R-2R DAC Kit

Jump to Latest
#3
Nowadays 'R2R' in relation to DACs has come to mean 'multibit' - if you didn't mean a literal R-2R DAC (with logic chips as switches and ultra-precise resistors) then I have one :

PhiDAC hex kits with pre-built filters

The DAC chips I'm using in this design are true multibit but instead of using resistors, use capacitors. The advantage is that the chip is immune to ageing because it constantly refreshes the charge on its capacitors to maintain accurate binary weighting.
 
#6
@SiliconTi, it is not easy to create your own R-2R DAC, you can buy one ready-made, or alternatively look for a chip based which is verified and works well ... often it won't be like a kit, but you buy your own parts, you don't need a lot of them there ... to create your own R-2R from resistors means to get resistors that are accurate enough, and not only that, you also need low resistance switches ... the sound will not be as good as you would achieve with a high quality chip
 
#7
You can find complete DIY R2R modules, a pair for $115 there:
???响

Not a complete kit, just fully assembled DAC modules with self calibrating CPLD control logic and the I2S or RJ protocol inputs. The rest is on your own to built a DAC.

I wouldn't do that, as on the same website you get complete DAC of very good sound quality for $350.
 
#8
altough it claims 24-bit, the reality is only about 14-bit?, for a logic reason: the resistors accuracy ... for 10-bit dac the accuracy 0.1% is expected, for 14-bit it is 0.006% ... as it gets deeper, the switches resistance accuracy does matter more 😀 .... for 24-bit R-2R dac the resistance accuracy of resistors and switches should be 0.000006% ... which is possible in chip on some advanced technology, but impossible as a solid state 😀 ...
soekris offers at least sign-magnitude which reduces glitch
 
#11
miro1360 said:
altough it claims 24-bit, the reality is only about 14-bit?, for a logic reason: the resistors accuracy ... for 10-bit dac the accuracy 0.1% is expected, for 14-bit it is 0.006% ... as it gets deeper, the switches resistance accuracy does matter more 😀 .... for 24-bit R-2R dac the resistance accuracy of resistors and switches should be 0.000006% ... which is possible in chip on some advanced technology, but impossible as a solid state 😀 ...
soekris offers at least sign-magnitude which reduces glitch
Click to expand...
Are you saying that DA-8 modules are not sign magnitude? Who told you this? Secondly, do you know how resistors accuracy is resolved in TDA1541/3, TDA1387?
 
#12
Am I saying that? Idk DA-8, I just mentioned that soekris has it, if DA-8 has than good for it.
TDA1541/3 and 1387 are multibit, but are not R2R ... 1541/3 is based on active elements, so the 1387 is continuous calibration ... therefore their operation internally differs from a pure R2R architecture, it uses a mixture of switched current sources for the lower bits and multiemitter transistor current dividers for the upper bits. I said something about difficulties in R2R accuracy of solid states.
 
Last edited:
#14
miro1360 said:
TDA1541/3 and 1387 are multibit, but are not R2R ... 1541/3 is based on active elements, so the 1387 is continuous calibration ... therefore their operation internally differs from a pure R2R architecture, it uses a mixture of switched current sources for the lower bits and multiemitter transistor current dividers for the upper bits.
Click to expand...

+1

All the TDA15XX and TDA1387 are current-source based rather than resistor based - one consequence is they have higher output impedances than R2R types being true 'current out' designs. R2R DACs can be designed to be 'current out' the trouble for them is then they have effectively zero output compliance. AD5547 is one such example.
 
#19
miro1360 said:
Am I saying that? Idk DA-8, I just mentioned that soekris has it, if DA-8 has than good for it.
TDA1541/3 and 1387 are multibit, but are not R2R ... 1541/3 is based on active elements, so the 1387 is continuous calibration ... therefore their operation internally differs from a pure R2R architecture, it uses a mixture of switched current sources for the lower bits and multiemitter transistor current dividers for the upper bits. I said something about difficulties in R2R accuracy of solid states.
Click to expand...
Your comment was about required resistor network accuracy and you didn't include a quoting text, but it is placed immediately after my message. Now your explanation makes it clear, but you can't blame the messenger.

My point was about the fact that calibration reduce requirements for resistor accuracy. It is why it is relevant to your (previously quoted) message. Whether it is a pure R2R (resistor) or I2I network (current source) network, it doesnt matter. my point is still valid, so I don't understand a reason of your arguing that TDAxxx is not R2R architectuire. This is also a ladder architecture and works on the same principle, so calibration makes sense.

Once the above is clear. It is advertised that Audio GD DA-8 modules do self calibrating. I am interested how it could be done.
 
#20
TDAs discussed are not i2i, but rather binary weighted and it is not the same principle 😀
In DA-8 another technique is used for the correction (based on description), idk details
I did not find any measurements or evidence that it could handle those 24 bits successfully.
 
You must log in or register to reply here.
Top Bottom