Earlier we posted a few passive IV circuits, using 27R passive IV resistor, followed by a 100x low-noise amplifier (LNA).
The latter can be based on AD797, AD8429, or discrete JFET, e.g.
https://www.diyaudio.com/community/...st-tht-i2s-input-nos-r-2r.354078/post-7160371
A question was raised :
Can they be used for other DACs, especially those with an offset current, like TDA1541.
The simple answer is yes.
What do you need to change :
1. Choose a suitable value for Riv = 28mV / Iout
2. Add a C-R high-pass filter between Riv and the LNA.
C is connected between Riv and LNA input.
R is connected between LNA input to Gnd.
R should be ~100k.
C can be 100n for AD8429 or discrete JFET, ideally MKP, FKP, Styroflex or PPS.
But C must be 2.2µ or larger for AD797, due to lower Zin.
Patrick
The latter can be based on AD797, AD8429, or discrete JFET, e.g.
https://www.diyaudio.com/community/...st-tht-i2s-input-nos-r-2r.354078/post-7160371
A question was raised :
Can they be used for other DACs, especially those with an offset current, like TDA1541.
The simple answer is yes.
What do you need to change :
1. Choose a suitable value for Riv = 28mV / Iout
2. Add a C-R high-pass filter between Riv and the LNA.
C is connected between Riv and LNA input.
R is connected between LNA input to Gnd.
R should be ~100k.
C can be 100n for AD8429 or discrete JFET, ideally MKP, FKP, Styroflex or PPS.
But C must be 2.2µ or larger for AD797, due to lower Zin.
Patrick
Last edited:
... yes, not ceramic caps if you can... very low inductance here is not always needed with 0402 smd size!
Though I would not go for higher than 0805 for the TDA1540/41 DEM caps 😉 the sooner they see the ground the better !
🙂 ... John from ECEDESIGNS showed that and measured it extensively.
Though I would not go for higher than 0805 for the TDA1540/41 DEM caps 😉 the sooner they see the ground the better !
🙂 ... John from ECEDESIGNS showed that and measured it extensively.
What has an input decoupling cap for a 100x gain stage to do with the DEM cap of TDA1541 ??
Patrick
Patrick
Last edited:
+Io from DAC @ Riv shift voltage phase to 180deg. So the analog output will be inverted. And stay inverted because of OP is i non-invetrted mode.
Put the OP in inverted mode to habe 0deg phase at the output... Cheers 🙂
.
OR second option to invert the data digital word at the input of DAC. THen at Riv comes -Io and with phase shift Voltage will be in right phase...
This coul be of help for balanced configurations.
.
Point is that simple Riv shifts tha phase opposite.
Put the OP in inverted mode to habe 0deg phase at the output... Cheers 🙂
.
OR second option to invert the data digital word at the input of DAC. THen at Riv comes -Io and with phase shift Voltage will be in right phase...
This coul be of help for balanced configurations.
.
Point is that simple Riv shifts tha phase opposite.
Humans are not sensitive to absolute phase.
It depends on the type of OP, in general the inverted OP has no common mode distortion. This is not a problem in many modern OPs, which have even less distortion when non-inverted.
But I really don't understand - if using the OP, then why not use a "regular" I/V stage?
It makes more sense to use resistor I/V with discrete or tubes.
I think yes because of the every single -A stage is shifting phase for 180deg
So after shift of 180@Riv second shift from gain stage (JFET or tube) brings back the phase like should be @ the output of every device - IN phase...
Maybe the more issue comes, that in inverted mode OP again "doing" V-to-I sort of conversion?
But final outcome has to be checked by simulations, real measurements and listening tests?
When the phase is a topic, always comes out this sentence?
Reativizing the phase issue leading to conclusion that it is even better for phase to be inverted?
Never mind... 🙁
Sure.
Who cares?
P.S. If someone cares, he can swap the wires to the speakers, but this will not change anything...
I have no intention of entering into a debate as to what topology makes the perfect IV converters.
Maybe there is none.
There are enough people who prefer passive resistor IV at the Iout node.
This can be followed by some 40dB gain stage (as can be found in MC phono preamps), or even MC step-up transformer.
https://www.audioasylum.com/cgi/t.mpl?f=tweaks&m=4182
https://www.diyaudio.com/community/threads/pcm63-max-compliance-voltage.126770/
My understanding of opamps is really poor.
But as far as I can tell, the virtual ground of the classic opamp IV is not truely zero.
