX6265


I got the idea of using the 2SC3381 last night as i was searching for a low noise NPN device.
As for the availability a quickie google shows different options at different price mand you can find it for cheap.
The only problem is if it the part is genuine or not.

You gave me a good hint with the cascode.
I therefore have fold cascoded the 2 transistors (as shown on the pic) although I don't think there will a great benefits by doing so since the cascoding helps lowering the input capacitance and thus increasing the linearity/stability ah high frequencies allowing the block to be driven by an higher load.

Speaking of cascode though, i have thought of cascoding the voltage gain stage of the MC section Q15 by using an high pinch-off N-JFET such as 2SK246 which should shunt the input capacitance down to 50pF: good enough!
(picture attached on the next post)

In order to achieve a gain of 70dB, i figured that it wouldn't be so smart to increase the current sense resistor R40 because this might cause a roll off of the high frequencies due to a lower cur frequency (even though with the cascoding should help quite of a bit here).

Instead what i have thought it would have been preferable here is:

a) parallel more transconductance matched n-jfet since the 2SC1844 can manage up to 100mA.
this would enhance the noise figure but with the drawback of more transconductance and supposedly more distortion at highs.

Since i don't have 2SK170GR but only BL types i can't follow this path.


This is my idea of how to increase the gain:

if i keep the 4 paralleled devices and use the BL type and leave the 22ohm source resistor i have a total of 26mA and each jfet is biased at 6mA which is the supposedly good spot (linear region) and use 470ohm as current sense resistor i achieve 70dB of total gain.

I think that keep the current sense resistor low is good and also bias the jfet up to their linear area beneficial.



that's it, any comment is welcome.

[Stefanoo]

and this is the cascoded gain stage of the MC

[Stefanoo]

the more i look at the folded current mirror and less sense this makes sense to me.
Do you guys think that this configuration might have any kind of advantages over the normal current mirror?

Since i don't have the output coupling cap anymore i was wondering if i still really need the R14/15/28/30 , C8 anymore?

I can't understand why this parts are there (i think that was to obtain a low cut off freq with the 10uF but not sure)

[MRupp]

Sorry, don’t have much time now, so quickly, and I can only give you my own personal opinion - though I have not personally tried or compared most of the options (yet):

with regard to the balance output: ...

I do not understand what the balanced output has to do with balanced input, are you not confusing the two? The Ono is a single ended phone stage that only creates a balanced output signal to connect to a balanced preamp.

MAX6265 ...

Can’t find this particular part but it can’t do any harm if it is a good voltage reference.

I therefore have fold cascoded the 2 transistors ...

The reason for the cascode is that simple current sources have a much lower impedance and lower bandwidth compared to a cascoded CS, see also the recent articles from Walt Jung on current sources. Not sure if this is audible but I do have BF245C (to be found on ebay relatively cheap) and a cascode is straightforward.

Speaking of cascode though, i have thought of cascoding the voltage gain stage of the MC section Q15 by using an high pinch-off N-JFET such as 2SK246 ...

I know Borbely is using this approachand and you can off course do this but I do NOT like the idea at all. From all I have learned so far the K170 is more linear at higher Vds of say 8 – 10 Volts minimum. Actually I read a WEB article where someone varied Vds in a similar setup and heard clear sonic degradation at 6 volts or lower; he stayed well above this threshold to be on the safe side.

In order to achieve a gain of 70dB, i figured that it wouldn't be so smart to increase the current sense resistor R40 because this might cause a roll off of the high frequencies ...

The Ono is using a 1K option for increased gain and I should be very surprised if this is a liability. Why do you not just raise the voltage to accomodate the voltage drop across a 1K drain resistor, possibly lower Vds to 10 Volts also.

[Stefanoo]

HI MRupp.

sorry i didn't explain myself very well.
Of course this is a single end stage.
I was thinking one thing and wrote another ...

you can find it on digikey

http://search.digikey.com/scripts/DkSearch/dksus.dll

type
MAX6225AEPA+-ND

It an high end voltage regulator.

Thanks, you have just answered to my previous question where i was wondering what advantages there are by having such a folded cascode current mirror.
I have cascoded it with the 2SC3321.
Do you think it is better to use the BF345 you mentioned?


yeah i am not sure that an higher current sense resistor would be worse now that i am thinking through actually a 1K won't change that much the input impedance of the gain stage over a 470 since there is a 4K in series the coupling cap.
Anyways it appears better to me to increase the current of the jfets: by doing SO you will push the jfets on the more linear region.

What you've brought up with regard to the 6V VDS is really interesting.
On my set up VDS of Q15 is only 500mV.
I might decide to drop this solution eventually.
i would also like a comment on this, if possible, from Nelson Pass or other members too.

i will make researches on this.