Review and Measurements of Wyred4Sound DAC-2v2 SE DAC

[jackenhack]

Since everybody here seems smarter then well established audio designers, why don't you start building your own dacs?
I think the owner was pretty clear here, the Dac sounds good, better then well measuring ones, so it seems we are just measuring the wrong things..

Well established audio designer? Who? A hack going from doing "gild the lily" kind of engineering to doing final implementation? If you can't muster up better or equal quality than the ESS DAC sabre example board, you have failed.

 

[vitalii427]

If there is any interest in measuring a Devialet Expert 120, please let me know. Shipping may be pricey...

I’m also interested very very much!

 

[finneybear]

Hello all,
Just wanted to drop a note how we have many things going against us here. For one, understand that these DACs have discrete output stages. The ones you rate so highly and compare it to are Op-Amp implementations which have an incredible amount of local and global feedback to provide good numbers. This doesn’t come free and that is why we opt to not use them where possible. Im not posting here to debate on the legitimacy of Femto clocks, naked Z-Foil resistors, discrete output stages etc. We all have opinions and if we dont agree, I understand that and we can move on. Given all this, we must remember that a system will only be as revealing as the weakest link!

First of all, welcome on board! It's always nice to talk to designers directly.

One thing I have to stress is I/we here are not trying to give you a hard time. I am actually more of a supporter of yours. It's not easy to run an audio equipment company with the braveness on trying things differently. I talk here just to see whether my humble opinions can make your products better (or not! LOL)

Not sure whether those will be of any benefits to you....

1. Why not just put 9038 on the analog output board and make the digital part another board? To make space, split the big transformer into two. One for analog, one for digital?

2. 9038 Pro eats a lot of current. It needs a fast regulator with the best load regulation. Feed the 9038 with local LT3045 and LT3042. One for analog, one for digital. Keep the discrete regulator for analog outputs only. The Crystek needs its own dedicated regulator. 970X is a bit better than 575. Pretty much the same price but a bit bigger. If you move to two transformers, you will have larger board space anyway.

3. I like this discrete IV/OP LPF/discrete buffer structure a lot. As far as the spikes are only harmonics, I am fine with no NFB. Human ears are not measurement instruments anyway. The key is execution here.

4. The 1KHz *booming*. From the photo I am not sure whether the I/V is a two K170 pairs or is derived from the Pass IV? At least try to do a good matching between left and right channels. For this price range, you have to do active device matching as best as you can. Ever thought about using LSK489? How about LSK170? The cost is high yet you have already spent so much budget on foil resisters? Another thing is compared to other DAC chips, 9038 has a lot of very high frequency noises. They are too much for K170 and would bring non linearity to it. This is why you see the 1KHz booming. Try to see whether you can put a small silver mica for decoupling around the input of IV. It will remove some HF noise and hopefully reduce the booming.

5. Since the IV is band limited and has removed some HF noises, your LPF looks decent.

6. The output buffer. I am not sure about the value of using foil here. How about Caddock's precision film? It looks cool, has good marketing value, too, and you can save half of the budget for something else?

7. I am not fond of Dale resistors. They have a bit of thermal noises. If you adopted them for marketing purpose. fine with me.

8. How about adding an AKM4137 to upsample everything below to DSD256? Not only you get better sound and the measurement number will look good. It's not a cheating as to those doing measurements, seldom they will open the box to see what's inside. Hey, Chord is doing the same thing, upsample everything to high rate DSD. Why cant you? Then you can get both good sound and good test numbers to brag about?

 

[finneybear]

Amir’s reviews really are incredibly valuable aren’t they?
Didn’t anyone measure anything from 1974 ...
Keith

We did! And a lot! PCM1974, AD1955, PCM63, TDA1541, TDA1547, etc, You name it, most likely we had already done it. In fact, I am listening to a pocket PCM1794 DAC now.

Here is the test of an AD1955 DAC we designed 10 years ago. Remembered at that time we did not have those super voltage regulators, just plain old 78XX and 317:

Notice other than harmonics, all noises are below -140dB with a floor around -150dB. And this is a 10 years old design!
Even more, this thing has discrete IV, discrete LPF, and discrete buffer. No OPs!!! No negative feedback!!
Yes, you can still get very good measurement numbers with discrete IV!
Those harmonics can be reduced a lot by adding negative feedbacks, BTW.

And this is done with AD1955! With all the hypes around ESS chips, the plain old AD1955 can be just as good! And what a great sounding DAC chip it is!!!

 

[amirm]

Notice other than harmonics, all noises are below -140dB with a floor around -150dB. And this is a 10 years old design!

There is FFT gain there so what is seen is not the actual noise floor. Depending on FFT length used, it could be as much as 30 dB.

 

[finneybear]

There is FFT gain there so what is seen is not the actual noise floor. Depending on FFT length used, it could be as much as 30 dB.

It's already compensated there, we did some trick there. It's an old AP. Ignore those typing.

 

[gvl]

Ok, so we've established that paying big bucks in hopes to get better performance is money down the drain as close to SOTA devices can be had for $100 give or take and they sound close enough to the point there are no audible differences given a level-matched controlled experiment. This doesn't necessarily fully match my subjective experiences but I'm comfortable with this notion as it makes perfect sense. Now, what about paying big bucks for a DAC that produces a specific sound signature that's is readily audible and happens to be preferred over that of a transparent DAC? Related to this, can we take a transparent DAC and make it sound exactly like a non-transparent DAC by adding DSP that simulates the latter? This is not an empty interest, I have a not outrageously but still expensive non-transparent DAC which I happen to like, and if I could make my Khadas board sound like it I'd be happy to unload it.

