lingDAC - cost effective RBCD multibit DAC design

v1 BOM

@JMF - thanks for the update.

Here's the v1 BOM with Taobao-sourced parts.

The Mouser BOM's underway, its looking like the parts cost (excluding DAC chip and PCB) will be around $44 for 1 off. The most expensive parts are the AD8017ARZs which go for over $5 each on Mouser. On Aliexpress though they're considerably cheaper ($1 or so) - given you've got to use an Asian source for the TDA1387s it might pay to get the ADs from there too.
 

Attachments

  • phiDACrevG.zip
    10.6 KB · Views: 99
The most expensive parts are the AD8017ARZs which go for over $5 each on Mouser. On Aliexpress though they're considerably cheaper ($1 or so)

Analog Devices offers price for 1000+ pcs about $2.25. Is it possible that a seller from aliexpress can get such a significant good price? Be careful with opamps from aliexpress and ebay, there are the cheapest opamps remarked and sold under different codes.
 
These aren't the normal kind of opamps, they're CFB so not as generic. If ADI are offering $2.25 on their website that's a 'Western' price - quite likely an 'Asian' price is going to be cheaper. Though more likely that Aliexpress/Taobao sellers are offering obsoleted/deprecated stock, in particular non-Z suffix parts (where Z indicates lead-free).

Have a look here for prices on non-Z : AD8017AR-REEL Integrated Circuits ANALOG DEVICES INC | AD8017ARREEL | Quest Components
 
Last edited:
A USB input can be accommodated by the use of a USB->I2S PCB like this one, which uses a TI PCM2706 IC to handle the translation (Taobao link) : PCB2706 USB?I2S?? ????Amanero ??????-???

The compromise with USB is the lack of isolation - given that the typical PC is a noisy brute (in terms of electrical interference) it makes a lot of sense to have an isolation barrier between a PC and your audio system. But isolators for full speed USB are available quite cheaply (Taobao again) : usb???/,usb to usb/?? ????? ADUM4160/ADuM3160-???

On the other hand, S/PDIF can have isolation included by means of a small signal transformer, then the only other worry is jitter. Such a board could be one like this (Taobao) : ????DIR9001?? ??/ ???? SPDIF?I2S??24bit 96khz-??? and the isolation transformer would be one of these (Taobao yet again) : ???PE-65612NL PE65612NL ????????? ????-???
 
That's something for me to bear in mind for a future design - ideally a single DAC board would be configurable to stereo/SE or mono/balanced.

I would love to have something like you mentioned. Just a single board for balanced output, and maybe a matching input connector for the ubiquitous Xmos/Amanero clone USB-I2S module on Taobao? I like the idea of ingegrated modules, rather than flying wires all over the build.

Just a Gerber file would be ok, I order boards weekly for my work.
 
Example PhiDAC application - USB input DAC

@thuanth43 - I'm toying with the idea of putting an MCU on the next revision of the PhiDAC, at that point we can use software to configure the board as single mono/balanced or stereo/SE. That's some way away though, perhaps late spring time as I need to dust off my firmware skills :)

In the meantime, here's an example of how to use PhiDAC as the foundation of a USB powered DAC. There are a few parts to be added - the USB input, a PSU and transformers for the output.

I can't find the link to the USB board I'm using at present, its perhaps hidden because its still holiday period. When it turns up I'll post it.

The PSU is an LM2662 configured as voltage doubler (rather than the normal inverter) and an LC filter afterwards to reduce the switching noise. Its fed direct from USB bus power - I'm using a spare lingDAC power PCB.

The transformers are PQ2016 cores (here : PQ2016???????6+8???????????PC40?????-???) wound with 0.08mm for the primary (1100T) and 0.09mm for the secondary (1100T). Transformers aren't strictly necessary but they do help to improve the SQ when the DAC's fed from such an electrically noisy source as a PC. If the PC you use is a laptop that's always powered from battery they're probably not needed.
 

Attachments

  • 2109442289.jpg
    2109442289.jpg
    280 KB · Views: 554
Progress update

Still working on completing the Mouser BOM. The extra time taken to do this is down to the fact that one of the critical components (the multilayer chip inductors) don't have the same characteristics when sourced from Taobao compared to when sourced from Mouser. How do I know this? I just received an order of similar inductors from Mouser and electrically they're different.

TDK's datasheet says that the typical ESR of its 27uH inductors is 0.5ohm but from measurements I determined that those from Taobao (more than one source) read 0.8ohm typically. Meaning the Q is significantly lower. Not beyond the datasheet maximum though, which is 1ohm.

The inductors (33uH) which recently arrived from Mouser show a typical ESR about 0.6ohm (better Q than Taobao sourced ones) however their measured inductances are all over the place, well beyond the 10% tolerance for the part (MLF2012C330KT000). This makes me wonder whether Mouser sent me the 20% parts (which are cheaper) in place of those I actually ordered (and paid for).

I chose 33uH over 27uH for the Mouser inductors as there's a chance to save a few cents - since they're all in series in the filter we need fewer Ls if we use a higher value. When I tried sourcing 33uH on Taobao I found worse Q than the 27uHs.

The upshot of all this is - I'm building a sample PhiDAC with the Mouser parts to see if the performance is close enough to the DACs I've characterized with Taobao parts. The BOM's going to be slightly different between the two to accommodate the different characteristics of the inductors.

Kits of the Taobao parts are coming together, likely be wrapped up by the end of this week. Those who indicated an interest do please let me know your shipment details by PM in the next couple of days and I'll respond with payment info. For those who've not yet indicated an interest there are a couple of kit sets spare - first come, first served.
 
I've not done a side-by-side comparison with a WM8740 but I've modded one or two based on that chip and found, even after modding they couldn't compete with earlier incarnations of my TDA1387 designs. In particular the HF was 'mushy' (a common characteristic I've found of S-D DACs) and the bass lacked dynamics. Modding them helped, most obviously with the bass but I couldn't find a mod to fix the HF.

As for ES9038 I've not heard a board using that chip yet but I do hope to get a comparison from a local DIYA member who is in possession of both a 'Khadas' (ES9038 based, $99) and a v0 edition of PhiDAC. Stay tuned and he might be persuaded to post some subjective comments up when he's had a listen.
 
I've not done a side-by-side comparison with a WM8740 but I've modded one or two based on that chip and found, even after modding they couldn't compete with earlier incarnations of my TDA1387 designs. In particular the HF was 'mushy' (a common characteristic I've found of S-D DACs) and the bass lacked dynamics. Modding them helped, most obviously with the bass but I couldn't find a mod to fix the HF.

WM8740 is voltage out, to start with, so comparisons here are impossible because all the I/V and filters is done inside the chip, no way to change that, AFAIK. I have a CD player with that chip I'd probably/eventually like to mod or use as an external DAC, so if what I said is wrong please let me know, I might be interested in ways to mod the WM8740. TY
 
What makes comparisons impossible when the DAC chip's got a different mode of output? Even though ICs have differing architectures, all DACs end up putting out of the box voltage (2VRMS is the standard).

Yes yes, that's true. But what I mean is that if you compare Iout DACs you can have the same I/V, filters, etc to compare "the DACs". Obviously you can use different I/V and filters in each and compare the end sound, which is actually what we're interested in. Sorry for the confusion.
 
Not all Iout DACs have the same max output current or the same output impedance or output compliance range. Also they vary by whether they give out a bipolar current or are unipolar. So there are quite a few reasons why direct comparisons between Iout chips aren't likely to be possible.