I guess it is more up to how much the digital source pollute the common ground. galvanic isolation would be solution here.
LT3042 makes a poor AVCC supply, but it can be used as a low-noise voltage reference for a proper AVCC supply. Makes a BIG difference in sound quality to do it right, just sayin'
LT3042 makes a poor AVCC supply, but it can be used as a low-noise voltage reference for a proper AVCC supply. Makes a BIG difference in sound quality to do it right, just sayin'
I mean yeah, im just replacing the LTC6655 with it, is that what you meant is bad? If so then ill have to get an LTC later down the road.
Honestly this whole thing spelt more like a practice in soldering than something that would improve the board.
They didnt have several resistor values (they didnt even have 1k) needed for the I/V so i had to mix and match parallel and series to get the right values, then measure with multimeter and try to get it as close as i can. All resistors 1%, didnt have 0.1%. No LME49720 so NE5532P, etc..
Still, i have no regrets, I've asked too many questions in the thread that i cant not do something with it now.
Using it as a substitute for LTC6655 should work okay, as discussed recently here in the thread.
Just want to make sure no one reading the thread gets the idea that LT304x LDO regulators might make a good AVCC power supply without opamp buffering.
Of course this approach to I-V will work.
How the AK4499 reacts to voltage swing at the DAC output is yet to be seen, or more importantly, heard (and measured).
here will almost certainly be added measured distortion and some kind of sonic impact.
To make things more complex, there is also the issue of the AK4499's extra
feedback connection. My guess is to just join the IOUT and OPIN terminals
together.
Also it's a fair guess that a low value load resistor with a step up transformer will probably work best to minimize the DAC's voltage swing and whatever
impact it has. Also consider the OP Z of transformer secondary is square of turns ratio x IP Z. So if you use, for example a 1:4 ratio transformer then the
OP Z will be DAC OP Z in parallel with load R x 16.
At the end of the day this *could work really well, BUT it will be a case of finding the best compromise of parameters that are more or less fighting
each other. The higher the transformer turns ratio, generally, the more colored it will be (distortion). But the higher the turns ratio will mean lower
DAC OP voltage swing which will mean less DAC distortion. The higher the turns ratio will also throw away dynamic range (S-N ratio).
So yes, this could work very well but there are a lot iterations to try.
You've got some work ahead of you!
I was thinking about making a simplified PCB for the 4499 chip operating in pin control mode with direct current output, seems like the most economical and efficient way (aside from the eval board) to experiment with AK4499. However long it takes for Chinese boards to become available chances are we will be bypassing whatever attempt they made at an active I/V on the board, and tinkering with the board is preferably kept to a minimum with such an expensive chip.
The chip is similiar in layout to the previous v out AKM DACs.
The only confusing part so far is feedback connection you mention,
OPIN is described as ''Common voltage input pin'' yet connected as an output in exactly the place you'd expect IOUT to be connected.
The chip is similiar in layout to the previous v out AKM DACs.
The only confusing part so far is feedback connection you mention,
OPIN is described as ''Common voltage input pin'' yet connected as an output in exactly the place you'd expect IOUT to be connected.
laserscrape, this time I will disagree with Terry. So far as I am aware, neither of you have any idea how good AK4499 can sound at its best (including for the output stage).
Unless you have deep pockets to risk damaging dac chips and or don't mind risking frequent pops in the sound, I would not recommend playing around tying pins together and trying voltage output.
Also, pin mode does not support DSD at all, nor many other setting options. No idea why they bothered to include it.
If one sticks with the design AKM had in mind, but also get the I/V opamp power supply removed from the general purpose +-15v rails, it can sound better than any Sabre dac you have heard. On the other hand, you can easily make it sound worse that many Sabre dacs if enough care is not taken. Regarding I/V opamp power, I think AKM was afraid of people blowing dac chips if I/V is on separate power (a warning is included in the data sheet).
Easy enough to make it fail-safe though, and the complexity is not much more than needed for a very well executed Sabre design. That level of complexity is mainly a problem for consumer markets that can't afford the cost of it. That ESS made Sabre able to work in so many market settings was very smart for helping sell lots of chips. It also meant that the best sounding designs had to turn off most of the Sabre features, such as internal ASRC, internal PCM filtering, and internal SPDIF receiver.
