> Also my listening room temperature is very variable
Which is why I posted the schematics in #784.
The TL431 Vref has a low tempco, and you can combined that with a low tempco R-CL, e.g. a 680R Dale RN60 in parallel with a trimmer.
The TL431 itself wants to see at least 2mA, so you should use a current regulating diode(e.g. E252) to replace the resistor at the cathode of the 431.
And I would use 2x BC550C with matched hfe as the pass transistor.
This should be both low noise and thermally stable, at least better than a single 2SK170.
Patrick
Which is why I posted the schematics in #784.
The TL431 Vref has a low tempco, and you can combined that with a low tempco R-CL, e.g. a 680R Dale RN60 in parallel with a trimmer.
The TL431 itself wants to see at least 2mA, so you should use a current regulating diode(e.g. E252) to replace the resistor at the cathode of the 431.
And I would use 2x BC550C with matched hfe as the pass transistor.
This should be both low noise and thermally stable, at least better than a single 2SK170.
Patrick
Time for a little update.
I have now listened to the SEN with the ES9018 for about a week (a few hours per day) and I am truly impressed. The SEN will stay.
To put it very simply it is as if a bottleneck has been removed from my system - a bottleneck I never new was there
I find myself listening at lower levels now as if a "loudness" button had been pushed. This actually comes together with a clearly expanded dynamic range and frequency extension so it is all positive. I have lately blamed my speakers for being a bit shy in the bass domain, but with the SEN it looks like I just lost an excuse for buying/making new ones
Deep, fast and precise.
The sound is beautiful with voices and the tonal balance is great.
The only area where I am not yet completely convinced is the treble. It does seem just a tiny bit harsh to me. I have the feeling that this depends on the quality of the recordings, but I need more listening to confirm this. I may also try filtering the SEN output a bit more.
I'm currently using the SEN with a modified Buffalo II directly into a set of UcD180HG HxR monoblocks (with regulated supplies) and my French Apertura ALTRA speakers. Attenuation in the digital domain ("volumite").
I look forward to a serious upgrade of the power amp
Initially, when cranking up the volume without music playing, I realized that I was having some kind of low-frequency hum resembling ground loop noise. As grounding is something I have been very careful about I was quite puzzled where I was getting that noise from
As I could certainly not exclude magnetic coupling (one transformer with a single primary) I decided to do as Patrick suggested from day one - use batteries!
I got a bunch of alkaline 9V batteries and the problem was completely solved! The only issue is that it sounds horrible when the batteries run out of juice.....
As I have plenty of power sources in the build I will try NiMH batteries and I may even end up using batteries and regulators. Or maybe separate R-core transformers for each channel.
Before I find the final home for the SEN, I will play a bit with the circuit and see if I can locate any new bottlenecks. A nice thing about these circuits is their extreme simplicity and thus a limited number of components to tweak
I will maybe also try out the CEN in a similar configuration.
Those of you who have the V18 evaluation kit lying around should really try it out. Highly recommended!
Cheers,
Nic
Hi Nic ,
I have a question.
is a big difference between CEN and Legato3?
How would you describe it?
I Buffalo2 (dual mono) with 1 legato3.
I am very happy as it sounds with F5X (EUVL)
But I am very determined to do and I / V CEN.
I WANT TO RID OF THE XLR capacitors (Legato3)
I PCB (Thanks EUVL
), search 2sk369V.Thanks
Attachments
EUVL, thanks for sharing this truly amazing and elegant idea! And I want to ask, is FET matching required in any case, or it is possible to use “bottom” CCS FETs with higher IDSS, then “top” ones, and introduce resistors to their sources so that their operation current will become lower by means of local NFB – just to match to “top” ones. And next question is, isn’t it possible to use BSP129 depletion mode MOSFETs in this circuit? My ears prefer not to parallel any device in signal path if possible, and one BSP can introduce capacitances comparable with 2x 2SK170 and much lower input impedance. Sealed lead acid batteries will do high current task easily I believe.
>is a big difference between CEN and Legato3?
I have only tried the SEN version of Patrick's circuit (not the CEN). It is indeed very different from the Legato.
>How would you describe it?
Try it. It is another world.....
>I am very happy as it sounds with F5X (EUVL)
You have the F5X up and running without posting any photos
>But I am very determined to do and I / V CEN.
You should be.
>I WANT TO RID OF THE XLR capacitors (Legato3)
Good idea. With the ES9018 and F5X there should be no caps in the signal path.
> I PCB (Thanks EUVL), search 2sk369V.
If EUVL (XEN audio) does not have the matched 2SK369 for you I can help them/you out. Ask EUVL.
I have only tried the SEN version of Patrick's circuit (not the CEN). It is indeed very different from the Legato.
>How would you describe it?
Try it. It is another world.....
>I am very happy as it sounds with F5X (EUVL)
You have the F5X up and running without posting any photos
>But I am very determined to do and I / V CEN.
You should be.
>I WANT TO RID OF THE XLR capacitors (Legato3)
Good idea. With the ES9018 and F5X there should be no caps in the signal path.
