CDP / DAC output stage

[karatestu]

Hey,

I have been messing about with my Naim CD3.5 and decided after reading the lampizator snip it page to bypass the anologue output stage. I must admit that his views do strike a chord with me regarding keeping things simple if possible.

I lifted one leg of a 10uf coupling cap in each channel, added a 100K resistor to ground and that was it. Wired to my fixed resistor potential divider for volume. I have removed (per channel) 3 opamps, countless resistors and caps, 2 coupling caps and the output relay from the signal path. And also a twin secondary 500VA toroid with Avondale tpr4's followed by two ALW super regulators.

I feel quite liberated to have erased that lot from the equation. Simple is better, right ? I expected to hear something horrible when I pressed play for the first time (no not Celine Dion's latest offering) but it was not horrible. More detail, more natural and less processed sounding, less fatiguing. It would appear that a 7 pole Bessel filter is a bit of a music killer. I did keep a 1st order filter straight off the dac consisting of a 680pf polystyrene cap between two of the dac pins. The dac is voltage out as it has an internal opamp covering IV conversion.

The bistream dac (TDA1305) is not very well regarded and is said to need lots of filtering to get rid of out of band crap. Well, it sounds good to me (cloth ears ?) with just a 1st order filter on the end. I wouldn't know what aliasing sounded like to be honest but I don't hear any nasties, just lovely joyful music.

I started to wonder - do i need a buffer so the poor dac is not loaded down so much (if it is in fact). I use a passive pre (well just a potential divider with 10k in series and 768R to 0V) so was a little worried. Things that popped in my head were to use a TVC or a cathode follower valve buffer at unity gain. But then again maybe i dont need them or in danger of over complicating things again

Anyone else here like simple cdp or dac anoogue output stages ? Or have any wisdom to impart ?

Thanks, Stu

[Wolfgang]

Karatestu,
you should be able to use the output without buffer. From the TDA 1305 data sheet:
"High precision, low-noise amplifiers together with the
internal conversion resistors RCONV1 and RCONV2 convert
the converter output current to a voltage capable of driving
a line output. This voltage is available at VOL and VOR
(1.5 V RMS typical).

But from my practical experience it sounds a lot better (details, transients) if you use a unity gain buffer (6H6, 6H1) or even better a buffer with gain with RCA 26, 300B, EF86 or the substitute 6267 etc. It also depends on the input sensitivity of your poweramp and your speakers.

[karatestu]

Hi wolfgang,

Thanks for your input. Yes, i have read the data sheet again and it ties in with what i am hearing. Why Naim plonked a 7 pole bessel filter on the end i dont know. Or maybe i do. The unit has to measure well or it wont sell or be taken seriously.

The data sheet says Zo (output impedance ?) Is 10 ohms.
RL (output load resistance) is 3k ohms
Rc (output load capacitance) is 200pf

Taking Zo to actually be the output impedance, i have no doubt that this is low enough to be able to drive a low capacitance cable and 10k volume control no problem. Not sure what the RL and RC parameters will mean in my case.

Power amps are nva diy jobs which are designed to be used with passives - high input impedance, high sensitivity i think. Speakers are diy isobaric, 3way
Semi omni. Bass and mid drivers are doped and have no crossover. Tweeter has padding resistor and 3.3uf cap in series. Very simple.
I will try a buffer at some point. What options have i got - cathode follower, opamp, transistor........

[Wolfgang]

Why don’t you try to connect the output directly and listen how it sounds? In the data sheet it says that the TDA 1305 doesn’t require an analog output filter.
If you drive the input of your amps “passive” you will lose some SQ which is available in the original recording . Unity gain buffers (like cathode follower) don’t help much in this respect although they add some tube harmonics. If you use a typical preamp with a driver and an output tube (like the TS MP for example) you will get more details and better dynamics but you still lose a lot of subtle information and transients. That’s not theory. I went through the whole spectrum of possibilities doing A/B tests because I wasn’t satisfied with the results and the (missing)“realism” in reproduced music in comparison to how real instruments and good studio recordings can sound.
After many comparisons and experiments the ideal solution for me is a single tube buffer with gain between DAC/phono pre and power amp. Ideally with digital volume control integrated in the DAC. But a stepped attenuator also works well in case of the phono pre which I still use with the RCA 26 buffer (never tried it with the EF86 buffer). I use a ladder type attenuator for obvious reasons.
After trying different tubes and concepts I finally found a solution that satisfies all of my demands. A configurable pentode/triode buffer with the NOS EF 86,/6267. The NOS RFT EF 86 sounds a bit softer and warmer, the 8267 extremely open and even more dynamic. New Electroharmonix tubes sound simply boring IMO. In the configuration in which it sounds best (extremely rich tone colors, tons of details & micro details,reverberation time of single tones a lot longer than with regular preamps, lightning fast transients) it’s a bit demanding regarding the IN/OUT configuration. I use it with the 6C33 OTL (100k input, shouldn’t be less) and the Soekris DAC (no opamps) with digital volume control . No grid stopper, no output resistor to ground. I hope Ray will work on the PCB layout when he finds more time as the control logic part is a little hard to build without PCB.

[karatestu]

Hi again Wolfgang,

I have been listening to the cdp without the output stage for a week already. I have 10uf coupling caps and 100k ohms to ground on the dacs output before the phono sockets. Also i left in a 680pf polystyrene cap which forms a low pass 1st order filter. It says in the datasheet that this is all that is required but does suggest 1nf. Naim put in 680pf i presume because the value worked best when combined with the 7 pole bessel filter (that i have now disabled). I wondered if replacing the 680pf with a 1nf as suggested in the data sheet would make much of a difference.

