If you're open to unconventional approaches then I can't recommend a CLC filter between DAC and I/V stage highly enough. Normally the I/V opamp is critical to the subjective SQ but with the filter installed in my own (multibit) DAC good SQ can be had from lower-spec'd I/V opamps. Which makes sense to me technically as with the filter, the I/V stage is no longer slewing at sample transitions.
how about opa861 for I/V
like http://designideas.cocolog-nifty.com/photos/uncategorized/2013/08/28/20130828_00615.jpg
or discrete
Качественный дискретный выхлоп для ЦАП
like http://designideas.cocolog-nifty.com/photos/uncategorized/2013/08/28/20130828_00615.jpg
or discrete
Качественный дискретный выхлоп для ЦАП
OPA1612 as I/V
Hi Mikett,
reading your thread about ESS9028PRO Moving from 9038Q2M to a 9028Pro board I thought that you are charmed from using OPA1612 as I/V http://www.ti.com/lit/an/sboa237/sboa237.pdf
I don't know why they Inject bias voltage and In pcm1794 specification http://www.ti.com/lit/ds/symlink/pcm1794.pdf there Is no biasing of NE5534 used as I/V
did you change that opinion
and how about using OPA1622 or ADA4898 at output as differential summing and low pass filtering
Hi Mikett,
reading your thread about ESS9028PRO Moving from 9038Q2M to a 9028Pro board I thought that you are charmed from using OPA1612 as I/V http://www.ti.com/lit/an/sboa237/sboa237.pdf
I don't know why they Inject bias voltage and In pcm1794 specification http://www.ti.com/lit/ds/symlink/pcm1794.pdf there Is no biasing of NE5534 used as I/V
did you change that opinion
and how about using OPA1622 or ADA4898 at output as differential summing and low pass filtering
Last edited:
OPA1622 is not DIY friendly, and the 8-pin DIP versions of OPA1622 from China, assuming they are not fake, do at a least fake using the ground pin on OPA1622 properly. I kind of assume they use some resistors to make a voltage divider to bias the ground pin half way betwen the + and - power pin voltages. But, that's not really ground, and the power pins may have noise, ripple, etc., on them that a true ground plane ground would not have.
Otherwise, OPA1612 is a very good choice when very low distortion is wanted, such as with high quality 24-bit dacs. If OPA1612 doesn't sound good, then there is a problem somewhere, maybe with power supply impedance not being low enough across the entire audio band up to around 100kHz, and or too much RF/HF leakage from the dac chip getting into the dac analog ouptuts and no corrective action taken in the dad board design.
Otherwise, OPA1612 is a very good choice when very low distortion is wanted, such as with high quality 24-bit dacs. If OPA1612 doesn't sound good, then there is a problem somewhere, maybe with power supply impedance not being low enough across the entire audio band up to around 100kHz, and or too much RF/HF leakage from the dac chip getting into the dac analog ouptuts and no corrective action taken in the dad board design.
Mark,
thank you for reply
good to know that you like OPA1612 as I/V for quality 24 bit dacs
do you recomend to try It with not just only pcm1794 but also pcm63, pcm1704
what Is your choice for power supply to pcm1794, analogue and digital : Salas shunt, LT3042, ADM7150, cap multiplier
what Is your choice for OPA used as diferential summator / low pass buffer - what do you think about OPA2132 or ADA4898
thank you for reply
good to know that you like OPA1612 as I/V for quality 24 bit dacs
do you recomend to try It with not just only pcm1794 but also pcm63, pcm1704
what Is your choice for power supply to pcm1794, analogue and digital : Salas shunt, LT3042, ADM7150, cap multiplier
what Is your choice for OPA used as diferential summator / low pass buffer - what do you think about OPA2132 or ADA4898
samoloko,
While OPA1612 is good for low distortion 24-bit dacs, it may be important what clock noise coming out of the dac looks like. For Sabre dacs it is very low level, but there is some HF/RF junk. It can be wise to filter out a good amount of any HF/RF junk before an I/V amplifier. Sometimes a 2,000pf cap can be enough to do the trick, connected from the dac output to ground (I/V input to ground), or maybe connected differentially for differential output dacs. Whichever way works best for a particular dac is best, so it may require trying both ways, and maybe trying a bit bigger caps. For older dacs with even bigger clock pulses, more pre-I/V stage filtering may be appropriate and further help sound quality. Also, for 16-bit dac chips, there may not be a lot of benefit from using an ultra-low distortion opamp. A faster opamp may be more tolerant when using dac chips with more substantial, higher amplitude clock or other HF/RF noise.
