The transparency of the output coupling caps is very important for discerning differences in the circuit preceding it. Substandard caps are more opaque. No output caps are preferable, if it can be done.
Téflon...not for I according my ears. As for tantalum R at I/V... very dépendant of the dac and whole i/V stage. Rhopoint worked for I as carbon comp (if low R, noise is not critical but sounds good).
where the cheap chineese and else dac fail imo. they are made to measure not to sound good first, some certainly do but when we see the outputs for instance, most of the time we saw mêmes op amps a, smd passive parts below 1000USD. There is a website which is made for them called ASR where people only scope to rank them. Btw if I think twice of the econoomic model of such site, seems to me it is feed by such brands....
Now some surely sound good and some chineese brands are good too. You will not know at looking at the measurements only !
Now some surely sound good and some chineese brands are good too. You will not know at looking at the measurements only !
Is it possible to use an adjustable JFET as a current sink at the output of the DAC, so that the quiescent current output of the DAC can be zeroed? Then passive I/V and no output cap since there would be no DC voltage at the output?
The problem with large body film caps is that they give an impression of transparency, but they tend to lose low level details of instrument sounds, room reverberation decays, depth cues, etc. The older smaller body, lower voltage film caps were less problematic but the availability of thin, low voltage film material more or less ended. That's why we see 200v, 400v, and higher caps; smaller 100v caps aren't being made so much anymore.
Using physically smaller, lower capacitance film caps can help a lot, but then a FET buffer may be necessary to maintain full FR. Once there is a buffer, then there is its sound to contend with.
Last edited:
With opamps that type of thing can be done, or as AKM explained in the original AK4499EQ datasheet its also possible to reduce the fraction of Vref used on the I/V opamp non-inverting input. That allows the opamp output DC offset to be adjusted down as needed. By that means the DC offset can be set to ground potential; it can even be servoed to ground if desired. Differential summing can accomplish DC removal as well (since the DC offset can be viewed as very LF common mode noise). Just depends if you like the sound of opamps used that way.
Last edited:
not to say the high pass needs to be low enough which avoid best caps when having two digits DC blocking caps...
It is like to use a too slow filtering for a MCLK to filter it... the noise is isolated, but sound is so-so.... and you have recloooooooooocking instead of reclocking. 🙂
It is like to use a too slow filtering for a MCLK to filter it... the noise is isolated, but sound is so-so.... and you have recloooooooooocking instead of reclocking. 🙂
Thanks for the info. Trying to avoid op amps. Maybe the Pedja Audial Aya II I/V design with the OPA861.
Last edited:
Works good to me, i just had pucked à different buffer.
It is not exactly in the buffer the way he is using it, btw
It is not exactly in the buffer the way he is using it, btw
Okay. That type of sorta-grounded base input circuitry is a rather non-linear and poor virtual ground load for a dac with a low output impedance such as an ESS dac. For current output dacs with high output impedance its probably a much better fit.
Its pretty clear that some dac chips such as those made by ESS are designed work with opamp I/V output stages. Period. Its the only way to get a stable, low-distortion virtual ground for a low impedance source. IMHO, it means if someone doesn't like the sound of I/V opamps, the real solution is to look for a different type of dac which can work better with different types of output stages.
Last edited:
In the text, he says the JFET offsets the quiescent current of the 1541A. " As a result, OPA861 emitter current is modulated by +/-2mA so it is not exposed to excess offset, and hence it works in the most linear range." Important because there is no feedback.
Okay, but it takes a high output impedance dac to tolerate the nonlinearity of the load impedance with small voltage changes. A base-emitter junction in a grounded base circuit is a lot like a diode in terms of having an exponentially varying load impedance. For very small changes in current we can make an assumption that the load impedance seen by the dac is piecewise-linear. However, that assumption is much less valid when the dac outputs a high-ish current (with large current variations) and at the same time the actual output current that can come out of the dac is very sensitive to any small change in the load impedance (as would be the case with a low output impedance ESS dac). It means the output current is going to be rather distorted by the nonlinear load. But that distorted current is the audio signal we are going to be listening to after it is converted to a voltage in the next part of the output stage.
Last edited:
@Markw4 : Maybe it is linear enough as it is NOS anyway in the original circuit from Pedja Rogic ? he uses Rf 1k5 and Cf 1 nF (so with a true ground) iirc. and so no upsampling or any reconstruction filter ot sync filter. Mine blowed a R2R by Soekris not by a small margin everywhere and mostly in the bass and snap aera. (with mods)