 

[rebbiputzmaker]

So I did some subjective listening tests. I grouped the Wyred4sound Dac 2V2 SE with my Topping DX3 Pro in Roon so they play in sync. I hooked up the (RCA) unbalanced output of the DX3 Pro to Massdrop THX AAA 789. And to give more of a benefit of doubt, I used XLR connection from Dac 2V2 SE to the THX 789. This then allowed me to switch between them using the input switch on the THX 789. I matched levels using Audio Precision.

I then went through my reference clips across many genres. There is no audible difference. Matching levels is key. I can make either DAC sound better by increasing its volume.

What are you listening for?

 

[amirm]

What are you listening for?

Meaning of life. How about you???

 

[stunta]

Related to this, can we take a transparent DAC and make it sound exactly like a non-transparent DAC by adding DSP that simulates the latter?

Probably, if you could reverse engineer the alteration made by the non-transparent DAC. It might be actually be DSP.

I prefer to leave the art to the artists which is why I don't wear sunglasses at museums.

 

[gvl]

Probably, if you could reverse engineer the alteration made by the non-transparent DAC. It might be actually be DSP.

I prefer to leave the art to the artists which is why I don't wear sunglasses at museums.

There is no DSP, in fact there is no digital processing that alters the sound at all. I'm thinking along the lines of making digital filters as it is done for room correction, however not sure if EQ is enough as also need to match distortion profile.

 

[Blumlein 88]

It's already compensated there, we did some trick there. It's an old AP. Ignore those typing.

What do you take us for? What's the size of the fft or do you think we'll believe you have sub -140 db noise?

If it's an old AP, probably 42db fft gain. So maybe -110 db noise for? Pretty good.

 

[finneybear]

What do you take us for? What's the size of the fft or do you think we'll believe you have sub -140 db noise?

If it's an old AP, probably 42db fft gain. So maybe -110 db noise for? Pretty good.

Cant you see the 3rd harmony is already -70db here? Now consider then gain.... LOL

Old AP is bound by its ADC and front end amp, some compression here and there but there are ways to get around it.

 

[rebbiputzmaker]

Meaning of life. How about you???

I guess when one needs to listen to dacs to find the meaning of life, life has little meaning.
I listen to the conversations of free range chickens.

 

[finneybear]

What do you take us for? What's the size of the fft or do you think we'll believe you have sub -140 db noise?

If it's an old AP, probably 42db fft gain. So maybe -110 db noise for? Pretty good.

So here is the TDA1547 inside a professional Yamaha D2X:

And here is the famous TDA1541, note the test date is back to year 2009:

My God! The 2nd harmony is already -50db! So based on your FFT gain thesis, what is the real noise floor for this TDA1541?

There are people pushing the performance envelope, OK? It's similar to the situation Mola Mola is running into with the current AP. It's already beyond what AP can do. And they have to figure out a solution.

 

[gvl]

We did! And a lot! PCM1974, AD1955, PCM63, TDA1541, TDA1547, etc, You name it, most likely we had already done it. In fact, I am listening to a pocket PCM1794 DAC now.

Here is the test of an AD1955 DAC we designed 10 years ago. Remembered at that time we did not have those super voltage regulators, just plain old 78XX and 317:

View attachment 21143

Notice other than harmonics, all noises are below -140dB with a floor around -150dB. And this is a 10 years old design!
Even more, this thing has discrete IV, discrete LPF, and discrete buffer. No OPs!!! No negative feedback!!
Yes, you can still get very good measurement numbers with discrete IV!
Those harmonics can be reduced a lot by adding negative feedbacks, BTW.

And this is done with AD1955! With all the hypes around ESS chips, the plain old AD1955 can be just as good! And what a great sounding DAC chip it is!!!

The fact you achieved -150dB with the lowest theoretical noise floor of -144dB for a 24-bit signal is nothing short of impressive. Not.

 

[finneybear]

The fact you achieved -150dB with a maximum theoretical noise floor of -144dB for 24-bit signal is nothing short of impressive. Not.

Many current DACs can reach a -150dB noise floor as well. It's FFT ok? Dont be confused, LOL

 

[gvl]

Many current DACs can reach a -150dB noise floor as well. It's FFT ok? Dont be confused, LOL

When nothing is playing, maybe. With the presence of a signal there is quantization noise, with 24 bits playing -144dB down is best case scenario.

 

[finneybear]

Ok, so we've established that paying big bucks in hopes to get better performance is money down the drain as close to SOTA devices can be had for $100 give or take and they sound close enough to the point there are no audible differences given a level-matched controlled experiment. This doesn't necessarily fully match my subjective experiences but I'm comfortable with this notion as it makes perfect sense. Now, what about paying big bucks for a DAC that produces a specific sound signature that's is readily audible and happens to be preferred over that of a transparent DAC? Related to this, can we take a transparent DAC and make it sound exactly like a non-transparent DAC by adding DSP that simulates the latter? This is not an empty interest, I have a not outrageously but still expensive non-transparent DAC which I happen to like, and if I could make my Khadas board sound like it I'd be happy to unload it.

Yes, level matching is one major trick HiFi shops used to confuse people. Even accurate level matching is not really fair. There are other factors involved.

Here is the Khadas chart:

It has an impressive performance for its price, and more than good enough for most people.
Still, I can see problem in this chart and I will not call it transparent. Some people may still not like it.
It also depends on system matching.

And no, you cant fix the problem with DSP.

 

[gvl]

Yes, level matching is one major trick HiFi shops used to confuse people. Even accurate level matching is not really fair. There are other factors involved.

Here is the Khadas chart:

View attachment 21149

It has an impressive performance for its price, and more than good enough for most people.
Still, I can see problem in this chart and I will not call it transparent. Some people may still not like it.
It also depends on system matching.

And no, you cant fix the problem with DSP.

I don't want to fix the problem. I want to recreate the problem.