> I PCB (Thanks EUVL
), search 2sk369V.If EUVL (XEN audio) does not have the matched 2SK369 for you I can help them/you out. Ask EUVL.
Last edited:
I think the bottom half of the circuit is just a constant current sink so I think you could use any constant current sink but the current in the bottom half needs to be matched to the idle current of the top half and that changes as the top half warms up so making the top and bottom half symmetrical and using a shared heatsink helps to keep the two halves tracking each other.
I think my implementation is matched to about 1% and I get the distortion profile I posted above. The distortion is different in each balanced half and stays put when I swap the servos around so it's either from my measuring circuit or from the mismatched JFETs.
I have made a working Sen with unmatched JFETs using a floating servo to control the gate voltage on the current sink and the distortion was in the same kind of range. I did this on breadboard so didn't get great measurements and don't know how it sounds but it can definitely be made to work.
I think my implementation is matched to about 1% and I get the distortion profile I posted above. The distortion is different in each balanced half and stays put when I swap the servos around so it's either from my measuring circuit or from the mismatched JFETs.
I have made a working Sen with unmatched JFETs using a floating servo to control the gate voltage on the current sink and the distortion was in the same kind of range. I did this on breadboard so didn't get great measurements and don't know how it sounds but it can definitely be made to work.
> or it is possible to use “bottom” CCS FETs with higher IDSS, then “top” ones, and introduce resistors to their sources so that their operation current will become lower by means of local NFB – just to match to “top” ones.
Yes, you can do this by trimming.
But I rather match and forget about trimming once and for all.
A degen resistor also introduce extra noise, even though academically small.
> And next question is, isn’t it possible to use BSP129 depletion mode MOSFETs in this circuit?
MOSFETs are too noisy for my taste.
And many people parallel JFETs for phono preamps, so I do not see the problem.
Patrick
Yes, you can do this by trimming.
But I rather match and forget about trimming once and for all.
A degen resistor also introduce extra noise, even though academically small.
> And next question is, isn’t it possible to use BSP129 depletion mode MOSFETs in this circuit?
MOSFETs are too noisy for my taste.
And many people parallel JFETs for phono preamps, so I do not see the problem.
Patrick
> You have the F5X up and running without posting any photos
He bought 2 of my finished modules years ago with Fairchild FETs.
Go back to the original F5 thread and you WILL find the photos.
Vitalica,
May I suggest you seriously consider building a pair of new F5Xs with Toshiba MOSFETs ?
Patrick
He bought 2 of my finished modules years ago with Fairchild FETs.
Go back to the original F5 thread and you WILL find the photos.
Vitalica,
May I suggest you seriously consider building a pair of new F5Xs with Toshiba MOSFETs ?
Patrick
Last edited:
>is a big difference between CEN and Legato3?
I have only tried the SEN version of Patrick's circuit (not the CEN). It is indeed very different from the Legato.
>How would you describe it?
Try it. It is another world.....
>I am very happy as it sounds with F5X (EUVL)
You have the F5X up and running without posting any photos
>But I am very determined to do and I / V CEN.
You should be.
>I WANT TO RID OF THE XLR capacitors (Legato3)
Good idea. With the ES9018 and F5X there should be no caps in the signal path.
> I PCB (Thanks EUVL
If EUVL (XEN audio) does not have the matched 2SK369 for you I can help them/you out. Ask EUVL.
Thank you very much for your answer ,Nic
I am determined now to do I/V SEN.
Best regards.
Yeah, have to second this.
I sorted out the circuit grounding and shield grounding yesterday and it's taken the sound to a new level completely.
Listening test is unquestionably better than my op-amp IV and the measurements are getting there too now.
I'm guessing I'm hearing the sound of my tube amp for the first time.
> Listening test is unquestionably better than my op-amp IV
That I almost consider an insult.
Nic will tell you how many other discrete IV's it has beaten, even in its box standard form.
And without any magics (like Blackgate NX, Vishay Naked, ...., etc.)
And thanks to the generosity of Jan Didden, you can now even download the article for free.
What more can you ask for !
Patrick
That I almost consider an insult.
Nic will tell you how many other discrete IV's it has beaten, even in its box standard form.
And without any magics (like Blackgate NX, Vishay Naked, ...., etc.)
And thanks to the generosity of Jan Didden, you can now even download the article for free.
What more can you ask for !
Patrick
When the AD converter does not alias a linear phase filter is the correct way when you want " transparent". When you measure that DA converter ( with a linear phase filter ) it will have pre and post ringing, symmetrically aligned, but you will not hear it.
When the AD has aliasing products that can be an audible because sound will appear "before" it it causal.
When the AD has aliasing products that can be an audible because sound will appear "before" it it causal.
Sorry Joachim if I've got this wrong, but from the above ...
This means that a D->A converter produces pre and post ringing, and you can't hear it, but you can measure it with a linear phase filter, yes?
So why do you need any filters at all, to get rid of something that you can't hear? Or is this something to do with transparency?
This means that a D->A converter produces pre and post ringing, and you can't hear it, but you can measure it with a linear phase filter, yes?
So why do you need any filters at all, to get rid of something that you can't hear? Or is this something to do with transparency?