lingDAC - cost effective RBCD multibit DAC design

[tjstyle]

I'm Interested to for the 10-off kit!

Here is the BOM.

I'm wondering about the bulk kits idea - I'd like to encourage more ultra-cheap DACs out into the world and Taobao sourcing is the only way I know of to get this DAC to below $3 in components. As PCBs tend to come in quantity 10 minimum to me it kind of makes sense to kit up for a build of 10 DACs. Then shipping costs won't swamp the parts cost. What I don't know is how much labour's involved in putting together kits so at this stage I can only make a rough guess. Would there be any takers for a 10-off kit priced at $50 (including PCBs)? I'd guess this might be attractive for people who'd like to build multi-channel DACs for digital XO'd active speakers. For those who don't need more than a single stereo DAC, they could potentially sell their 9 remaining built-up PhiDACs to those who prefer not to get involved with soldering, more than recouping the cost of their own DAC.

 

[matt_garman]

I'd also be willing to jump on the 10-lot train.

 

[miro1360]

Where to get working TDA1387? The one, which is not recycled? Recycled SMD components from the electronic garbage are my worst ebay experience so far, only DIPs components are working because they are not pulled out by overrated hotair (unfortunately TDA1387 is only SMD).
I have strip of 40pcs from ebay and no wonder that they do not work (even though bias from iout is 0.5mA, Vref is 1/6 from Vdd, but there is no reaction on I2S signal - standard I2S (I have not tried LJ - EIAJ, because datasheet says standard I2S)).

 

[miro1360]

I found a few working pieces ... time to try something ...

 

[miro1360]

I tried this combination without any filters:
xmos i2s --> TDA1387 --> basic opamp I/V (lets say best were OPA1688 or ADA4898) --> audioCap --> TPA6120 for headphones (powered from batteries)

... well, the result from this simplicity is not the way (in terms of dynamics and detail) ...

Who knows what would happen if one took abraxalito's filter+I/V and put it forward to the AD1862 or PCM1704 ?)

 

[abraxalito]

... well, the result from this simplicity is not the way (in terms of dynamics and detail) ...

I'm curious - what does 'not the way' mean? I'd really like to know how that combination you tried sounded.

Incidentally why not use the TPA6120 as the I/V stage, seems simpler to me, omitting an opamp? I'm getting very good results from using AD815 as I/V and according to TI, TPA6120 is an upgrade on AD815.

 

[miro1360]

Not the way means ...., this simple connection lacks details through the frequency range (compared with similiar simplicity of AD1862 or PCM1704). The most noticeable is the loss of dynamic in low frequency and fewer noticeable, but still be, few details in higher frequency. Perhaps paralleling TDA1387 is a good idea to improve dynamic in low frequencies (but for price of increasing capacitance on the output, adding errors, ... overall it can bring some improvement).

Well, the TPA6120 directly as the I/V is in oscillation. It needs some tuning to get off the oscillation ... I did not get this oscillation completely off (the chip is than slower).

Do you have an internal concept/schematic of AD815?

PS: when you connect your I/V to the AD1862, it can be interesting

 

[abraxalito]

Did you delete the feedback capacitor on the I/V when you tried the TPA6120 in that position?

I have no internal AD815 schematic but judging by the DS for the AD8017 its most likely the 'Nelson' architecture : Current-feedback operational amplifier - Wikipedia

What are you powering the TDA1387 from? I don't have any AD1862s to try here.

 

[miro1360]

Yes, I removed the feedback cap (this slightly decreased the oscillation). I tried a solid plane GND on the PCB and the same result was present. The oscillation is in few of MHz and heats up the chip. The simulation in ltspice is ok, but the reality is not.

 

[abraxalito]

I found AD8017s oscillated with a few cm of wire between the +ve input pin and my Vref. A 220pF cap between +ve in and -V (on TPA6120 this would be between pin4 and pin1) fixed the oscillation.

 

[miro1360]

Thank you for this hint, I will try it during the next measurement and I will post you my result

 

[abraxalito]

Updated PhiDAC schematic v1

The PCB re-layout is now done and hopefully will go off to manufacture tomorrow. Here is the revised schematic for PhiDAC v1, I've shaved a few components off the BOM through deleting unnecessary repetitions of circuit elements related to the Vref. A few decoupling caps I omitted on v0 I found were necessary on the first prototypes, so they're now included.

The PCB had an error due to the schematic not being correct, the grounding has been improved through testing the DAC direct into headphones - this helped identify common-impedance coupling on the ground tracking. Headphones of 50ohm or above can be driven by PhiDAC but they work better if the 15ohm series output resistor is replaced with a couple of FBs and the 470uFs upgraded to 2200uF.

 

[PJN]

I'd also be willing to jump on the 10-lot train.

Hi Matt,

If you get that 10 kit I'd be interested in taking a couple off of your hands.

Paul

 

[TNT]

I'd like to get hold of one also ;-)

//

 

[hollowman]

Digital filter option

The PCB re-layout is now done and hopefully will go off to manufacture tomorrow.

Oh no!!!!!!!!!!
Dear Toots:
Youz fergots 'bout da pads fer an OPTIONAL digital filter.
Oh .... so close ... but no cigar.....

 

[Markw4]

Still think it might be a good idea to get one here and add it into the listening tests we do with other dacs. Unlikely to come out on top (I'm guessing), but if it is good enough to compete with dacs costing a few hundred dollars as Abraxalito has described, at least we could compare its value to the much more costly Sabre dac implementations. And I would be happy to recommend it to people looking for a low cost solution if I find it sounds good to me.

 

[hollowman]

Free samples ....

Still think it might be a good idea to get one here and add it into the listening tests we do with other dacs. Unlikely to come out on top (I'm guessing), but if it is good enough to compete with dacs costing a few hundred dollars as Abraxalito has described, at least we could compare its value to the much more costly Sabre dac implementations. And I would be happy to recommend it to people looking for a low cost solution if I find it sounds good to me.

With all the content we've added to this thread, I think Braxy owes us some FREE samples. Pre-assembled units, of course. I ain't payin' for no stinkin' solder!!
PM me Braxy ...

 

[Markw4]

Nothing owed to me. Haven't had anything to contribute. Abraxalito seems to know what he's doing. By the way, Abraxalito, don't know if you saw it or not, but you got a brief mention in Blowtorch III.

 

[abraxalito]

Youz fergots 'bout da pads fer an OPTIONAL digital filter.

So I did, right. And you forgot to answer my question.

Oh .... so close ... but no cigar.....

Couldn't call your forgetting 'close' though, after all you had reminders

 

[abraxalito]

By the way, Abraxalito, don't know if you saw it or not, but you got a brief mention in Blowtorch III.

Yes, I had a heads-up from another contributor to that thread. I saw George mentioned me and you said something about 'clock noise' which I wasn't sure about. For clarity, the filter's not for getting rid of 'clock noise' primarily though it of course does help with that, its for eliminating the bulk of the images.

Here's what I saw on my SA from a 'raw' TDA1387 output. The clock noise dominates for sure at higher frequencies but plenty of the lower frequency images (below ~400kHz) are higher level than the clock.