tda1387 dac pcb "front end"

[matt_garman]

Based on this tda1387x8 thread, I developed a bit of an affinity for the sound of the old tda1387 DAC chip. The topic of the linked thread was a standalone DAC that had eight of those chips paralleled; it also had optical or coaxial SPDIF input (and optional USB input), single-ended output, and a basic op-amp IV circuit. Some fairly simple mods (detailed in the wiki) made it go from just OK to quite engaging and enjoyable (at least to my ears).

Taking all the lessons learned from modding that DAC, and some of my personal goals, I wanted to create a basic "front end" DAC with these chips. Attached are images of what I came up with, with the following features/goals in mind:

• I2S input: use an SBC, SPDIF receiver, XMOS, etc.
• User-supplied power supply: be as cheap/simple or fancy as you want.
• Adding capacitance to pin7 of the tda1387 chip was one of the biggest gains in sound quality improvement (particularly bass); this board has room for 2x 8mm pin7 caps per tda1387 chip.
• Solder pads on the bottom of the board for up to four 10mm caps for the tda1387 chip's VDD supply (pin5); this allows tda1387 chips to be "stacked", boosting current output (e.g. this could be made into a headphone amp).
• Uses a logic inverter chip (74HC86) to optionally invert the I2s DATA signal to half the tda1387 chips. This allows for balanced output. Note this is optional, a jumper setting allows the user to select single-ended output mode (i.e. don't invert the data).
• Has jumper pins for user-selectable simple passive I/V, or direct current-output for a custom I/V scheme.

I think this is pretty flexible: you could get up and running quickly with only one tda1387 chip and passive IV; or you could go crazy and max out all the caps, stack multiple tda1387 chips, and implement the fanciest IV scheme you can imagine. Or anything in-between!

Also, the package and pinout for the tda1543 is the same as the tda1387. I see no reason why this board couldn't be used for the tda1543 as well. Edit: this is incorrect, the package sizes for tda1543 and tda1387 are different.

Note: I only just now rendered the gerber files, and looking at the bottom of the board, I see the big pin5 cap footprints interfere with the little pin5 SMD caps. Should be easy enough to fix.

Happy to hear any comments, critiques, suggestions.

[abraxalito]

Hope this isn't too much of a 'fly in the ointment' but the SMT package of TDA1543 is rather different (SO16 wide) from that of TDA1387 (SO8). As a result I reckon it would be very tricky to try to fit them here. The pin orderings and positions are the same though, not that that's too much consolation.

[matt_garman]

Originally Posted by abraxalito

Hope this isn't too much of a 'fly in the ointment' but the SMT package of TDA1543 is rather different (SO16 wide) from that of TDA1387 (SO8). As a result I reckon it would be very tricky to try to fit them here. The pin orderings and positions are the same though, not that that's too much consolation.

D'oh, I just saw that the pinouts were the same, and didn't realize that the package size was different. No matter though, tda1543 compatibility wasn't an explicit design goal---I just (wrongly) assumed it was a useful side-effect.

Although I suppose an enterprising individual could get creative with little jumper wires and the soldering iron and make it work. That person ain't me!

[matt_garman]

In hindsight, perhaps designing this for tda1543 from the start would have been the smarter move. It would be much easier to fit the tda1387 into the tda1543 footprint, rather than the other way around. All think about doing that, as I'm working on learning KiCAD (in place of Eagle), and that would be a good exercise. Available time will be the main factor though.

At any rate, here's the latest and greatest. Passes all automatic checks, and I don't see any obvious errors. Thinking of submitting to the fab house.

[matt_garman]

FYI, here are some useful diyAudio links on the tda1387 chip, for anyone who might be interested:

TDA1387 x8 DAC: let's check its design, mod it -or not-, play music -or not! -

TDA1387 X8 NOS DAC (Wiki)

TDA1387 continuous calibration dac

Improving passive I/V for Pi dac 8 x TDA1387

[matt_garman]

First, a couple more relevant links that might be interesting:

Dirty Dac, 16x TDA1387 dual mono symmetrical.

help me with my dac design?

Second, I've re-drawn this board using KiCAD. The built-in 3D viewer alone is worth the price of admission (KiCAD is free/open source, so the only cost is the learning time).

Anyway, here's a "version 2" of the board. Or maybe "version 1.1" is more appropriate, since it's only a minor tweak on the original design. Only one pin7 cap per tda1387 chip, but with a bigger (10mm) footprint. Also, all components are on the top of the board. And I created three separate ground planes, one each for digital, analog, and power.

[abraxalito]

Nice job Matt - turning this thread into a TDA1387 resource. I am looking forward to building a DAC from one of your PCBs...

[matt_garman]

I received the boards for the first design (haven't submitted "v2" yet). I had these made by Elecrow. They happened to be having a sale when I submitted: $4.90 for 10 boards (up to 10x10), and $4.80 air mail shipping to USA. Under $10 for 10 boards---and they actually sent me 13 boards!

If anyone is interested, I probably only want to build one or two of these at most. So if you cover shipping, I'll give away the board itself. If you're in the Chicago area, stop by and grab a couple!

[hollowman]

Although I've got five or so in my bin, I've used only a single TDA1387 in my experiments (all in Philips CDP that orig. used a 1543).

I know much of the romance with this DAC chip began with the parallel x 8 design orig. offered a few years ago.

How much sonic benefits actually (subjectively) accrue when more than one TDA1387 are stacked?

I'm assuming parallel x 8 is some sort of sweet-spot?

[hollowman]

Originally Posted by abraxalito

I am looking forward to building a DAC from one of your PCBs...

Brax is gonna need a couple thou. Maybe less than 1k if Brax really maxes out on all the 1387 footprints on Matt's boards.