This old chip is interesting with its simplicity, you only influence on output RC filter and power supply (which is critical), any i/v stages and active filters, where each component affects on overall sound character. And it sounds really good with proper implementation of power supply and a good i2s source.
Hey anatolii_A,
When you do this, does the i2s then have to be 1,8v or is 3,3v still ok?
Have you added a op amp output stage or only using the simple rc filter shown in the datasheet?
Hi! Can't say about i2s (it seems in my case there is 1.8v), but high livel at DEMP/FLT/FLT pins must be 1.8V in case of using 1.8V DVDD. It is hard to broke this chip, in any unnormal conditions it goes in mute and you will see ground at analog outputs.
I haven't added anything at output, but the components of RC filter matters a lot. For minimum it must be good thin film resistors. Different type of capacitors matters too, btw we can slightly increase value of cap.
Also I have updated some parts, 1.8V regulator now TPS7A2018. To keep the concept of simplicity I decided to use single polarity power supply, for negative source I use TPS60403, and then TPS7A4701/TPS7A3301 after it for positive/negative 3.3V DAC rails. It is good enough for analog output supply.
just edit post, it may be not ok to have 3.3V at i2s, because I haven't checked it
Last edited:
Coincidence! Just have ordered these for experiments in similar direction (balanced though) but this time El Cheapo way with 4 x AMS1117-X.X as I have these in abundance for experimenting and they are sturdy. 1 x 5V, 1 x 3.3V and 2 x 1.8V...3.3V. Also one of the latter possibly to be used as -3.3V. Single board of course.
http://www.advanced-monolithic.com/pdf/ds1117.pdf
http://www.advanced-monolithic.com/pdf/ds1117.pdf
Attachments
Last edited:
Okay, could I use 3.3v on the DVDD pin and then 1.8V on the LDOO pin or would it then be better to just use the internal LDO as specified in the datasheet?
I am working on a board to replace the cheap china modules I currently use in my DIY DSP: https://www.diyaudio.com/community/...-based-on-adau1466.426241/page-2#post-7985793
For now I am just aiming for a decent sounding DAC. Not the highest tier of HiFi. This is mostly a proof of concept of all the connectivity and basic functionality. When it works I will take my shot at a more fancy AKM4493S based DAC using LT3045 and LT3094 for the analog supplies.
I also cannot decide wether or not my implementation justifies Wiima Film caps or if COG's will sound the same. For resistors I use MELF thin film resistors.
I may be able to post my schemattic and layout later if you wouldn't mind shooting it down 😀
Then my schematic looks something like this. Power supplies:
DAC:
Does that look decent @Anatolii_A ?
DAC:
Does that look decent @Anatolii_A ?
Checked i2s, there is 3.3V levels at oscilloscope and it works with 1.8V DVDD. Looks very decent, what is main power source? Relay is unnecessary, PCM5102A mutes output to GND by itself. What will be i2s source? For filter capacitors you can make many different holes/layout for tht or smd. If you make your own board, think about layout DGND/AGND. I'm not sure internal PLL will work with disabled charge pump and think it anyway better to have MCLK from good quality oscillators (powered by separate regulator).
Really? well its not within specifications at least.. but thanks for testing it!
The main power source, for now is directly from USB-C PD with some filtering, or seperate 5V wall wart. later it will be a more fancy supply, but not there yet..
The relay is unnecesarry yes, but I will need it for the AKM design, so just want to try it out and see how I can control it and so on 🙂
The I2S source is my own DSP: https://www.diyaudio.com/community/...-based-on-adau1466.426241/page-2#post-7985793
That board is currently V0.1 and I already have made some significant changes to include a better clock and buffer based clock distribution to be able to power all boards (DSP, ADC and DAC's) from the same 24.576MHz chrystek clock. Buffers will be one LMK1C1104 followed by three additional LMK1C1104's to achive my required 9 outputs (4 i2S lines in 4 i2s lines out and the DSP itself).
Yeah my idea was also to just make holes for the film caps to fit so I could try it. Do you think it will make sense? or are the film caps overkill?
Oh, good point on the PLL, didn't considder that.. I have at least 2 mclk outputs on my current DSP board so I should be able to power two DACs at least then..
This is what the layout currently looks like. Obviously not done yet, but you can get an impression of what I'm trying to do. its 4 layers. top layer is 99% of it. two internal layers are GND. Bottom layer is mostly GND, some power routing and the relay. I am not trying to split GND, but just physically keep analog stuf on the analog side of the DAC and everything else on the digital side. Does it make sense?
Closeup of the interesting part
Looks solid for me, I haven't learned how to model boards yet, now trying to learn it with KiCad to make own board version of dual PCM1794 NOS DAC.
