I'm coming into this thread late, but it appears that you're using Eagle. If so, make sure you are using the ERC (electrical rule check) and DRC (design rule check) to check your schematic and layout for obvious errors.
I also wanted to suggest that you try out www.FreeDFM.com, who will test your PCB design for manufacturing and send a report on any errors found. This goes beyond the DRC in Eagle. 4PCB will also give you a quote, and might even phone you, but you are not obligated to buy your PCB from them just because you use their FreeDFM service.
I also wanted to suggest that you try out www.FreeDFM.com, who will test your PCB design for manufacturing and send a report on any errors found. This goes beyond the DRC in Eagle. 4PCB will also give you a quote, and might even phone you, but you are not obligated to buy your PCB from them just because you use their FreeDFM service.
Thanks for that piece of info. I was already using ERC and DRC for debugging, although some rules in the DRC are not helping me very much. For example it was showing me that I don't have enough clearance between the pads of a ssop28 footprint...
I will try freeDFM, sounds like an interesting options. I wasn't aware of it.
The project took an interesting turn in the meantime. I have an almost complete DAC made on a breadboard with crude voltage regulators and grounding. I will use it for making sure I understand how the DAC works and what control pins I need.
After this the board will be reviewed and I will include some other features:
1. I did some research and it appears that the VREF input needs a very high quality source. I plan to feed it through a buffer around a low noise op-amp, a la Buffalo dac.
2. The same holds for VCOM pin. This one needs DVDD/2 so I need another voltage reg. and I plan to also feed it through an op-amp buffer.
3. I plan to add some slots for experimenting with other sources and regulators.
All in all this should be quite a reference board once finished!
I will try freeDFM, sounds like an interesting options. I wasn't aware of it.
The project took an interesting turn in the meantime. I have an almost complete DAC made on a breadboard with crude voltage regulators and grounding. I will use it for making sure I understand how the DAC works and what control pins I need.
After this the board will be reviewed and I will include some other features:
1. I did some research and it appears that the VREF input needs a very high quality source. I plan to feed it through a buffer around a low noise op-amp, a la Buffalo dac.
2. The same holds for VCOM pin. This one needs DVDD/2 so I need another voltage reg. and I plan to also feed it through an op-amp buffer.
3. I plan to add some slots for experimenting with other sources and regulators.
All in all this should be quite a reference board once finished!
That is not uncommon when using the default DRC. Keep in mind that DRC cannot be a universal test, like ERC. You must create a set of Design Rules that match your PCB fab house parameters. Some places, like Sunstone Circuits, will actually provide an Eagle .dru file that you can use.
Basically, when the SSOP28 footprint fails, it means that your Design Rules specify spacing that is actually larger. It's actually quite complex to translate PCB maker specs to .dru, but you can get boards to pass usually. Otherwise, you might have to modify the footprint to match the capabilities of your fab, or modify your .dru to truly reflect the capabilities of your fab.
I have several .dru files in my database, one for each product pricing level available from the PCB houses that I work with. Then, when working on a given circuit board, I can load the corresponding .dru according to the project budget, and run DRC to make sure that the board design will work. There are even reasons to change the .dru on the same board. For example, you might use a cheap PCB service for prototyping, and then change to a higher quality for production once you get the volume discount.
Thanks for the additional tips rsdio. Unfourtunately the fabs around don't seem too much into eagle, I'll ask for .dru files though. I've also used the one available from sparkfun.
5th element, yes, you are right, I mixed things up. I guess the best decoupling tehnique remains a good cap. Is there any "active" solution for decoupling?
By the way today I've started the ak4396 with exaU2I. It sounded great although the implementation is crude comparing to the one I am planning here. Still, it was a good idea to do this as the PCB will suffer plenty modification. By the way, not all of the control pins are having internal pull-up as I thought. Check below the schematic I arrived for starting this DAC. Maybe will be useful for others too.
I'll come back with the revised board also.
5th element, yes, you are right, I mixed things up. I guess the best decoupling tehnique remains a good cap. Is there any "active" solution for decoupling?
By the way today I've started the ak4396 with exaU2I. It sounded great although the implementation is crude comparing to the one I am planning here. Still, it was a good idea to do this as the PCB will suffer plenty modification. By the way, not all of the control pins are having internal pull-up as I thought. Check below the schematic I arrived for starting this DAC. Maybe will be useful for others too.
I'll come back with the revised board also.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
It's always worth it to ask, but I must admit that few fabs will provide the .dru files for you. You have to learn how to create them yourself.
It might be worth comparing the default Eagle DRU with the one from Sparkfun, just to learn how they are different.
I have also noticed that certain assembly shops want a CAD file, which is easily produced within Eagle using their GenCad.ulp file (from the CadSoft website).
Basically, the more you work with a certain fab and/or assembly house, the more you will learn about how to create the files they need. Sometimes you'll get lucky and there will be specific Eagle versions of the files available, but mostly you have to create them yourself.
I would use an op-amp buffer to drive Vcom. It's a 1/2 VA divider (from internal 750-975 ohm resistors) and there would be sonic benefits. Otherwise, at least leave a pad or something to try this.
The AK4396 defines VCOM as an Output only. Nowhere in the data sheet does it allow driving VCOM from an external source. Recommendation is to attach parallel capacitors for filtering, and there is also mention that a Time Constant is set by the RC network formed by the internal resistance and the external capacitance. It seems highly likely to me that driving this pin from an external op-amp, which is a very low impedance source, would defeat the intended function of that pin and might even cause damage.
