DIY discrete dac chip.

Status
Not open for further replies.
Jump to Latest
#81
Re: Re: Re: unanswered questions

Kuei Yang Wang said:
Konnichiwa,



No, it uses (used) UA, actually, custom made with passive IV off board.

Sayonara
Click to expand...

Yep, you are right Kuei. I got it mixed up with Levinson
30.5 / 30.6. 30.5 had UA module, 30.6 uses PCM1704/OPA627.

All these discrete dacs (lavry + UA) use a sample and hold
OP clocked by the XO. Probably this is because they could
not build a sign magnitude architecture and infringe Burr Browns
patent. As such they need the S+H output to have clean state
changes.

Terry
 
#82
Using Signed magnitude for DIY

Can the signed magnitude approach be used for DIY designs ?

What about sharing a schematic for a DIY dac using signed magnitude ?

If signed magnitude can't be used, an approach were the MSB bits are split up into smaller "steps" can be used. Or is something like this patented ?
 
#84
New idea's minimizing jitter

To minimize jitter, and get fast settling time the DAC should consist of two DAC's whith the wanted resolution(16 bit).
A ultra fast switch should switch the active DAC for each sample period.

The settling time for each DAC will then not be critical.

The switch which I think is the most critical part in the DAC can now be highly optimized because only one extremly fast switch instead of 16 is needed.

The disadvantage is that the two DAC's must give same absolute current output.

This is very simular to S/H in a conventional DAC.

What do you think ? is there other ways to make fast(and relative cheap) switch in a Weighted Current source DAC ?
 
#85
Lgrau said:
I was wondering if anyone have made a discrete konverter chip themselves.

Should be a farly simple task. The circuit will give a lot of opertunities for optimizing the perfomance.

Sorry for my poor english but my english spelling check isn't installed yet.

/Lennart
Click to expand...

I found a Burr Brown DAC80 chip yesterday. It has a 12 bit D/A, all separately accessible on bit-level. Adding just 4 bits with current sources (FETS) would really make a rather discrete 16 bit DAC. Excellent settling time.
Any ideas someone?
regards albert

http://www.analog.com/en/prod/0,,774_1082_ADDAC80,00.html
 

Attachments

  • dac80 sheet.jpg
    dac80 sheet.jpg
    97.6 KB · Views: 915
#89
Bernhard said:
Yes, the summing.
Click to expand...

Now an odd suggestion:
as pin 20 is plain out (in the logical diagram, I admit, not the physical) thus would provide Real Current for external I/V IF and only IF the internal OpAmp would not provide the current dump (in its I/V work).
Now, to do that, this would imply shorting it with a 1 Hz filter. If you understand what I mean.

albert
 
#93
percy said:
Hi, I am glad to have found this thread while searching for information on building a completely discrete r2r diy dac!

So whats the progress so far ? I hope encouraging!
Click to expand...


Pondering.
yes, good links from ' rfbrw '!

Lets look again at the 1543. Good,dynamic. But like a donut:
'Keep your eye on the (w)hole and not on the donut' - it's if I quote correctly, an american proverb.
In our case: Focus on the whole and not on the hole in the middle.

So, having said that, what would the reason be for a hole in the middle? Some L-R interference on the powerline? Maybe due to the multiplexing of the same output device on the two output channels?

So a discrete dac like the burr brown would allow a real solid line to be set. 12 bits, and either go to 14 bits with 4x oversampling (could sound as good as NOS 16 bits in my opinion).
2 bits, on the bottom (the LSB is now 400 so much micro amperes. So it can easily be extended there.

And yes, if there is some nice converter to 16 channels, then just pumping these devices at 44khz to 88 khz or even 186khz would not hurt them a bit. --> Discrete 16-bit DAC (CS8412 input receiver, mode 5 output format) for start with... does anyone have a board that outputs this and can be fed directly inot the Burr-Brown (the 12 singular bit entries)?

2) Stil thinking how this -vVersion of the DAC-80 can be converted into -iMode! Actually just floating the output on 10k or even 100k would just swing it to either rail. This should give the current out in all freedom.:angel:

--> if that were true we also would have a solution to get the 1-bit chips free of the output summing opamp that is in the chip!!
 
#97
Any news ?

I have collected 16 DAC chips with parallel data input, Iout and access to the Vref.

Connecting all DACs - each using another bit only - to a single Vref using tight adjustment pots Sfernice ceramic multiturn could work.

Accuphase had a discrete DAC in their first transport / DAC combos: 16 bit 2x os ( I have borrowed from a friend to make measurements ) and 20bit 8x os.
They used ICM8019 integrated current source / switch 4 bit per chip.
Plus pots for the 4 MSBs.
The whole thing is very temperature sensitive...
 
#99
oshifis

How about using a chip that inputs serial and outputs parallel each bit separately. I found this chip: NJU3716M, 16 bit output producer. (an 18bit version also exists.)
This would provide an output directly, allow adding an output stage of any configuration you like.
The chips works up to 5 Mhz input rate, fast enough I think.

regards
albert
 

Attachments

  • nju3716m klein.jpg
    nju3716m klein.jpg
    50.3 KB · Views: 610
Status
Not open for further replies.
Top Bottom