Guess we will see how an economy version of AK4499 sounds when Topping D-90 comes out. Other than that, some of the major Japanese manufacturers have decided to go with their own dac solutions. They have to do something besides stay with Sabre sound forever, and AK4499 is just too complex for many uses. On the other hand, FPGA prices are coming down, and tools like Matlab are getting better, so some companies that can afford DSD engineers on staff will be doing more of rolling their own dac solutions. They have to do something, after all. DAC technology is been mostly static for too long, and ESS secrecy has badly slowed down advances in design knowledge.
Regarding AKM, now that they have shown they can make a better sounding dac than Sabre they need to come out with one that is about as good as AK4499, but simpler for designers implement into a product. Maybe a 2-channel part. We will have to wait and see.
Unless you have deep pockets to risk damaging dac chips and or don't mind risking frequent pops in the sound, I would not recommend playing around tying pins together and trying voltage output.
Also, pin mode does not support DSD at all, nor many other setting options. No idea why they bothered to include it.
If one sticks with the design AKM had in mind, but also get the I/V opamp power supply removed from the general purpose +-15v rails, it can sound better than any Sabre dac you have heard. On the other hand, you can easily make it sound worse that many Sabre dacs if enough care is not taken. Regarding I/V opamp power, I think AKM was afraid of people blowing dac chips if I/V is on separate power (a warning is included in the data sheet).
Easy enough to make it fail-safe though, and the complexity is not much more than needed for a very well executed Sabre design. That level of complexity is mainly a problem for consumer markets that can't afford the cost of it. That ESS made Sabre able to work in so many market settings was very smart for helping sell lots of chips. It also meant that the best sounding designs had to turn off most of the Sabre features, such as internal ASRC, internal PCM filtering, and internal SPDIF receiver.
Guess we will see how an economy version of AK4499 sounds when Topping D-90 comes out. Other than that, some of the major Japanese manufacturers have decided to go with their own dac solutions. They have to do something besides stay with Sabre sound forever, and AK4499 is just too complex for many uses. On the other hand, FPGA prices are coming down, and tools like Matlab are getting better, so some companies that can afford DSD engineers on staff will be doing more of rolling their own dac solutions. They have to do something, after all. DAC technology is been mostly static for too long, and ESS secrecy has badly slowed down advances in design knowledge.
Regarding AKM, now that they have shown they can make a better sounding dac than Sabre they need to come out with one that is about as good as AK4499, but simpler for designers implement into a product. Maybe a 2-channel part. We will have to wait and see.
Last edited:
A little more, if I may...
Regarding signal transformers, so far I can't find any reason to justify using them in a good reproduction system unless one is trying to solve ground problems (which pretty much everyone has). Best solution IME is still a used Monster HTPS 7000 MkII. It was designed mainly by Richard Marsh with some work done by Demian Martin (Audio1). Not too long ago Richard posted the schematics in the Blowtorch thread. Only thing that might be as good or better is one of those regenerator things that are also no longer made.
Regarding the Monster unit, I have everything in my test system hooked up to it. That includes computers, Neurochrome HP-1 headphone amp, all the AK4499 power supplies and the scope (when used). No ground problems at all, it is very clean. Every signal transformer I have tried sounds worse that no transformer, and that includes a couple of transformers Jam gave me that were used in some high-end audio product. They always produce audible IMD on vocal harmonies, which if you can't hear that means you have other problems to work on. And, they never make it sound better.
That said, Jam tells me he is working on a new high end product that will have transformers that he says sounds extraordinarily good, better than what he designed in his last job at you-know-where. I haven't heard it yet, so no opinion from me now. Assuming I will hear it at some point before it is released as a product, I promise to let you know if I have changed my opinion about transformers then. One issue Jam has pointed out with the power conditioners I use and the regenerator he uses is that they can't handle enough current to use them with really big power amps. In a similar vein, Howie Hoyt who designs recording studio and radio station studio electrical systems says the Monster unit I mentioned is the best solution for home use, but studios require larger grounding grounding solutions that have to be custom engineered, and they do use signal transformers for ground issues in those big complex system environments despite knowing the drawbacks that signal transformers add distortion. They have no other choice, basically. However, most of us in home use are not in that situation.