From what source did you decide that using an op-amp buffer would be beneficial, much less even allowed in the first place?
There are DAC designs where an internal reference voltage is available, but it can be disabled in favor of an external reference. Sometimes in these cases there is a dual-purpose pin, which serves as a connection for a bypass capacitor when the internal reference is active, but it serves as an input when an external reference is used. However, in such cases the pin is clearly marked as I/O. Also, there is typically a second pin to specifically disable the internal reference, and thus switch the dual-purpose pin from output to input.
In other words, I don't see any documentation or even any possibility that the AK4396 allows VCOM to be used this way. I would recommend following the data sheet from Asahi Kasei.
In other words, I don't see any documentation or even any possibility that the AK4396 allows VCOM to be used this way. I would recommend following the data sheet from Asahi Kasei.
Thanks rsdio and 5th element, this seems the right way to handle this pin too. I'll stick with opamp buffer only on the vref.
By the way, I've been doing further listening with the DAC and the exa device. I really think I'm on something here. I need to replace those noisy VR's, and then I would have an even better image of what this dac is capable of..
By the way, here's a comparison of high quality VR's. Unfortunately the really nice ones are hard to solder.
However the old LM723 puts out some nice figures don't you believe?
By the way, I've been doing further listening with the DAC and the exa device. I really think I'm on something here. I need to replace those noisy VR's, and then I would have an even better image of what this dac is capable of..
By the way, here's a comparison of high quality VR's. Unfortunately the really nice ones are hard to solder.
However the old LM723 puts out some nice figures don't you believe?
Choosing regulators has always been one of my pet hates. Loads of parameters to juggle around with, varying degrees of circuit complexity and sometimes widely variable prices. Not to mention a plethora of different manufactures and product availability.
I've used the TPS7A4501 to good effect, as well as the well known LT1763. I think everyone's built a few regulators around the LM117/317 part too. Now what 'to good effect' means I don't really know. The regulators work and the devices they power work flawlessly too. I've mainly stuck to super regulators for any serious application involving analogue audio, the TPS and LT are powering a DSP and another digital circuit. I know clean power is supposed to be vital for low jitter applications, but to what degree changing from one regulator to another will have on the sound/jitter I do not really know, especially I would assume, if one has provided CLC filtering after the regulator anyway.
I use two TPSs for the DSP and the DSP works brilliantly, it doesn't provide any clocks mind you, it's slaved to external ones. I do however believe that the actual implementation of the DSP or say a DAC chip is probably far more important towards an end products performance then what power supply you feed it. Or said another way, a lousy implementation could well probably mask any small improvements that a better PSU could provide.
If you're experimenting, I'd say go with something simple, something easy to solder and something that doesn't break the bank.
I've used the TPS7A4501 to good effect, as well as the well known LT1763. I think everyone's built a few regulators around the LM117/317 part too. Now what 'to good effect' means I don't really know. The regulators work and the devices they power work flawlessly too. I've mainly stuck to super regulators for any serious application involving analogue audio, the TPS and LT are powering a DSP and another digital circuit. I know clean power is supposed to be vital for low jitter applications, but to what degree changing from one regulator to another will have on the sound/jitter I do not really know, especially I would assume, if one has provided CLC filtering after the regulator anyway.
I use two TPSs for the DSP and the DSP works brilliantly, it doesn't provide any clocks mind you, it's slaved to external ones. I do however believe that the actual implementation of the DSP or say a DAC chip is probably far more important towards an end products performance then what power supply you feed it. Or said another way, a lousy implementation could well probably mask any small improvements that a better PSU could provide.
If you're experimenting, I'd say go with something simple, something easy to solder and something that doesn't break the bank.
Thanks, I've been buzzing members of this fourm with my questions for about 6 years now with nothing to share back and I am glad this thread actually generates some valuable info.
And a special notice: I might just played 352Khz content on the AK4396 with the exa U2I device. It's still possible that I am fooling myself, but if interested head over to the exaU2I thread on this sub-forum also for discussions. If this DAC really plays at higher sampling rates then wow...
I found SunRa also posted the thread here... DIYHiFi.org; View topic - Layout question for AK4396... Please share what you find.
In general, DAC IC's have an internal 1/2*AVDD node. This is used to bias their output stage and for some IC's it's a common reference. There are also switched-capacitor filters and the integrators' analog power that can be brought to outside pins. All of these (pins) need to have the lowest impedance possible to keep noise to a minimum. Capacitors do this but there are limits to their size and the DAC IC's current sources. An op-amp can provide push-pull drive and since feedback is involved, it can provide a very a low impedance for the DAC pin, assuming frequencies are in-band.
I haven't worked with this AK part and suggested at trying it. You won't damage the IC; it's an "output" pin for the DRC check.
Why is it okay to do this to REF pin?
I haven't worked with this AK part and suggested at trying it. You won't damage the IC; it's an "output" pin for the DRC check.
Why is it okay to do this to REF pin?
Okay, it seems to me that an active circuit could be used on VCOM to null in-band noise or signals appearing there, essentially, acting like an active decoupling capacitor. I believe that the datasheet says VCOM is a high impedance output. However, an actual capacitor will not interfere with the D.C. level of VCOM while providing an A.C. ground. An op-amp on VCOM may need to be configured as an A.C. only regulator, and not regulate D.C. Perhaps, no damage or performance problems would otherwise result. One way to find out, but at the risk of the DAC chip. In the end, a sizable, but not ridiculously so, audio grade electrolytic cap. may be preferable for it's simplicity and lack of risk.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.