A NOS 192/24 DAC with the PCM1794 (and WaveIO USB input)

Thank you clivem but I kindly ask everyone to address all questions about WaveIO in dedicated thread located here. Let's try not to pollute Doede's discussions about his fine work! Speaking of that, I found plenty of PCM chips for my future decks but still waiting for an answer about the numbers... I guess four are enough but in your opinion Doede, eight are redundant?
Thank you,
L
 
Last edited:
thanks for feedback. I checked the article from John and it was interesting in terms of getting keen on doing some next experiments with software filtering and up sampling and use some other players. Actually I never tried to listen and measure what happens if you up sample 44.1 to 176.4 or 192 and how it sounds or measure.

this week I finally get my new toy, an AP system two, so I am hoping to get some more insight by doing signal tests as well.

On a sad notice, if you read the thread of John, there is a lot of talking and opinions, but no real DIY which provides an easy route in improving sound for most hobbyists.... Nice for reading time, but not for "listening time"

I only have 44.1/16 music and now upsample it all to 176.4 (rather than 192 as 176.4 is a whole number multiplier of 44.1) using Secret Rabbit Code Level 0 (the highest quality) all performed in software within mpdPup to a WaveIO. It is excellent and gives a major boost to SQ without IMHO any unpleasant artifacts.

I was also very interested in John Swenson's article - but actually more from the point of view of the Ethernet transfer isolating the DAC from any other equipment in the chain. However, as you observed, there has been no news for some months on any DIY progress. I can only assume that is because the concept is being taken forward exclusively as a commercial product?
 
Hi Doede and all other diyers !

I am very interested by this DAC, so much interested that I will buy one of them ! :p

But, as there is no oversampling nor digital filtering, what's about ultrasonic noise (quantification noise) at the output ?
Is it reasonable to think that ultrasonic noise could interfere with the audio band and generate some intermodulation distorsion ?
Or could this ultrasonic noise damage some amplifiers or some tweeters at high volume ?

And, if I understand correctly, without oversampling it is practically impossible to put an analog lowpass filter at the output.
But, if one would oversample by software in the computer (using SOX for example) so all the files are played at least at 176,4 or 192 KHz sample rate (just like jrling does), would it be possible (and useful) to put an analog lowpass filter wih reasonable slope ?
Or would this filter destroy all the benefits of the technical choices of this DAC ?

I don't have any experience with non oversampling DAC's, so please all the explanations are welcome ! ;)

Chers, staki
 
Last edited:
You put 133R at 1794 outputs. I guess this is optimal value for biasing? Do you know if 130R would be right also?

I think Doede might be away on business at the moment.

IMHO, 130R I/V resistor is fine. Doede chose the resistor value to obtain a reasonable voltage output while keep distortion low. Decreasing the resistor size will reduce output voltage by a proportional amount. I think Doede stated a peak output current of 6.3mA per DAC, so......

133R: 2 x 6.3mA * 133 / 1.41 = 1.1885V RMS out.
130R: 2 x 6.3mA * 130 / 1.41 = 1.1617V RMS out.

100K is matched with 2.2 µF?

The value of the output cap? 2.2uF into 100k load will give you a -3db point @ <1Hz. Again, fine.
 
Have in mind to try discrete opamps (Borbely) at output also.

I'll post more when I have time later in the week, but I've already had a go with the Broskie Unbalancer tube output stage and ended up dumping the differential gain input stage (that I was running with a much lower value I/V resistor) and keeping the Broskie Cathode Follower and the original 133R I/V resistors. Output is still 1.2V, so on paper the only theoretical advantage would be if there was anything to be gained by cancelling noise common to using both the +/- output rather than just using the + output direct from the board, assuming a SE amp input. But to get the thing as quiet as possible I ended up with both regulated B+ and heater supplies for the tubes. (Aside from noise on the HT, these ears only find anything vaguely cathode follower related to sound good when regulated, so it's not just for the sake of quiet.)

