Zen -> Cen -> Sen, evolution of a minimalistic IV Converter

[regal]

Originally Posted by heb1001

I made some cable by hand - 3 cores braided, aluminium foil shield, copper braid around the foil, PVC heatshrink tubing outside.

Same connection strategy as with the usb cable before and I haven't changed anything else.

It sounds good again but 2nd harmonic is also back up (by 36dB!) to -100dB 3rd harmonic is down from -87.7dB to -91.6dB.

Not really sure what's going on. Could twisting DAC +ve and -ve together prevent the balanced circuit from canceling out the 2nd harmonic properly?

Its seems that the 2H is probably closer to reality, maybe the earlier measurement was a fluke ?

[EUVL]

If you read the article at Jan Didden's website, you will know that we had a hard time measuring the circuit to -120dB (or 2.8µV).
And that was with an AP System1 in balanced mode.

So when people reporting measuring to -130dB or even -140dB, they might have super professional equipment that none of us have.




Patrick

[heb1001]

Originally Posted by regal

Its seems that the 2H is probably closer to reality, maybe the earlier measurement was a fluke ?

No fluke - there is second harmonic in each measurement of the two halves of a balanced pair and it clearly cancels out when they are mixed together. See attached pictures.

So the total circuit DAC - Sen - ADC in balanced mode definitely had very low 2nd harmonic with the usb cables.

I saw this on all the measurements with the usb cables and I did at least 4 or 5 - it varied a bit, sometimes I could see a little spike at around -130dB.

[heb1001]

Originally Posted by EUVL

If you read the article at Jan Didden's website, you will know that we had a hard time measuring the circuit to -120dB (or 2.8µV).
And that was with an AP System1 in balanced mode.

So when people reporting measuring to -130dB or even -140dB, they might have super professional equipment that none of us have.




Patrick

I'm not claiming to be able to measure the distortion from the Sen circuit in isolation, I'm just reporting the numbers of the overall chain.

The components in the chain might be conspiring together to cancel out distortion but the end-to-end result is pretty clear in the digital data as you can see from my previous post.

My ADC is a PCM4220 and I was doing FFTs with 10M samples using Audacity.

Not much progress this weekend - I tried increasing Riv but my circuit gets too hot with Riv > 256R. I'm not sure whether I just need to add heatsinks to make it work or if it's heating up because increasing Riv is causing it to oscillate (perhaps a problem with the servo which was working really well with Riv=256R).

I did get some more measurements with increased Riv in the window before it becomes unstable with a lower noise floor which you aren't likely to believe either

[samoloko]

In earlier posts (post 421) Joachim has recommended using blue BLOCK split bobbin pcb transformers In noise sensitive enviorment
Is It sure that Block pcb transformers are split bobbin
I can not find where It Is specified that they use split bobbins

[heb1001]

Originally Posted by samoloko

In earlier posts (post 421) Joachim has recommended using blue BLOCK split bobbin pcb transformers In noise sensitive enviorment

Has anyone actually made a non-battery PSU for Sen that they are happy with?

[zoka959]

Originally Posted by EUVL

> next project should be a hassle-free, plug and forget LiFePo4 charging scheme

I don't use them, so there ain't going to be one any time soon.
One reason is that they have limited lifetime, even unused.
My current preamp NiMHs are from 2007, and they are still fine.


Patrick

Patrick
what do you think bout his battery PSU from
THEL AUDIOWORLD
Akkunetzteile für Audio-Module
Zoran

[EUVL]

The link to THEL shows ordinary NiMH AA batteries soldered onto a PCB.
You can do the same yourself.
And there are many chargers on the market for NiMH.
So I leave that to you to choose from, including from THEL.


Patrick

[EUVL]

We have a number of PM's asking us for the same thing -- how to do SEN for PCM1704.

As you can download the article from Linear Audio, you will find that the SEN was designed originally for PCM1704.
Thus, you will need the standard evaluation kit, and 4x of matched quads 2SK170BLs. (i.e. totally 16 JFETs).
Alternatively, you can also use 2x (or 4x if you want lower Zin) of matched quads of2SK369V.

We do not sell matched 2SK170BLs. But you can get them from Spencer.

The rest you can get from us.
All you need to do is download the service request form and fill it in.
Then send it to Gogowatch and he will process for you.

xen-audio
xen-audio

The content of the evaluation kit are listed at the beginning of this thread in post #3.
We have everything in stock.

All information required are public. Please help yourself.


Regards,
Patrick

[needsp]

Hi Patrick

This is my first post here, but I’m well on the way to assembling all the parts I need to build a 2SK369 based SEN to use with my Twisted Pear Buffalo II ESS9018 based DAC.

Nic has kindly provided me with a set of closely matched FETs. At the moment I’m using a Twisted Pear Legato I/V stage (a little modified) and its op amp–based balanced to single ended converter to feed my single ended valve preamp. Rather than use the Bal to SE converter with SEN, I’ve decided to fit Jensen JT-11P1 input transformers to my preamp. Properly terminated, these will provide an ideally balanced- much better than an op amp, as discussed in a white paper on the Jensen website- low distortion input. (I’ll also use them on my SE inputs, where their superb CMRR should provide additional sonic benefits.)

But, of course, I need to avoid applying any dc to the Jensens. Nic’s matching will win half the battle here. But I’m concerned about using a reference voltage of ½ Vacc (1.65 V). Unless I’m mistaken, the 681k/470k voltage splitter you propose will only result in 0 V (analogue ground) on the output when the floating supply is exactly 18 V. In practice, with NiMH batteries, the supply voltage will be vary from 16.8 V to 14.4 V (your data) as the batteries discharge, resulting in a small but increasing dc voltage on the output, relative to ground. Now, if the power supplies to both SENs which form the two halves of the balanced I/V converter for the ESS9018 discharge at the same rate and from the starting point, there will be no dc on the combined output- but can we rely on this, and, in any case, won’t there be a small dc current through the each I/V resistor?

I notice that Russ White of Twisted Pear has abandoned the practice of using a ½ Vacc reference; in the latest Legato circuit, the DACs see 0V instead, with supposedly no performance deficit (I think ESS feel the same way). So, unless I’ve got this wrong (and all this only just occurred to me!) it might be best to use a zero V reference for SEN- ESS9018 too, with two 680k (or whatever) in the voltage divider? Then all my concerns vanish- if the Fets are well matched.

A couple of other things. Why do you favour using bipolar caps for C1 and C2, rather than polar electroytics? And would you agree- initially at least- I would be best to omit any I/V low pass capacitor, as the Jensen transformer itself rolls off above 95 kHz?

Finally- even though I’ve yet to try it- many thanks for your elegant design!

Paul Needs