Regarding signal transformers, so far I can't find any reason to justify using them in a good reproduction system unless one is trying to solve ground problems (which pretty much everyone has). Best solution IME is still a used Monster HTPS 7000 MkII. It was designed mainly by Richard Marsh with some work done by Demian Martin (Audio1). Not too long ago Richard posted the schematics in the Blowtorch thread. Only thing that might be as good or better is one of those regenerator things that are also no longer made.
Regarding the Monster unit, I have everything in my test system hooked up to it. That includes computers, Neurochrome HP-1 headphone amp, all the AK4499 power supplies and the scope (when used). No ground problems at all, it is very clean. Every signal transformer I have tried sounds worse that no transformer, and that includes a couple of transformers Jam gave me that were used in some high-end audio product. They always produce audible IMD on vocal harmonies, which if you can't hear that means you have other problems to work on. And, they never make it sound better.
That said, Jam tells me he is working on a new high end product that will have transformers that he says sounds extraordinarily good, better than what he designed in his last job at you-know-where. I haven't heard it yet, so no opinion from me now. Assuming I will hear it at some point before it is released as a product, I promise to let you know if I have changed my opinion about transformers then. One issue Jam has pointed out with the power conditioners I use and the regenerator he uses is that they can't handle enough current to use them with really big power amps. In a similar vein, Howie Hoyt who designs recording studio and radio station studio electrical systems says the Monster unit I mentioned is the best solution for home use, but studios require larger grounding grounding solutions that have to be custom engineered, and they do use signal transformers for ground issues in those big complex system environments despite knowing the drawbacks that signal transformers add distortion. They have no other choice, basically. However, most of us in home use are not in that situation.
Pin control mode is useful if you're just interested in a PCM DAC or cant program, true that it is really strange they would include it in this DAC, no one is ever going to release an Ak4499 based DAC that doesnt support DSD but its a good option to have.
Although I wasnt a fan of PCM to DSD conversion compared to PCM with the Sabre DACS it may be different for AKM, with filter and volume bypass function of Ak4499 its probably excellent, only thing then is that it requires a good analogue attenuator. The problem with DSD in general is relying on software when the only one worth using costs a fortune (HQPlayer). other than that most features in register control can be done without.
Although I wasnt a fan of PCM to DSD conversion compared to PCM with the Sabre DACS it may be different for AKM, with filter and volume bypass function of Ak4499 its probably excellent, only thing then is that it requires a good analogue attenuator. The problem with DSD in general is relying on software when the only one worth using costs a fortune (HQPlayer). other than that most features in register control can be done without.
Last edited:
Happy to help/assist anyone who wants to program AK4499. No NDA on this one, so we can freely talk all we want.
Regarding DSD, the eval board comes with AK4137, same SRC I used with dac#2. It does a pretty good job of conversion to DSD256, and cheap ones that work fine can be had on ebay or aliexpress.
You didn't appear to read my post.
I stated "How the AK4499 reacts to voltage swing at the DAC output is yet to
be seen, or more importantly, heard (and measured). There will almost
certainly be added measured distortion and some kind of sonic impact."
What I'm stating here is pretty clear.
WRT joining the current OP / FB pins, these are almost certainly internally
connected. Perhaps you could pop a multimeter across these pins and
confirm this for Laserscape.
Having voltage swing at the DAC OP should not damage the DAC. Switched
unity weighted resistor architecture is well defined. It should also not have
bearing on pops or clicks.
BUT - as I clearly stated, there will be a sonic and measured impact from
increased distortion.
The distortion mechanism have been covered elsewhere. John W mentioned
voltage dependent distortion of the internal resistors and I mentioned
previously logic switch generated distortion.
After all is said and done, it's worth noting Iancanada's results using a low
ratio step up transformer on Sabre. From reading that thread he states it's
the best sound achieved so far. The 4499's OP architecture is very similar
to Sabre being unity weighted, switched resistor so it's fair to assume great
results could also be had from 4499 with the right combination.