Leave more space between PCM and output connectors for bigger caps variants. Also I think that i2s lines should be as short as possible (and what about damping resistors). With 4.7nF caps in RC we move phase at 10kHz approx for 8 degrees, but I can't say it is audible.
Also try tantalum nitride thin film resistors in RC filter. Sounds pretty good.
Leave more space between PCM and output connectors for bigger caps variants. Also I think that i2s lines should be as short as possible (and what about damping resistors). With 4.7nF caps in RC we move phase at 10kHz approx for 8 degrees, but I can't say it is audible.
Also try tantalum nitride thin film resistors in RC filter. Sounds pretty good.
Last edited:
How can you know these things if you never made a PCB :O
I use KiCad. I'm a mechanical engineer so its been quite a learning curve the last 6 months.. the DSP was my first ever PCB.. a miracle it works xD
But now, I'm starting to get the hang of KiCad, and I really like it... except for stitching vias.. it can't do that it seems 🙁
Yeah, will make room for the film caps, will be fun to be able to try it.
I do not belive the i2S lines need damping resistors at the recieving end? There are some right at the DSP exit on the DSP board.
I also don't worry to much about the i2S line length.. Honestly, i2s seems pretty robust. if the spaghetti monster I have now works, without any GND refference at all (just LRCK, BCLK and DATA, not a single GND wire next to them - only the power wire has a GND next to it) then the new design with a 1mm pitch ribbon cable with gnd between each signal as well as this 4 layer board with GND below the signals must be like livin the dream for i2S xD
Tantalum nitride, are they better than MELF? I thought they were the S***?
Again, bare with me.. I'm no EE by far xD
Cheers, Daniel
I use KiCad. I'm a mechanical engineer so its been quite a learning curve the last 6 months.. the DSP was my first ever PCB.. a miracle it works xD
But now, I'm starting to get the hang of KiCad, and I really like it... except for stitching vias.. it can't do that it seems 🙁
Yeah, will make room for the film caps, will be fun to be able to try it.
I do not belive the i2S lines need damping resistors at the recieving end? There are some right at the DSP exit on the DSP board.
I also don't worry to much about the i2S line length.. Honestly, i2s seems pretty robust. if the spaghetti monster I have now works, without any GND refference at all (just LRCK, BCLK and DATA, not a single GND wire next to them - only the power wire has a GND next to it) then the new design with a 1mm pitch ribbon cable with gnd between each signal as well as this 4 layer board with GND below the signals must be like livin the dream for i2S xD
Tantalum nitride, are they better than MELF? I thought they were the S***?
Again, bare with me.. I'm no EE by far xD
Cheers, Daniel
Why so categorricaly right away ? I saw people were using them as I/V stage with good results so decided to try it in simple analog circuit. It so easy to swap some few components you have an interest to try it. My first PCBs were painted with varnish and acid etched, later were some iron-printer way made. All another stuff was chineese DIY-level boards or modifying already finished things. Anyway now I can solder BGA chips but can't design PCBs, it happens 🙂
Last edited:
It is not hard, just start. Makes way more sense than soldering many existing ChiFi devices as with the gained knowledge you can make a nice design. Exactly as you envision it.
Depends. If an I2S signal and its dedicated ground wire forms a transmission line of nonuniform characteristic impedance, then some distributed damping may be helpful.
The ones I tried added some distortion and or correlated noise. Some people seem to like the effect for coloration of output stages. I prefer my output stages to be more accurate instead. Then if it doesn't sound right, try to see if there is some way to fix it without adding a coloration effect (which may be more just to help mask a problem rather than to fix it properly).
My personal opinion only, of course.
Last edited:
Can you please elaborate on this. I'm still very new to all electronics, so not sure what a real dual rail supply is vs. What I have here.
And what can you hear? A pop or click? Or what do you mean? 🙂
Will definitely make room for some filter caps!
Sometimes its simply a veil over the sound. In other cases it can be more obvious. That's why you really need to A/B the difference between good analog and common SMPS power; its to start learning what such artifacts can sound like. Its not that SMPS can't be made to be as good, but its not necessarily easy to do well from an engineering perspective, and some of the more random-like effects can be hard to measure.
Last edited:
Well as I said I don't expect this to be perfect. But this specific board is a proof of concept in a project that's going to evolve over the next years. Gotta start somewhere xD
Next step after this board will be deep diving into power supplies. I would very much like to use SMPS for the size and efficiency. I know I need a better PSU before starting on op amp designs as required for an akm DAC.
All the more reason to start at the beginning, rather than at a more advanced level.
Like so many things in life, there are things that motivate taking shortcuts (such as skipping math classes), while hoping to jump right into some of the more interesting areas of theoretical physics.