Looking to compile a list of every R2R ladder DAC available. I will add what I know of and hopefully everyone will chip in.
Denafrips: Ares, Pontus, Venus, Terminator
Metrum: Menuet
Schiit: Gumby, Yggdrasil
Audio Mirror: Tubadour
Holo: Spring KTE
Denafrips: Ares, Pontus, Venus, Terminator
Metrum: Menuet
Schiit: Gumby, Yggdrasil
Audio Mirror: Tubadour
Holo: Spring KTE
Perhaps this list will be more useful if the chip and discrete solutions are kept separate, as there are only a handful of discretes and hundreds of chip based dacs.
Exactly = PCM 1704 🙄 , while still love my WADIA 27 spline based interpolation sound (tube alike) than this pre/post ringing AKM/ESS and so one.
Also the soldered R/2/R stuff will get humid and poisoned over time with a guarantee of oxidation and leakage. In other words a thin laser trimmed film based substrate is hard to beet.
Hp
Found another maker of several ladder DAC models, some with tube-buffers, at MHDT Lab in Taiwan.
Website here: mhdtlab.com
Review of the "Stockholm 2" model here:
Mhdt Lab Stockholm V2 USB DAC
Website here: mhdtlab.com
Review of the "Stockholm 2" model here:
Mhdt Lab Stockholm V2 USB DAC
According to the TDA1541 datasheet, page 5: "True 16-bit performance is achieved by each channel using three 2-bit active dividers operating on the dynamic element matching principle, in combination with a 10-bit passive current-divider, based on emitter scaling."
TDA1541 Datasheet pdf - Dual 16-bit DAC - Philips
I'm not sure why they call a circuit based on emitter scaling a passive current divider, but it sounds to me like the TDA1541 is based on scaled current sources rather than R2R networks. It might use an R-2R network to divide the current between the sources for the lower ten bits, though, depending on how they implemented that "10-bit passive current-divider, based on emitter scaling".
The PCM1704 indeed has R2R networks according to its datasheet (page 6, left column, pcm1704.pdf ): "Two DACs are combined in a complementary arrangement to produce an extremely linear output. The two DACs share a common reference, and a common R-2R ladder for bit current sources. The R-2R ladder utilizes dual balanced current segments to ensure ideal tracking under all conditions."
TDA1541 Datasheet pdf - Dual 16-bit DAC - Philips
I'm not sure why they call a circuit based on emitter scaling a passive current divider, but it sounds to me like the TDA1541 is based on scaled current sources rather than R2R networks. It might use an R-2R network to divide the current between the sources for the lower ten bits, though, depending on how they implemented that "10-bit passive current-divider, based on emitter scaling".
The PCM1704 indeed has R2R networks according to its datasheet (page 6, left column, pcm1704.pdf ): "Two DACs are combined in a complementary arrangement to produce an extremely linear output. The two DACs share a common reference, and a common R-2R ladder for bit current sources. The R-2R ladder utilizes dual balanced current segments to ensure ideal tracking under all conditions."
Last edited:
I just read Rudy van der Plassche's 1976 article about dynamic element matching. Apparently R-2R was a very common technique for splitting the currents in current source DACs, so chances are that Boydk and Zoran are correct and that the TDA1541 uses an R-2R network for its lower ten bits.
Not exactly understand why take the discrete R2R implemention route. To achieve even 16 bit ENOB, you need a 0.0001% resistor. Expensive but possible with integrated, laser trimmed thin film but good luck with the discrete resistors. Almost all discrete R2R DAC uses 0.01% or worse resistor that theoretically can only achieve 13 bit ENOB even if there is no noise (only non-linearity) at all.
Whats important for Audio is not absolute linearity, but good level linearity, so you don't need that precise resistors for a good discrete R-2R DAC, just use the Sign Magnitude principle (same as PCM63 and later parts)....