This is 'DIYaudio' - different people like to entertain different approaches to
all things audio. We should be encouraging and helping them, not
discouraging them.
T
Terry,
I read what you said, what I meant was that you don't know how good AK4499 can sound without a transformer as a starting point, so you could then know just how bad voltage mode messes it up.
Regarding Iancanada, I don't take that story as an encouraging sign like you seem to.
There is switching of the feedback I/V feedback loop inside AK4499, which is briefly described in the datasheet, IIRC. They don't say exactly why they are doing it, but their way of dealing with pops especially for DSD seems very different from ESS. In some cases it is up to the register programmer to put it in software-reset before doing something that would produce a pop. If using volume bypass DSD there may be more issues, don't recall at the moment.
Most of my recent and ongoing experimentation involves power supply effects on sound quality. All those power supply substitution banana jacks and jumpers on the eval board are there for a reason. Designing a dac board without understanding what the power supplies do to the sound would likely not be a very satisfying experience.
I read what you said, what I meant was that you don't know how good AK4499 can sound without a transformer as a starting point, so you could then know just how bad voltage mode messes it up.
Regarding Iancanada, I don't take that story as an encouraging sign like you seem to.
There is switching of the feedback I/V feedback loop inside AK4499, which is briefly described in the datasheet, IIRC. They don't say exactly why they are doing it, but their way of dealing with pops especially for DSD seems very different from ESS. In some cases it is up to the register programmer to put it in software-reset before doing something that would produce a pop. If using volume bypass DSD there may be more issues, don't recall at the moment.
Most of my recent and ongoing experimentation involves power supply effects on sound quality. All those power supply substitution banana jacks and jumpers on the eval board are there for a reason. Designing a dac board without understanding what the power supplies do to the sound would likely not be a very satisfying experience.
Eventually learning how to do some basic programming of a DAC would be great, better to leave till the DAC gets a stronger DIY following and there is more of an audience who can benefit.
I never understood the use of the SRC, if the DAC essentially converts to DSD already. the idea with pre-conversion to DSD was using powerful PC processor to run more complex filter and modulator algorithms.
The DAC is limited in its processing power and cant see how SRC would be any different, is that off the mark?
IME I think a transformer would bottleneck the Ak4499.
I found 1:4 transformer distortion unacceptable with 9038Pro DAC with little improvement over just a resistor, the 9038Pro did not like anything but op amps for I/V.
If using only a resistor for I/V for AK4499 does not give acceptable performance then I think with a trafo either the ratio will be too low to be effective or the distortion will be too high.
The TDA1387 worked decently with resistor I/V and that trafo happened to be a better match for it. However with TDA1387 paralleling chips and lowering the I/V resistor reduced the positive effects from the trafo step up, at 20 or so chips the DAC sounded better with just a resistor than with the trafo, at that point trafo distortion was greater and more offensive. The TDA1387 sounded its best like that, even OPA1612 I/V couldnt compare.
More than likely the 4499 will be the same as 9038Pro, but they arent identical so who knows.
to be clear the board would basically be a custom breakout board just for the chip. Power supplies, I/V etc. would be external.
Last edited:
I don't really want to deflect this thread way OT but I'd like to have clarification here - you were running 20 paralleled TDA1387s in conjunction with a 1:4 step up trafo? If so I suspect you were ending up with too much output voltage variation on the DACs. I'm using a 1:25 trafo at present, with 36 chips - a 1:4 trafo would need far fewer chips. You had the trafo giving the standard 2VRMS output?
Yes, off the mark. SRC4137 does much like what HQ Player does for sound quality, there are just fewer processing algorithms to choose from and the most computation intensive ones are not included. Yet, it can sound quite good if well implemented into a dac design.
Regarding dac chips like Sabre and AK4499, they do not exactly convert everything to DSD anyway as you seem to think. If you have a Sabre data sheet maybe take a look at the filter graphs for PCM and DSD. If the dac converted everything to DSD then all the interpolation/reconstruction filters would be of the DSD type which they are clearly not.