At least it is frequency dependent and inductive, dependent on its open loop characteristics.
https://forum.allaboutcircuits.com/threads/virtual-ground-impedances-connected-to-grounds.140347/
In a BJT based IV converter circuit, such as that from the late Jocko Homo :
https://www.diyaudio.com/community/threads/easy-to-build-i-v-stage.6121/
while the current input node can be quite low at a few ohms, the base current from the current mirror is a source of added distortion.
In the original CEN IV circuit, the gate current is dependent on internal JFET capacitances and is quite a bit lower than the above base current.
https://www.diyaudio.com/community/...lution-of-a-minimalistic-iv-converter.195483/
The input impedance is the effective source impedance of the input JFETs in parallel, and is about 14 ohm.
By choosing a passive Riv of 27R, we are very close to the commonly mentioned 25mV DAC voltage compliance.
And a 100x low noise amplifier (LNA) is not too difficult to make.
Of course, if your DAC has an output current of +/-2mA, you can lower Riv further to 14R.
Whether it has the lowest possible distortion over the entire audio band, or if it sounds good, I leave that for you to judge.
When using an opamp, maybe even a MC-Pre, that you might happen to have around, as the LNA, all you need is a 27R resistor to try out for yourself.
That is all what it is about, having fun.
PS if you use balanced-output DACs, as I do, phase reversal is no issue.
Or as mentioned above, just swap the speaker cables.
Patrick
.
Maybe there is none.
There are enough people who prefer passive resistor IV at the Iout node.
This can be followed by some 40dB gain stage (as can be found in MC phono preamps), or even MC step-up transformer.
https://www.audioasylum.com/cgi/t.mpl?f=tweaks&m=4182
https://www.diyaudio.com/community/threads/pcm63-max-compliance-voltage.126770/
My understanding of opamps is really poor.
But as far as I can tell, the virtual ground of the classic opamp IV is not truely zero.
At least it is frequency dependent and inductive, dependent on its open loop characteristics.
https://forum.allaboutcircuits.com/threads/virtual-ground-impedances-connected-to-grounds.140347/
In a BJT based IV converter circuit, such as that from the late Jocko Homo :
https://www.diyaudio.com/community/threads/easy-to-build-i-v-stage.6121/
while the current input node can be quite low at a few ohms, the base current from the current mirror is a source of added distortion.
In the original CEN IV circuit, the gate current is dependent on internal JFET capacitances and is quite a bit lower than the above base current.
https://www.diyaudio.com/community/...lution-of-a-minimalistic-iv-converter.195483/
The input impedance is the effective source impedance of the input JFETs in parallel, and is about 14 ohm.
By choosing a passive Riv of 27R, we are very close to the commonly mentioned 25mV DAC voltage compliance.
And a 100x low noise amplifier (LNA) is not too difficult to make.
Of course, if your DAC has an output current of +/-2mA, you can lower Riv further to 14R.
Whether it has the lowest possible distortion over the entire audio band, or if it sounds good, I leave that for you to judge.
When using an opamp, maybe even a MC-Pre, that you might happen to have around, as the LNA, all you need is a 27R resistor to try out for yourself.
That is all what it is about, having fun.
PS if you use balanced-output DACs, as I do, phase reversal is no issue.
Or as mentioned above, just swap the speaker cables.
Patrick
.
Attachments
Last edited:
Just a general statement.
Was not referring to anyone in particular.
Certainly not you.
🙂
Patrick
Was not referring to anyone in particular.
Certainly not you.
🙂
Patrick
Long thread, I'm searching to switch two I2S signals 2:1, please could you link to schematic and BOM if exist?
TIA
Felipe
TIA
Felipe
So how does it perform ?
If you use high-ish bias, 18V rails, and output decoupling cap, you will get close to -90dB (in Spice) for "Vrms out.
Not difficult to build, and cap will take care of DC drift for you.
Probably too hot / large for SMD, so through hole I think.
Patrick
.
But then using the same idea, we can perhaps modify the SMD FC CEN IV to a single-ended one ?
The advantage :
1) possible for 12V rails
2) no unobtanium
3) good thermal tracking when using dual JFETs for DC stability
4) no output coupling cap
5) added output buffer (source follower)
Distortion about the same as the Jocko.
Patrick
.
The advantage :
1) possible for 12V rails
2) no unobtanium
3) good thermal tracking when using dual JFETs for DC stability
4) no output coupling cap
5) added output buffer (source follower)
Distortion about the same as the Jocko.
Patrick
.