I would say that if ishizeno powers the evaluation board from linear +-15v, and LDO 5v and 3.3v, supplies, and if he can get DSD512 working with HQplayer at its highest quality settings, and if he has a very clean power amp or headphone amp, he can almost have the experience of hearing the dac at, or very close to, its best. Adding some film caps to the +-15v rails helps clean up a bit of less than optimal smoothness from the OPA1612 I/V opamps.
With all the foregoing and some way to cleanly convert the eval board's differential outputs to single-ended, he will probably have an idea of what the dac is capable of, that is, if he has learned how to listen critically like a very experienced recording engineer. (If he has not learned how to listen critically properly, then there will likely be a problem in that musicians and recording engineers likely to use his product may be able to hear any flaws better than ishizeno can.)
With all the preceding, if there is also a good newer higher end AP analyzer and knowledge of how to use it, then ishizeno may have everything needs to reproduce the performance of the eval board in a smaller footprint.
That is, he should be able to hear and to measure (not necessarily in that order) if he still has work to do to get it right. That he should nail it in a short time without a lot of prior experience in that area seems unlikely to me. Of course, I could always be wrong, and expect you will think that I am. That's okay. We can wait and see how it goes.
When the time comes and ishizeno has an eval board, I would be happy do discuss what I have learned up to that point.
Already, there are some issues with some USB to I2S boards playing DSD512 that I am working on. Know that Allo recently had problems and had to work with ESS and XMOS to solve a DSD512 problem. Turned out to be one line of XMOS code, don't know if the fix is in a public release yet or not.
Last edited:
Mark,
Thank you so much for your mentoring. I can't tell you how much I appreciate it. I will make sure to get help from experienced sound engineers and musicians. I won't be the one telling whether the DAC sounds good or measures well. Other people can do that a lot better than I ever could, and I do not pretend that I could acquire the necessary knowledge in a short time. Some things just take time, and that's the way they are, and it's probably best that way. What I am doing is a little bit different: leveraging the awesome power of an open source community to do things that would be impossible otherwise.
Cheers!
It's hard to say what will be stable without testing, but you could try ADA4898.
I like these! They seem to combine a lot of the benefits that I was looking for. One question though: I am wondering if I could use them for the 470μ capacitors that I need for MVDD and MVREF. With ESR=70mOhms, the unit price is $6.26, but with ESR=20mOhms, it goes up to $15.40! Here is the datasheet for the expensive one. Alternatively, would the SP-CAP be an acceptable option? I am trying to save space compared to the conductive polymer hybrid aluminum electrolytic capacitors.
Last edited:
Best to be careful that filter caps you may select are appropriate for voltage regulators used. Very low ESR caps may cause a Jung regulator to oscillate, for example. That could be destructive to the dac chip if sufficient care is not taken. Some zener diodes might mitigate risk, it may even be that AKM will include them on Rev 1 eval boards, don't know.
Thanks a lot for the tip! The selection process for capacitors is a lot more complex than I was expecting. It's a fun puzzle though. I have put a selection matrix together. I will make it a bit more precise in order to take your warning into account.
Capacitors Analysis
Here is an updated version of our selection matrix for capacitors to be used by the DAC. This document currently focuses on capacitors with a capacitance of 10μF or more. For this selection, we are trying to optimize the following parameters:
- Low ESR (but not too low)
- Low leakage (for battery-powered operations)
- High ripple current
- Small area
- Low height
According to this analysis, tantalum polymer capacitors from Vishay seem to provide the best options for all the required capacitors, at the exception of the 100μF 15V, for which a polymer electrolytic capacitor from Kemet is probably better. That being said, the Vishay alternative would work too and would be half as expensive.
The good news is that our selected capacitors are really small: the largest takes 31.4mm² of PCB space and the tallest is 4.3mm high. As a result, we should have plenty of space to put everything we need on the PCB.
