Return-to-zero shift register FIRDAC

...adding the 22uH reduced the noise at high frequencies...
However, it may be substituting one kind of noise for another. Ferrites in a digital reclocker power filter just caused me a LOT of wasted time chasing around sonic problems in Marcel's dac. Why didn't I think to look at the ferrites first? Don't know, but I do know lots of EEs seem to be in denial that it can be a problem. Too many EEs think its just audiophile BS.
 
Unless it is an insanely big permalloy core monster like the WE111C repeater coil or bigger (we are talking transformers that push a kilogram or more (WE111C is 4.5 lbs) you have an issue with that.

Level vs frequency and distortion for an extremely respectable line input transformer (10k:10k)

View attachment 1315502

from:

Jensen JT-11P-1 Transformer Data

As we see, as little as +4dBu / 1.23V input voltage exceeds 0.001% THD or -100dB below 250Hz, which is the music's foundation. Due to the underlying mechanics this HD will be H3 and odd order.
HD goes up pretty linear level for any given frequency.

So for +10dBu (2.46V) distortion will double for a given frequency.

A lot more details on all this can be found here:

https://www.onallbands.com/wp-content/uploads/2019/08/Non-LinearXformerBehavior.pdf



Hmmm, I suspect compared to the main signal, the RF is not material.

You should have a pair cap's on each phase to ground ahead of the transformer. Analyse the complex impedance of the circuit first and make sure it is critically damped or add a snubber on the secondary to kill resonance's.

I find killing a resonance flat sounds a little flat, just 1-2dB ultrasonic peaking seems the sweet spot.

Thor
A quick note on their (Jensen / Cinemag) transformers, the line OP traffos have up to 10 x lower LF distortion when driven appropriately from a very low Z source.
These are multi-filar, sans electrostatic shield so if P-S wide band isolation is required not really suitable.... Is the high winding capacitance in a DAC app your friend or your enemy... ?

I've only used them for Pro Audio applications where they work very well. Ironically I often use the 50 / 50 Nickel Iron core to *add some color.
What did Homer Simpson say -> Neve / other old stuff, the cause of and solution to all our (audio) problems. 🙂

PS Edit - Thorsten, I just read you reply to Mark WRT transformers so you prolly are well aware of all this, still may be useful info for others
 
Last edited:
  • Like
Reactions: gpapag
Regarding zero-field (in the audio band), not sure if that's the issue with the transformer I have. It more like I wonder how the RF is interacting with the high-nickel core. Obviously the core is lossy at RF but the RF is also sort of like tape bias. Too much RF bias is not necessarily better.

A passive pre-filter might be worth trying. Or, even better, I would like to have a one on one with the transformer designer. But that may not happen because of human factors.
Mark, if you can / choose to reveal, are you using Toroid or conventional core transformers? I have a toroidal winding machine coming to the shop and will be experimenting with Toroid signal transformers. Should be interesting.
 
...conventional core transformers...
One prototype unit arrived here not in its mu-metal can, and was promptly sent back to the factory. Not toroid. Not exactly conventional either, but closer to that. Interestingly the Quadratic division of Cinemag office is not to far from here. Same for Pass Labs factory. Mostly a coincidence, but its an interesting place to be around.

Regarding toroids, we don't really want too good HF coupling between primary and secondary. Not above a couple of hundred kHz anyway, if that. So conventional-ish might be fine for this. 🙂

The prototypes here are the first transformers for hi-fi that I actually like the sound of. Besides the step up transformers out now, would like to some of the other goodies being lined up get into production. Should be pretty interesting.

Getting back to Marcel's dac, once I made a little progress with the ferrite power filters on the reclocker board, the fully stock Marcel dac started to sound very precise, punchy, and detailed. The line level transformers are sounding very good too. Today I heard details in familiar music I never heard before. Been a long time since that's happened. Maybe things will get even a bit better once the reclocker circuit is fully sorted. Only then will I know if Marcel's dac needs more development work or not.

Also it may turn out that no, or at least not much, of an active output stage will be needed. The transformers, with or without a little passive pre-filtering, then maybe followed by an output buffer, may serve quite well. Problem with buffers is that the really good ones IME tend to be not all that cheap, easy, and or simple, so it adds to cost (good transformers and good clocks already cost enough). Have to wait a bit and see how things turn out on some of that stuff.

All of this talk reminds me that the reason I decided to focus on dacs when I originally joined the forum was because I thought there was need for better diy dacs to be available at lower cost so that more people could benefit from better sound. Also it seemed like a not-fully mature area of audio to do some work in. Unfortunately what I have found is that really good digital audio is possible, but doing it at low cost is a hard nut to crack. Just doing it really well at all is hard enough.
 
