Well, the output of IC1 can drive as much as the buffer, so no need for the buffer. And that IC1 has the same offset as the servo opamp, so the servo doesn't make things better, so no need for the servo. Just use the IC1 output, with maybe a small resistor of 47 ohms to isolate cap loads.
You have lots of space on that PCB 😉
Jan Didden
You have lots of space on that PCB 😉
Jan Didden
Hi Stef and Janneman,
If I were to build similar to this for a project I am working on, with the CS3310 (or I think a PGA3210 is better??) in the circuit, would the final buffer (2nd half of OPA2134) be needed, or I was thinking the Vol. Control to be balanced as well though, so fully balanced from the DAC to the output would require different op amps... any thoughts there?
I think the filter parts MFB circuit would be ok, is the OPA1632 good enough for output circuit?
I will be designing for 8V RMS output from +/-15V rails.
If I were to build similar to this for a project I am working on, with the CS3310 (or I think a PGA3210 is better??) in the circuit, would the final buffer (2nd half of OPA2134) be needed, or I was thinking the Vol. Control to be balanced as well though, so fully balanced from the DAC to the output would require different op amps... any thoughts there?
I think the filter parts MFB circuit would be ok, is the OPA1632 good enough for output circuit?
I will be designing for 8V RMS output from +/-15V rails.
Pulse-R said:
In your case I would use the filter/balanced-to-se opamp after the DAC, then into the CS3310 that can drive 600ohms directly. Note that the CS works on +/-5V so max Vout is about 2.5VRMS, which would be enough for power amps anyway. You can then also supply the opamp from +/-5V.
If you want 8VRMS (for what?) you need an extra gain stage plus separate supplies.
You can also use two CS3310 channels directly balanced from the DAC but then you need coupling caps to get rid of the 2.5VDC on the DAC outputs.
Jan Didden
janneman said:
Actually, you can do much better then that! Take a balanced opamp after the DAC, like OPA1632 or THS413x. You can do a 2nd order filter on it as well, and be rid of the 2.6 V DC (at least it will be 90 dB down 🙂 ). Then you can simply feed the signal into the CS3318 (balanced). and de output of the CS3318 to the XLR terminals.
Just one problem: at full output the DC voltage of the CS is about 5 mV at +22 dB setting, however you can only go to +10dB anyway (9V supply on CS3318) without clipping on the output. You can either get rid of it with coupling caps, a DC servo or another balanced opamp, or you just leave it be...
Anyway, that's how I want to make my output for the DCX. Just about the last stage I'm not sure yet
Pulse-R said:
The PGA2320 is may be more accurate for what you're looking. +-15v analog power permit more output swing. The 0.5dB step is also very great in real life, and the DSO-16 box easy to solder.
I replaced the CS with PGA in my Tag AV32 without problem. I don't noticed any sound difference but specs are better on the paper... One thing may be. On the Tag with the CS, sometime I had small "clic" moving volume. This disappeared after the PGA upgrade.
Stef...
.
stef1777 said:
PGA had better specs then CS 😕
Where did you look. You should go for the CS3318 anyway and leave the 3310! CS has better THD+N, more levels, better dynamic range, is easier to use, less noise, less offset. The only thing that is actually better is in the PGA channel separation and voltage swing (do you need that?).
Edit: Correction: THD+N is the same 😉
And one last thing: the CS3318 (21 euro, 8 channel) is much cheaper than the PGA2320 (20 euro's, 2 channel).
4real said:
I'm too old for 48 LQFP package.
For the dac output. Does something will be fine?
I've some space left on the PCB. Can we add some circuit to be best?
😉
Last revision. Also include the first beta of the output board. All seems to fit on the board. I tried to define the right filter values. Hope this will be fine.
miniANALOG beta9
.
miniANALOG beta9
.
4real said:
Actually, you can do much better then that! Take a balanced opamp after the DAC, like OPA1632 or THS413x. You can do a 2nd order filter on it as well, and be rid of the 2.6 V DC (at least it will be 90 dB down 🙂 ). Then you can simply feed the signal into the CS3318 (balanced). and de output of the CS3318 to the XLR terminals.
Just one problem: at full output the DC voltage of the CS is about 5 mV at +22 dB setting, however you can only go to +10dB anyway (9V supply on CS3318) without clipping on the output. You can either get rid of it with coupling caps, a DC servo or another balanced opamp, or you just leave it be...