Here is an updated version of our selection matrix for capacitors to be used by the DAC. This document currently focuses on capacitors with a capacitance of 10μF or more. For this selection, we are trying to optimize the following parameters:
- Low ESR (but not too low)
- Low leakage (for battery-powered operations)
- High ripple current
- Small area
- Low height
According to this analysis, tantalum polymer capacitors from Vishay seem to provide the best options for all the required capacitors, at the exception of the 100μF 15V, for which a polymer electrolytic capacitor from Kemet is probably better. That being said, the Vishay alternative would work too and would be half as expensive.
The good news is that our selected capacitors are really small: the largest takes 31.4mm² of PCB space and the tallest is 4.3mm high. As a result, we should have plenty of space to put everything we need on the PCB.
Last edited:
The Vishay T59EE477M016C0020 is every bit as good and less than half the price. But we'll need to establish that its ESR of 20mΩ is not too low.
Mounting of LT3045
I am still trying to figure out the best way to mount the LT3045 regulators on the DAC board. If we go with low-profile tantalum capacitors like the ones we've selected so far, we will want to keep the PCB as short as possible, which makes the use of vertically-mounted regulator PCBs with heat sinks rather unattractive, especially if we need to stack two DAC boards on top of each other, either to add more outputs or to provide more room for all the components that need to fit on it.
Bottomline: should we still use the LT3045-78XX but use straight headers for horizontal mount and a lower-profile heat sink? Or is there another option that we should consider?
I am still trying to figure out the best way to mount the LT3045 regulators on the DAC board. If we go with low-profile tantalum capacitors like the ones we've selected so far, we will want to keep the PCB as short as possible, which makes the use of vertically-mounted regulator PCBs with heat sinks rather unattractive, especially if we need to stack two DAC boards on top of each other, either to add more outputs or to provide more room for all the components that need to fit on it.
Bottomline: should we still use the LT3045-78XX but use straight headers for horizontal mount and a lower-profile heat sink? Or is there another option that we should consider?
The LT3045 data sheet shows the recommended board layout in the applications section. You should read the entire applications section carefully to make sure you understand the issues.
The LT3045 has a exposed pad (gnd) which should be soldered directly to the pcb.
I find in all my DAC’s where I mount the regulators in this manner no other heatsinking is necessary.
However you need to look at the in/out voltage difference, current draw etc & calculate
the heat dissipation. As Markw4 said best have a good look at the data sheet as all that info is there. With careful planing & pre regulation you should be able get away without heat sinks.
I find in all my DAC’s where I mount the regulators in this manner no other heatsinking is necessary.
However you need to look at the in/out voltage difference, current draw etc & calculate
the heat dissipation. As Markw4 said best have a good look at the data sheet as all that info is there. With careful planing & pre regulation you should be able get away without heat sinks.
Mark,
There is quite a bit going there indeed... Definitely not something that I would like to improvise about. This further confirms the idea that using a pre-built module like the LT3045-78XX is the way to go, at least for the first version. And the cost of this module is very reasonable compared to the cost of the parts that go into it.
This begs one question though: if I were to mount these modules parallel to the DAC board and away from the DAC chip, would it be okay for it to lay above capacitors and resistors mounted on the DAC board?
That's good to know.
My current approach is to build a PSU board similar to the OSVA AAPSU01 in order to pre-regulate all the power supplies, and use the LT3045-78XX, which already mounts the LT3045 in the way that you are describing. I hope this should be enough (to be verified and tested). I've also asked Alex at LDOVR if he ever considered building an LM7805-compatible module with two or three LT3045 chips in parallel: this would positively remove the need for the heatsink for most applications. Putting two on a 30mm × 15mm board would be perfect for what I need...
I would really suggest to get the eval board first. It is designed so you can try alternate voltage regulators. You can get an LT3045 module and see if it makes any difference to dac performance. if not, maybe you don't need it. Or then again, depending on incoming power quality, it may need some cleaning up. Maybe LT3045 can do part of that. In other words, I think you are jumping the gun here. You should first figure out what matters a lot and what matters very little, then expend effort and board space where it is useful.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.