For something like the clock or the DACs reference voltages, I think the LT304x is better, at least theoretically. It has lower noise at low frequencies (with 22uF Cset cap).
For the op-amp supplies and other analog supplies, I doubt there is any real difference between the two, they are both great.
They do have lower leakage and lower ESR, but I don't think they will actually outlive polymer or MnO2 tantalum caps. The hybrids still have a partially liquid electrolyte from what I read. Probably doesn't matter too much which you use for bulk power supply applications.
I would personally stick with the MLCC, or at least, add a cap but don't remove them. They have much lower ESL and lower Z at high frequencies. I haven't ever experienced a microphonics issue with X7R on power rails.
1OK sounds like a good idea to stick with MLCC - thanks Chris For the record these are the Tantalum's I have had good results with. T520B476M010ATE035 KEMET | Mouser Australia
The KEMET Organic Capacitor (KO-CAP) is a solid electrolytic capacitor with a conductive polymer cathode capable of delivering very low ESR and improved capacitance retention at high frequencies. KO-CAP combines the low ESR of multilayer ceramic, the high capacitance of aluminum electrolytic, and the volumetric efficiency of tantalum into a single surface mount package. Unlike liquid electrolyte-based capacitors, KO-CAP has a very long operational life and high ripple current capabilities.
The KEMET Organic Capacitor (KO-CAP) is a solid electrolytic capacitor with a conductive polymer cathode capable of delivering very low ESR and improved capacitance retention at high frequencies. KO-CAP combines the low ESR of multilayer ceramic, the high capacitance of aluminum electrolytic, and the volumetric efficiency of tantalum into a single surface mount package. Unlike liquid electrolyte-based capacitors, KO-CAP has a very long operational life and high ripple current capabilities.
Last edited:
Yeah, there's nothing wrong with them on their own. Sciencesky might want to get away from them because tantalum (from coltan) is considered a conflict mineral by some.
They don't have many advantages over the aluminum polymer caps now though. Maybe ultimate longevity and capacitance for a given volume (I'd have to check on that one).
They don't have many advantages over the aluminum polymer caps now though. Maybe ultimate longevity and capacitance for a given volume (I'd have to check on that one).
I like LDOs just fine for the digital supplies. Maybe even for certain of the analog supplies, but not all. Based on the eval board design, looks like most of the power supplies aren't all that critical. However, those that have directly audible effects are not LDO types. I'm sure Chris will disagree with AKM's decision, as well as with my opinions. That's okay. He's still a good guy, IMO.
Last edited:
I’d prefer if they specified the requirements for the pin instead of just dictating a solution. I haven’t seen the eval board schematic, but I guess it’s similar to the old ESS op-amp buffered reference method from what I read.
I would keep in mind that there are orders of magnitude difference between LT3045 and a regular decent LDO. The LT3045 has lower voltage noise than many op-amps used in audio if you use a large set pin cap. It’s <=2nV/rtHz from 200 Hz and up with a 22uF cap there. Plus, you know it’s going to be stable as it’s designed and compensated to have medium to large ceramic caps on its output. Last I checked, the specs are as good or better than the dedicated ESS low noise regulator chip they now recommend instead of the op-amp solution. It’s the only voltage regulator I’d use in this way for these DACs. It is certainly possible they don’t use it because designers would assume they can substitute any regulator for another, and no one likes being restricted to a single source.
I did say it’s safer to just do what they do if you want to guarantee datasheet performance.
I would keep in mind that there are orders of magnitude difference between LT3045 and a regular decent LDO. The LT3045 has lower voltage noise than many op-amps used in audio if you use a large set pin cap. It’s <=2nV/rtHz from 200 Hz and up with a 22uF cap there. Plus, you know it’s going to be stable as it’s designed and compensated to have medium to large ceramic caps on its output. Last I checked, the specs are as good or better than the dedicated ESS low noise regulator chip they now recommend instead of the op-amp solution. It’s the only voltage regulator I’d use in this way for these DACs. It is certainly possible they don’t use it because designers would assume they can substitute any regulator for another, and no one likes being restricted to a single source.
I did say it’s safer to just do what they do if you want to guarantee datasheet performance.
Last edited:
I don't know what kind of filtering you've tried on the PoE input, but a common-mode choke is a good idea. There are also some nice 3-terminal "feedthrough" caps that Murata makes we have used to good effect. Could also look into using LDOs with better rejection up to 1 MHz, which seems to be around where the best IC regulators top out.
The attenuator should be fine for your use case as clay-speakers said. In theory you can improve on digital volume control, but in practice there are a lot of tradeoffs and I think digital attenuation is the lesser of all evils.
I know you're set on the AKM converters, and they are good, but you might consider doing a "cheaper" version with one single ES9038. You get 8 channels with one IC and still overkill-level performance. The SNR/DNR is what improves in mono mode, and I genuinely have no idea what I would do with that much SNR
.DAC Components Placement
The attached picture of the PCB for the A&ultima SP2000 is interesting: it shows that two AK4499EQ chips can be mounted side-by-side fairly close to each other. Of course, in our dual-mono DAC brick, each AK4499EQ will be mounted on its own PCB, and the layout will be different (left/right instead of top/bottom). This layout is justified by the attached connection diagram taken from the publicly-available datasheet. It shows that most of the components sit on the West and East sides of the chip (note that when the indexing dot is at the top-left corner, the chip's label reads vertically from bottom to top, as shown on the A&K board). Furthermore, the evaluation board shows that the only large components that need to be mounted on the North and South sides are four large capacitors. The chip being 16mm wide, we should have no problem populating them on a 35mm wide board, with board-to-board headers being placed on the West and East sides of the chip (the 35mm wide sides of our 70mm × 35mm board). With that design, the four regulators will also be mounted on the West and East sides of the chip, close to the four operational amplifiers. Doing so, we should probably mount the two DAC boards directly onto the XMOS board (we will use one XMOS chip to feed two AK4499EQ DACs).