Anyway, that's how I want to make my output for the DCX. Just about the last stage I'm not sure yet
Yes, that' s also the basic solution I used in my DCX remote control output board.
Jan Didden
janneman said:
I've some place left on the PCB and some very nice Wima MKS-XL 10uF and 22uF caps in my box.
http://www.wima.com/EN/mks2xl.htm
What do you think to add 4 of them at DAC output? I can also add some caps print in the pcb for various filter configuration setup (marked Cx, C1 and C2 removed in this case).
An externally hosted image should be here but it was not working when we last tested it.
Hi,
the 10uF seems a bit low value for the input impedance that follows. I wonder if a 100uF//10uF would help the bass to pass?
R1 Cx forms a passive Q=0.7 filter. This combined with the MFB with Q=0.7 creates your Bessel (Q=0.5) overall filter.
R9 Cx is a final passive filter, Q=0.7, that if close to the turn over frequencies of any of the previous filters would ruin the Bessel characteristic. If the two passive filters each create Butterworth then this achieves Bessel if the MFB is set to Q=1.
But, the impedances around the two passive filters will change the filter effectiveness and component values will need to be corrected (is this pre-distortion?) to achieve the required roll offs.
Is my assessment correct? I'm not too good at this filter stuff!
If the Tag was very good, then why have you deleted the second MFB filter? Will you need to swap the two inputs to correct the output phase which the second MFB would have done?
the 10uF seems a bit low value for the input impedance that follows. I wonder if a 100uF//10uF would help the bass to pass?
R1 Cx forms a passive Q=0.7 filter. This combined with the MFB with Q=0.7 creates your Bessel (Q=0.5) overall filter.
R9 Cx is a final passive filter, Q=0.7, that if close to the turn over frequencies of any of the previous filters would ruin the Bessel characteristic. If the two passive filters each create Butterworth then this achieves Bessel if the MFB is set to Q=1.
But, the impedances around the two passive filters will change the filter effectiveness and component values will need to be corrected (is this pre-distortion?) to achieve the required roll offs.
Is my assessment correct? I'm not too good at this filter stuff!
If the Tag was very good, then why have you deleted the second MFB filter? Will you need to swap the two inputs to correct the output phase which the second MFB would have done?
I really don't see any advantage for the coupling caps except for dynamic range if you would feed the opamp from a low supply like +/-5V. But with +/-15V, the point is moot.
Also, the filter caps from the 430 ohms to gnd don't attenuate the diffferential noise BETWEEN the two outputs, and that's what counts. C1 is already doing that.
The output filter doesn't do a lot either; R9 needs to stay low value, and any reasonable cap after that works at too high frequencies to be of help.
Jan Didden
Also, the filter caps from the 430 ohms to gnd don't attenuate the diffferential noise BETWEEN the two outputs, and that's what counts. C1 is already doing that.
The output filter doesn't do a lot either; R9 needs to stay low value, and any reasonable cap after that works at too high frequencies to be of help.
Jan Didden
Ok, I restart with a new drawing. 😉
Forgot dac output cap and the R9/Cx cap.
I've have the space to place all the 4 Cx caps marked on this following drawing. But was is appening with phase if we wire all Cx caps?
Or I can go back to the original drawing without Cx caps.
As well, all values need to be recalculated and adapted with E series.
The TAG filter have 2 opamps (no space for me). The input impedance is very high (ok with jfet) and Q very small (CS3310 follow).
Forgot dac output cap and the R9/Cx cap.
I've have the space to place all the 4 Cx caps marked on this following drawing. But was is appening with phase if we wire all Cx caps?
Or I can go back to the original drawing without Cx caps.
As well, all values need to be recalculated and adapted with E series.
The TAG filter have 2 opamps (no space for me). The input impedance is very high (ok with jfet) and Q very small (CS3310 follow).
An externally hosted image should be here but it was not working when we last tested it.
The Tag circuit as posted had two inverting stages resulting in 4pole rolloff and phase correct.
Your single stage inverts the phase and is 3pole rolloff (if the values are corrected).
Your single stage inverts the phase and is 3pole rolloff (if the values are corrected).
AndrewT said:
ok, to summarize:
- if I use Cx -> need to switch OUT+ and OUT- to correct phase
- if I don't use Cx -> can stay with this way.
But Cx can't be optional due to the phase problem!
What is the best? With Cx and phase inverted, without Cx, or with Cx but without C1 and C2 as in the tag schematic?
I'm starting to become crasy with this filter...
.