I like the sound very much but want to improve it if possible. There is something i am missing compared to before but cant put my finger on it. There is however more detail and sounds more natural and less processed.

I have a resistive volume control with 10k in series with the signal and this has about 150mm of wire between it and the dac chip. From the volume control there is 1 metre of very low capacitance interconnect to each power amplifier.

What ever i add on the end of this dac (if anything) must not take away the increased detail i am now experiencing and i dont want to over complicate things if i dont have to. It is just missing a certain "je ne sais quoi" if you get my drift.

Stu

[Nick]

What 10uf cap are you using? They don;t all sound the same, maybe consider a film bipass?

[Wolfgang]

Stu,

I wouldn't worry about the 680pF or 1nF. Don't think you will hear any difference. But you might improve things if you use a very good 1uF cap (like for example Mundorf Supreme ) instead of 10uF. 10uF is overkill for lowest -3dB point.
It looks like you use a shunt type attenuator which presents a variable impedance for the DAC output depending on the volume settings but I don't think this is a problem, unless you would hear some kind of changes in the sound depending on the volume settings?

I got the biggest increase in details/micro details/space around instruments, transients when I stepped away from passive volume control and unity gain buffer/preamp solutions. That's all I can say. I had experienced a similar problem as you describe. The missing "Je ne sais quoi" boiled down in my case to what separates good HIFI from how a piano, guitar or any other instruments sounds when you listen to it live (no PA system) sitting next to it.

A single tube buffer with gain is actually the simplest thing that is possible in this position (DAC-amp) that would make sure that the signal doesn't suffer losses from being passed through resistors and that there is more than enough power to drive the input stage of the amp so that it really responds to the tiniest details. With a passive volume control you cannot achieve this. In case of a DAC you would get the best possible sound using a digital volume control. But the difference - depending on the resistor quality - is rather negligible if you use a stepped attenuator.

We can discuss this further (circuit drawings) if you think this is something you want to try. But maybe the 1uF cap will already be enough improvement?

Wolfgang

[karatestu]

What 10uf cap are you using? They don;t all sound the same, maybe consider a film bipass?

Hi Nick,

10uf, 50v coupling caps are wima film ( mks 2xl)

Stu

[karatestu]

Hi Wolfgang,

Thanks for your time with this. I only went with 10uf as that was what i had in there already -the naim stage had 2pairs of 10uf tantalum coupling caps - one pair straight after dac output and another pair after the bessel filter and before the output relay. I just kept the first pair of caps after the dac and used them by lfting one leg .

I will try a 1uf part then and see what happens. I only wanted to block dc as my power amps have no input coupling caps.

As far as active preamp stages go i already ripped an avondale active stage out and replaced with 10k potential divider with the resistor to ground being variable. I cant change volume at the moment as all i have is two resistors making a potential divider. With the power amps i now have it really was a big step forward getting rid of the active pre. But, that is with NVA power amp boards. You may not know anything about them (Nick certainly does) but they are specifically designed for use with passive preamps .

I will get a circuit up of what i have at the moment (when i have tried a 1uf coupling cap) and i will take anyone's offer of advice, thanks.

I have thought of all sorts of possible things to try including solid state, valve cathode follower and transformer volume control (with or without gain). I need to think about my options and you are certainly helping, thanks.

Stu

[karatestu]

1uf paper in oil any good for coupling caps ?

[pre65]

1uf paper in oil any good for coupling caps ?

What make/type is it ?

Some of the Russian PIO caps are not paper in oil.

[Greg]

Yes they are. Some people’s coupling cap of choice. They tend to have a warmer sound compared to polypropylene types. With age, watch out for becoming leaky, both electrically and physically (oil leakage).

[karatestu]

How about a couple of these babies ….

https://www.ebay.co.uk/itm/1-5uf-Paper- ... rk:23:pf:0

I use a 3.3uf size of this cap as a high pass filter for my tweeters. Does a good job in that role

[pre65]

"Russian K75-10 Hybrid Paper and Polyethylene Terephthalate In Oil Capacitor
This Russian military capacitor is quite intriguing in its reported construction and in appearance that resembles a small grenade. Its dielectric is claimed to be a hybrid of paper and polyethylene terephthalate, which is essentially a type of Mylar, saturated in oil. There is some debate as to which Russian “oil” capacitor is the best, and while K40Y is widely recognized and used, some report K75-10 is the premium oil capacitor out of the Russian military.

I can report that K75-10 sounds extremely unique, quite different from K40Y PIO and any other true PIO one cares to mention, including Vitamin Q, Jensen, and AmpOhm. It has a very saturated, colorful, textured, and detailed midrange that is thick and juicy. While PIO’s are known for their natural and smooth midrange, K75 seems to add some jest, pop, and color to the proceedings, yielding an apparently more detailed and forceful presentation. Treble and bass is pretty good, especially for an oil cap, but the leading edges are not razor-sharp and precise like premium film caps. Since the midrange is so involving and palpable, the lack of equal force in treble makes it seem a bit dark overall, but as one increases the volume, this becomes less of an issue, and poorly-recorded music is much more tolerable at higher volume.

The sound is rather addicting in its presentation and can even be called delicious in what it does well, kind of reminiscent of biting into a ripe peach. The line dividing “colorful” and “colored” is thin, and while K75-10 likely leans a bit to the latter, I quite enjoy its substantial sound and prefer it to the thin, plasticky sound many other capacitors represent."

Copied from

http://www.enjoythemusic.com/diy/0411/capacitor2.htm

[Wolfgang]

I was wondering why there is a need for a coupling cap in this position? If there is no DC (and it shouldn't) at the output of the TDA 1305 and no need for some kind of filtering the opamp (1,5V ) can be directly connected to the next stage. What's so special with the input stage of the NVA amp?