Last edited:
  • Like
Reactions: Terry Demol
A quick note on their (Jensen / Cinemag) transformers, the line OP traffos have up to 10 x lower LF distortion when driven appropriately from a very low Z source.

True. Even lower if driven with negative impedance to compensate primary winding resistance.

In a DAC that often doesn't apply however.

I don't particularly like to comment on others products, but here an illustration what a not so well implemented transformer in voltage mode output on a DAC does:

https://www.stereophile.com/content/ps-audio-perfectwave-directstream-da-processor-measurements

Is the high winding capacitance in a DAC app your friend or your enemy... ?

Neither. It's just something to take into account.

Ironically I often use the 50 / 50 Nickel Iron core to *add some color.

Yes, high nickel transformers can be quite transparent.

As for "conventional", I would call all laminated core transformer's conventional. Smaller cores are often LL, especially in German designs for microphone levels.

Something like Metglass I'd call unconventional.

Thor
 
However, it may be substituting one kind of noise for another. Ferrites in a digital reclocker power filter just caused me a LOT of wasted time chasing around sonic problems in Marcel's dac. Why didn't I think to look at the ferrites first? Don't know, but I do know lots of EEs seem to be in denial that it can be a problem. Too many EEs think its just audiophile BS.
Then you will have an even bigger problem with the 100uH in Marcel’s filter, don’t you 😊

Hans
 
However, it may be substituting one kind of noise for another. Ferrites in a digital reclocker power filter just caused me a LOT of wasted time chasing around sonic problems in Marcel's dac. Why didn't I think to look at the ferrites first? Don't know, but I do know lots of EEs seem to be in denial that it can be a problem. Too many EEs think its just audiophile BS.
Most EEs acknowledge that any possible issues caused by ferrites in a digital reclocker filter are measurable. Unless measurements are shown to prove otherwise most EEs draw a conclusion that it is just audiophile BS.
 
Unless measurements are shown to prove otherwise most EEs draw a conclusion that it is just audiophile BS.

Then the EEs may be more arrogant than is good for them. Depending on the phase of the noise, it can look like flat white noise in the frequency domain. That's true even it its an infinite amplitude Dirac pulse.

BTW, nobody claimed anything was immeasurable. Another strawman to knock down.
 
However, it may be substituting one kind of noise for another.

Unless it is receiving power from somewhere or transmitting power to some cold place, a passive two-terminal device without any bias current should produce the thermal noise corresponding to the real part of its impedance, no more and no less. Otherwise you could make a fridge that requires no power by connecting it to a normal resistor.
 
Splitting the 22uH seems like an excellent idea to let permanent DC current flow into two 10uH inductors.
But what I don’t get is the benefit of having permanent DC current flowing into an inductor.

Looking at a BH curve the more DC current, the more non lineair the inductor becomes and the larger the hysteresis for a superimposed signal.
So to me it would seem to be a better idea using inductors in an audio chain that are rated for much higher currents, but still letting flow as little DC current as possible.
What is wrong in my assumption ?


Hans
 
Unless it is receiving power from somewhere or transmitting power to some cold place, a passive two-terminal device without any bias current should produce the thermal noise corresponding to the real part of its impedance, no more and no less. Otherwise you could make a fridge that requires no power by connecting it to a normal resistor.
True. But in normal operation the inductors in the filter will have some voltage applied to them due the audio signals being processed. Thus some current will be flowing.
 
I doubt that it would make anything more linear,

Marcel, the issue is the hysteresis loop on the magnetic response of the inductor core. Think of it as something a little akin to crossover distortion. It generally can be measured.

By applying DC current bias we move to one of the sides of the hystresos loop and never "cross zero".

This is similar to how SE output transformers manage lower distortion than those in PP Amplifiers.

The 22uH if ferrites cored will add distortion. Air core is better.

Switching to a classic folded cascode FDA improves stability and can improve noise. Zetex has some super low noise SMD PNP's.

Using N-Channel MOSFETs in the folded cascode has a few advantages and improves the circuit.

I suspect 10kHz OLG will end up similar or better dir FCFDA.

But you could replace the 22 uH inductor with two inductors of half the value if you like (10 uH is close enough)

IF cored, probably worth doing. FWIW, I tend to use planar PCB coils at these values, air cored, with 4 layers you get a fair bit inductance.

Thor
 
Thor's suggested inductors and biasing does not seem to work well in practice:

It works very well. The Unit measured in SP had a firmware that escape (I did not test it or greenlight it) and as all the settings (tube or J-Fet input, amounts of NFB are under software control.

For example, the system was always operating in "inner loop feedback" that is the input stage is excluded from the NFB Loop.

That said, the analogue stage is not designed with huge amounts.of NFB and the DAC Chip in use has its own relatively high HD, which inductors however exceeded.

Thor