While OPA1612 is good for low distortion 24-bit dacs, it may be important what clock noise coming out of the dac looks like. For Sabre dacs it is very low level, but there is some HF/RF junk. It can be wise to filter out a good amount of any HF/RF junk before an I/V amplifier. Sometimes a 2,000pf cap can be enough to do the trick, connected from the dac output to ground (I/V input to ground), or maybe connected differentially for differential output dacs. Whichever way works best for a particular dac is best, so it may require trying both ways, and maybe trying a bit bigger caps. For older dacs with even bigger clock pulses, more pre-I/V stage filtering may be appropriate and further help sound quality. Also, for 16-bit dac chips, there may not be a lot of benefit from using an ultra-low distortion opamp. A faster opamp may be more tolerant when using dac chips with more substantial, higher amplitude clock or other HF/RF noise.
Last edited:
samoloko,
I am using OPA1612 for IV and differential summing/filtering. The +-15v opamp power needs to have low impedance across the audio band up to 100kHz or so, IMHO. That can be done passively with enough fairly large and expensive film caps, or it could be done with an active voltage regulator. Active regulators work by having a lot of gain in a feedback loop, so depending on how much phase shift is caused by components in the load, such regulators may risk instability and oscillation if sufficient care is not taken. They also need to sense the load voltage where the voltage is used or very close to that, not long wires away on another power supply board. Our fellow diyaudio member eziitis has reported what he finds to be good results using the three transistor version of Nazar regulators, which are pretty cheap and easy to build. eziitis also says he hasn't had any stability problems if he uses only X7R ceramic bypass caps at opamp power pins, and skips any tantalum or electrolytics on the load side of the regulator. Probably, a set of dedicated + and - Nazar regulators located on right next to a 3-opamp output stage would be appropriate and likely to work well. However, I must say for the record that I have not directly compared a set of the Nazar regulators eziitis likes with what amounts to about $100 worth of Wima mks4 film caps. Of course, there are other regulator options too, such as perhaps classic Jung regulators (diyadudio store sells bare PCBs to make a set of them). I will attach a schematic below of +- Nazar regulators recommended by eziitis and add notes offered by him as follows:
"The shunt transistor type depends on how much power it should deal with, so current set by R15/39 and output voltage. their value can be calculated R=(1.25-Ueb)/(0.3/Uout)+Iload
Ueb typically 0.65V, 0.3 stands for the max wattage of the transistor in this case. so if one wants to pump more current then something like BD139/140 should be used. LMs need some 4V overhead, and the whole thing radiates a little heat. anyway, the noise is in uV range, but the impedance is what rules. elcaps better not very low ESR, the reg may start to oscillate. ceramics perhaps better np0, but i am fine with x7r so far."
Regarding my own use of passive film caps, more info at: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-380.html#post5690994 ... and a few posts following that one. My most successful experiment with film caps was using Wima mks4, with 2 identical sets of five film caps soldered in parallel (2 independent sets of 5-soldered-together-parallel-caps), and with one of the sets connected from each of the +-15v rails to ground. Values used for each of the sets of parallel caps are: 10+10+22+33+33 ( all values in uf). As I said, it costs about $50 per rail, so not cheap.
I am using OPA1612 for IV and differential summing/filtering. The +-15v opamp power needs to have low impedance across the audio band up to 100kHz or so, IMHO. That can be done passively with enough fairly large and expensive film caps, or it could be done with an active voltage regulator. Active regulators work by having a lot of gain in a feedback loop, so depending on how much phase shift is caused by components in the load, such regulators may risk instability and oscillation if sufficient care is not taken. They also need to sense the load voltage where the voltage is used or very close to that, not long wires away on another power supply board. Our fellow diyaudio member eziitis has reported what he finds to be good results using the three transistor version of Nazar regulators, which are pretty cheap and easy to build. eziitis also says he hasn't had any stability problems if he uses only X7R ceramic bypass caps at opamp power pins, and skips any tantalum or electrolytics on the load side of the regulator. Probably, a set of dedicated + and - Nazar regulators located on right next to a 3-opamp output stage would be appropriate and likely to work well. However, I must say for the record that I have not directly compared a set of the Nazar regulators eziitis likes with what amounts to about $100 worth of Wima mks4 film caps. Of course, there are other regulator options too, such as perhaps classic Jung regulators (diyadudio store sells bare PCBs to make a set of them). I will attach a schematic below of +- Nazar regulators recommended by eziitis and add notes offered by him as follows:
"The shunt transistor type depends on how much power it should deal with, so current set by R15/39 and output voltage. their value can be calculated R=(1.25-Ueb)/(0.3/Uout)+Iload
Ueb typically 0.65V, 0.3 stands for the max wattage of the transistor in this case. so if one wants to pump more current then something like BD139/140 should be used. LMs need some 4V overhead, and the whole thing radiates a little heat. anyway, the noise is in uV range, but the impedance is what rules. elcaps better not very low ESR, the reg may start to oscillate. ceramics perhaps better np0, but i am fine with x7r so far."