I nearly jumped in on the questions from staki above about ultrasonic noise, but IMD, aliasing, images and noise generated by the sigma-delta modulator, seemed like too many cans of worms to be opened at the same time in one post. ;) Anyway, while I was working on the tube output stage, I did have the chance to look at FFT on the outputs. Now admittedly my decent test gear is in storage and this was with a ten year old digital "sound-card" scope, (complete with useless software user interface that I suspect was designed by someone who has never used a scope), it is pretty obvious that the DAC chip itself has some on-chip filtering built-in. From what I saw I wouldn't bother with any form of external LPF. But having said that, there is always a corner case. From my point of view, using NOS DAC's with tube amps and transformers, where bandwidth ends up being limited....... But a couple of years ago I lent a NOS DAC without LPF on the output to a friend who lent it to a friend. They couldn't figure out why the amp they connected it to kept tripping the protection circuitry and shutting down. Of course, it was solid-state, and turned out the HF -3db point was something ridiculous like 300kHz. It didn't like my DAC! LOL.

Tweeters..... Unless you are talking about super tweeters, most of them tail off pretty significantly after 20kHz. It's a bit of a old wives tail that your blow your tweeters by using a NOS DAC without a LPF.

The bottom line. My modus-operandi, (following the "incident" above with the solid state amp), if I'm going to do any filtering at all, has been to put a small cap in parallel with the I/V resistor, when running into a tube gain stage without feedback, and aim for -3db @ 60kHz or so. Of course, as the OP says, if you do wish to use a "proper" analogue LPF, then up/oversampling is the way to go, but then you bring digital filtering back into the equation, regardless of whether you do it in hardware or software. I was actually quite pleased with what I saw (or rather didn't see) on the output of this DAC, to the point where unless I knew the DAC was going to be used with a solid-state amp that was likely to be a problem, I wouldn't bother with any additional analogue filtering at all. Having said that, my ethos tends to be KISS and that curing one so called "problem" tends to cause another. (And don't forget, assuming a 44.1kHz input, that with a NOS DAC you'll be be 3db down at 20kHz before anyone starts to get all bent out of shape about a stupid amount of attenuation being required at at 24.1kHz and how to avoid that biting into <=20kHz.) YMMV.
 
And don't forget, assuming a 44.1kHz input, that with a NOS DAC you'll be be 3db down at 20kHz

When Doede made the listening tests of his DAC, he was very impressed by the sound of the 176,4 or 192 KHz files. The 44,1 KHz sounded less impressive.

Could the "3 dB roll-of at 20 KHz" for 44,1 KHz files explain the perceived difference in sq ? :rolleyes:
Or could it be that the quantization noise at 44,1 KHz is much more in audio band than the one at 176,4 or 192 KHz ? :rolleyes:
 
Well, if I understand, a IV conversion circuit like OPC did (for an ESS DAC)followed by an active stage (my Borbely opamps or a tubes stage) would not be the perfect choice. That NOS converter has been (probably) "thought" having passive output.

Sorry, not familiar with what OPC did or Borberly opamps.

It doesn't have to be passive I/V. It's up to you. Doede, stated his aims. (And also did his own custom transformer design with Sowter.) One eye on cost and another on simplicity. It's not like you have only +/- 1mA to play with, where you would have to add some kind of gain stage. And the 1794 seems friendly, (0.4% distortion, ISTR Deode said), to a value of resistor allowing a decent output. Transformers, opamps, tubes, jfets..... Or stick with a single I/V resistor. Pick your poison. ;) I have tried a couple of options and I'm sticking with the BCF.
 
Could the "3 dB roll-of at 20 KHz" for 44,1 KHz files explain the perceived difference in sq ? :rolleyes:

Not IMHO.

Or could it be that the quantization noise at 44,1 KHz is much more in audio band than the one at 176,4 or 192 KHz ? :rolleyes:

No. That's not it either.

Don't think it is anything to do with not using a LPF and the images being shunted higher in frequency, either.

I have a theory that I've spoken to Doede about. I think he will be back this weekend and now that he has new test gear toys to play with, I imagine he will be busy testing and measuring.
 
Having had the chance to fiddle with various output stages for the DDAC1794, I've decided to stick with the Broskie Cathode Follower. (I make no apologies for being a tube guy.)

One thing I have found is that there is a benefit to be had using the +/- output from the DAC. Although I have not tried a transformer, (don't have anything suitable to hand), I expect that the benefit would be the same. You could even consider the BCF to be a sort of poor mans "active" tube transformer, converting a balanced output to single ended, but step-down, 2:1. (So if you had 1.2V single-ended from using one of the outputs, you'll still have 1.2V SE using both of the outs via the BCF.) Aside from any on-paper arguments about CMRR and noise cancellation, the bottom line.... more focus and less grain combining the outputs, as opposed to just using one single ended output. (Of course, if you prefer opamps...... ;))

Although I was using a differential gain stage in front of the BCF, I've gone back to the 133R I/V resistors and am just using the BCF to obtain a SE output from balanced.