The attached picture of the PCB for the A&ultima SP2000 is interesting: it shows that two AK4499EQ chips can be mounted side-by-side fairly close to each other. Of course, in our dual-mono DAC brick, each AK4499EQ will be mounted on its own PCB, and the layout will be different (left/right instead of top/bottom). This layout is justified by the attached connection diagram taken from the publicly-available datasheet. It shows that most of the components sit on the West and East sides of the chip (note that when the indexing dot is at the top-left corner, the chip's label reads vertically from bottom to top, as shown on the A&K board). Furthermore, the evaluation board shows that the only large components that need to be mounted on the North and South sides are four large capacitors. The chip being 16mm wide, we should have no problem populating them on a 35mm wide board, with board-to-board headers being placed on the West and East sides of the chip (the 35mm wide sides of our 70mm × 35mm board). With that design, the four regulators will also be mounted on the West and East sides of the chip, close to the four operational amplifiers. Doing so, we should probably mount the two DAC boards directly onto the XMOS board (we will use one XMOS chip to feed two AK4499EQ DACs).
Attachments
The attenuator should be fine for your use case as clay-speakers said. In theory you can improve on digital volume control, but in practice there are a lot of tradeoffs and I think digital attenuation is the lesser of all evils.
I know you're set on the AKM converters, and they are good, but you might consider doing a "cheaper" version with one single ES9038. You get 8 channels with one IC and still overkill-level performance. The SNR/DNR is what improves in mono mode, and I genuinely have no idea what I would do with that much SNR
I totally agree with you, the ES9038 would be plenty enough, but this would kill a lot of the fun. There is something really exciting about using a brand new chip. Also, I really like the idea of using the same supplier for both ADC and DAC converters. And now that we have dramatically simplified the overall design, we should be able to make a small 80mm cube with a Sitara SoC and two AK4499EQ DACs for $300. With a totally open source design, that is something that might get a few people excited. In other words, while the overall project of building this reference setup is batshit crazy, the smaller projects of building this small DAC and its upcoming ADC brother are quite manageable, especially with the amazing amount of collective experience providing guidance on this thread.
If you could not tell already, I'm having a real blast!
Last edited:
I totally agree with you, the ES9038 would be plenty enough, but this would kill a lot of the fun. There is something really exciting about using a brand new chip. Also, I really like the idea of using the same supplier for both ADC and DAC converters. And now that we have dramatically simplified the overall design, we should be able to make a small 80mm cube with a Sitara SoC and two AK4499EQ DACs for $300. With a totally open source design, that is something that might get a few people excited. In other words, while the overall project of building this reference setup is batshit crazy, the smaller projects of building this small DAC and its upcoming ADC brother are quite manageable, especially with the amazing amount of collective experience providing guidance on this thread.
If you could not tell already, I'm having a real blast!
Yeah, I totally understand. If you are going to go to these lengths you might as well. I always have some internal conflict with my practical side - at some level, even really cheap DACs are probably good enough, but if you are going to do it then might as well go all the way. I normally don't post that much on project threads but I thought this one was pretty ambitious and worth watching
.Yeah, I totally understand. If you are going to go to these lengths you might as well. I always have some internal conflict with my practical side - at some level, even really cheap DACs are probably good enough, but if you are going to do it then might as well go all the way. I normally don't post that much on project threads but I thought this one was pretty ambitious and worth watching
That's the spirit. Please keep watching (and posting)!
Reference Voltage
As Markw4 mentioned, the reference board for the AK4499EQ includes "fairly sophisticated Reference Voltage power supplies, similar to Jung-Didden. One supply for each channel." These, alongside the four regulators (NJM78M05) are mounted on the base board, not the DAC board. With that in mind, I keep thinking about whether I should mount all this on the DAC board or on a separate PSU board. Any preference?
As Markw4 mentioned, the reference board for the AK4499EQ includes "fairly sophisticated Reference Voltage power supplies, similar to Jung-Didden. One supply for each channel." These, alongside the four regulators (NJM78M05) are mounted on the base board, not the DAC board. With that in mind, I keep thinking about whether I should mount all this on the DAC board or on a separate PSU board. Any preference?
Normally I would say the DAC board. The long wire / trace to the board normally hurts the load regulation, but if the reference voltage is mostly quiescent, then I guess it doesn't matter.
I'm a little surprised they are using 7805 regulators still on such an expensive board / part...
So am I... This reference board is quite interesting. Outside of the Reference Voltage power supplies, everything is quite basic, which tells me that the chip must be really good on its own. But there are many parameters that one can play with. As clay-speakers and Markw4 mentioned here and on stereonet, it will take a while before people figure out how to best use this chip, but it should take some extra efforts to make it sound really bad.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.