Regarding my own use of passive film caps, more info at: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-380.html#post5690994 ... and a few posts following that one. My most successful experiment with film caps was using Wima mks4, with 2 identical sets of five film caps soldered in parallel (2 independent sets of 5-soldered-together-parallel-caps), and with one of the sets connected from each of the +-15v rails to ground. Values used for each of the sets of parallel caps are: 10+10+22+33+33 ( all values in uf). As I said, it costs about $50 per rail, so not cheap.
Attachments
Last edited:
I've bought the same PCB. I'm looking how to use it...
Have you read this? http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sboa237&fileType=pdf
Have you read this? http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sboa237&fileType=pdf
i have similar problems with ada 4898-1 in i/v...it works for a minute...then it melts, but i used asymmetrical rails(+9v-12v)...
What i liked very much was this here :
http://lampizator.eu/lampizator/fetishizator/JFET_CDout.pdf
A rather discrete approach, that i liked much more than the hyper neutral new age opamps.
http://lampizator.eu/lampizator/fetishizator/JFET_CDout.pdf
A rather discrete approach, that i liked much more than the hyper neutral new age opamps.
Experience shows not to use LT3045/94 for opamps. However, okay to use LT3045 for the digital/clock areas of dacs.
Never came across a published article, if that's what you mean. Mostly people trying the LDOs, then trying something else and noticing a non-subtle improvement in sound quality. It's why shunt regulators are still popular for high end audio. Twisted Pear Placid HD, Jung-Didden, Salas, Nazar, etc. The other thing that often works well for opamps are lower gain series regulators with some big film caps near the load. Some people like to use DC Link film caps because some are pretty low cost. Some discussing over a couple of pages at: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-423.html#post5753075 ...for example. Suggest you read around and try some different options yourself.
EDIT: Even for digital and clock loads in a dac (or AVCC in Sabre dacs) ADM7150 is usually preferred over LT3045 by those who have tried both.
Last edited:
Did anybody try lm7805(78L05) or the lt1086 with a negative reference to get lower voltages?I found the 7805 series being very good for anything digital or audio no matter the speed.Lt1086 (83,84) was widely used in computer motherboards...I think the secret with regulators is to use separate resistive loading of at least a quarter...half the value of the needed regulated current.Ideally , if you need 100ma you load it separately with another 100ma.That is empirical practice, but i've seen it here and there...
Too much noise for LM317/337. I'd suggest any LED referenced pre-regulator, even the simplest (LED + pass transistor) is adequate.
Could you elaborate further on your subjective impressions of the sound character of that circuit? Does it sound like a triode based circuit to you? Thanks.
Very cool board!
Here is a little tech note I wrote about the output stage of the PCM1794A. This mostly focuses on the gain balancing from the first and second stage. Another thing to consider is that you do not necessarily need symmetrical supplies for the first stage. The inputs stay at the virtual ground, so that stage does not really need a high voltage positive supply.
http://www.ti.com/lit/an/sbaa333/sbaa333.pdf
Here is a little tech note I wrote about the output stage of the PCM1794A. This mostly focuses on the gain balancing from the first and second stage. Another thing to consider is that you do not necessarily need symmetrical supplies for the first stage. The inputs stay at the virtual ground, so that stage does not really need a high voltage positive supply.
http://www.ti.com/lit/an/sbaa333/sbaa333.pdf
Hi, Paul,
Can you can answer the following question which frequently arises for those of us who model discrete I/V circuits for use with the PCM1794A? What is the value of the device's analog output impedance? Presumably, it's relatively high since it's based on switched active current-sources, and probably code dependent, but is there an nominal value we can use with our I/V circuit simulations? Thanks.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.