200V HT from Tom Christiansen 21st Century Maida Regulator to the tubes. Heater supply is regulated too. ECC88's, 220R cathode resistors.
50k grid stopper and feedback resistors. That's all there is to it really.

I also swapped my original "junk" 7812 reg that was powering the main DAC board for a Salas shunt. To be honest, I couldn't hear any difference. Then I tried an eBay reg module based on the LT1764, claiming 40uV of noise. Couldn't hear any difference with that either. So it has stayed in by virtue of being last into the circuit. (I've ordered another to power the WaveIO board.)

I'll put it all into a case over the Christmas holidays.
 

Attachments

  • ddac1794_bcf.jpg
    ddac1794_bcf.jpg
    275.4 KB · Views: 2,380
Hi,

Why don't you have a resistors on the Iout going to a negative Vcc voltage to ensure the negative bias current of -6.2mA at bipolar zero ? Or am I missing some info in the diagram?

Page 25 of data sheet for pcm1794 show this. If you provide this resistor and negative bias current then you can trim the Vout to almost zero across your passive I/V resistor at Bipolar zero. And the you can use larger value of I/V resistor to obtain higher Vout, before THD rises, due to the conducting of the ESD diodes on Iout pins. The voltage on the I/V resistor will actually have a +/- swing with this negative bias resistor.

I use this principle with a negative bias resistor on a modded Asus Essence One, that uses PCM1795, that I have changed to passive I/V, I get very low THD.

What is the DC voltage on your 133ohm I/V resistor at zero output?

Cheers
Flemming
Denmark
 
Last edited:
Why don't you have a resistors on the Iout going to a negative Vcc voltage....

Doede must still be away on business. He can answer for himself. (I suspect that he will say the same as when 1704 was mentioned, needing additional -Vcc supply, that he wanted to avoid an additional power supply.)

What is the DC voltage on your 133ohm I/V resistor at zero output?

Doede is using a Vcc of 8V and a 6.1k ref resistor, resulting in 10mA at zero. Mono mode, so x2, 20mA x 133 = 2.66V. Full scale current is 6.3mA peak per DAC, so swing up from ~2.6V to ~4.4V. (He determined that output clips at 4.5V.)
 
Hey guys, I am wanting to go balanced out from this dac into balanced input into my preamp.

Does that mean I can bypass the coupling capacitor? How about the R load resistors?

cheers,
Tuyen

PS. your temp setup looks really interesting Clive. How does the tube rectification/output stage compare to the standard passive IV output? Do you use a valve preamp as well? IE. will I benefit from going an active tube output stage like yours if I already am using a decent valve preamp?
 
Does that mean I can bypass the coupling capacitor? How about the R load resistors?

My advice would be, unless you know that the inputs to your pre are cap coupled, stick with the coupling caps on the outputs of the DAC.

The I/V resistors? The DAC has current outputs. You need to keep the I/V resistors.

PS. your temp setup looks really interesting Clive. How does the tube rectification/output stage compare to the standard passive IV output? Do you use a valve preamp as well? IE. will I benefit from going an active tube output stage like yours if I already am using a decent valve preamp?

Well, the BCF I'm using although active, isn't really active in as much as it doesn't provide gain. It just combines the balanced outputs into a single output for SE use. In your case, where you'll be using both +/- outputs into your pre, there is no need to add anything if you are happy with the concept of passive output from the DAC board.

The problem with adding active tube stages to DAC's is that without feedback, they better be very low distortion, or you end up in the situation, where the passive out has reasonably low distortion to start with, that you increase it just by adding a tube stage. And without using very high transconductance tubes, you'll probably be adding more noise to the output than is there with the passive out.

IMHO, there is a benefit from using the balanced outputs from the DAC. Without getting into any audiofool clichés, eg. "it seemed like a veil was lifted", I didn't find it subtle. If you have a balanced setup to start with, you don't need to do anything. If you are single-ended, then it is worth thinking about transformers or the BCF.
 
Last edited: