The Rload of 2k4 is a bit low for this buffer, also I'd suggest having it AC coupled.
The ferrite beads I've measured normally come in around 1uH, 21nH seems too low (not that it'll change the sim result).
I note that you've got 4k7 in the base circuit whereas I put 4R7 - meaning 4.7ohms.
The ferrite beads I've measured normally come in around 1uH, 21nH seems too low (not that it'll change the sim result).
I note that you've got 4k7 in the base circuit whereas I put 4R7 - meaning 4.7ohms.
I made a mistake on the 4K7, corrected now. Also changed ferrite beads to 1uH.
I followed pavouk's schematic when setting Rload to 2K4.
What value would you suggest for R1-R4?
Sim with 1uH and 4R7:
AC coupled? Please explain this. I'll try to google it in the meantime.
Edit: so it's simply blocking DC with the right value capacitor. On the input of the buffer before L2 or on the output before Rload? Somewhere else in the circuit?
Last edited:
On AC coupled I meant to put a capacitor in series with the Rload as the buffer's not designed to be DC-accurate. At present the Rload is shunting away current from the CCS (Q1 and Q3).
Given a clean sheet to design the filter stage I'd go for 100k resistors for R1-R4.
Given a clean sheet to design the filter stage I'd go for 100k resistors for R1-R4.
Ok, without too much changes to the filter stage, what values can I use for Rload? I've ordered most parts for the WM8740 board already so I'd rather not change too much. MKP's etc aren't cheap.
If you want to keep the 2k4s in circuit you'll need to increase the bias current of the CCS so that it stays in classA the whole time. It'll lead to poorer sound I reckon due to the increased loading on the DAC chip and opamps.
I'm confused by your question because I earlier on suggested a change from 2k4 to 24k. Are you now saying that much of a change is too inconvenient? I don't recommend using MKPs, rather NP0 SMT ceramics which are relatively affordable and the smaller they are the cheaper they get.
I'm confused by your question because I earlier on suggested a change from 2k4 to 24k. Are you now saying that much of a change is too inconvenient? I don't recommend using MKPs, rather NP0 SMT ceramics which are relatively affordable and the smaller they are the cheaper they get.
No, I'd rather not change too much as I've got alot of the parts on the way already. I've ordered film capacitors for this project.
Like the 3n3 FKP2, 100nF MKP418, 680pF FKP2 etc etc.
I guess I could use them for some other project maybe in the future. I ordered these parts before you suggested the BC817-40 buffer.
So if I where to change R1-R4 to 24K, what else would need changing?
I'm sorry if my questions are a bit stupid, this is the first dac I build that I don't change parts in a kit or remove parts to add a different solution(example Edcor transformers out from my CS4398 instead of opamps).
If given a fresh start at the filter, opamp part...what values would you use including the buffer etc
Schematic again:
I just went over to Mouser to check prices of NP0 ceramics. Both 330pF and 68pF in 0805 cost $4 for a strip of 100. Not exactly wallet busting.
None of your questions so far have seemed stupid to me. As explained earlier, with R1 - R4 increased by a factor of 10 (10X bigger) the associated capacitors become 10X smaller. The other resistors in the filter also increase by 10X (so 1k5 for what was 150ohms). No need to change the decoupling capacitors - that's the ones connected to the opamp power rails.
From a clean sheet I'd go R1 = 100k, R2 = 100k, C29 = 82pF, R5 = 100k, R6 = 100k, R9 = 6k2, R10 = 6k2, C31 = 15pF, C33 = 15pF. Right channel similarly.
Connect an output buffer to J1-3 with bias about 4mA so R3 = 150ohm, R2 = 68k for your schematic shown in post #22. AC couple its output with a capacitor 220uF/16V and 220k to GND on the output side. Positive side of capacitor towards the buffer.
Connect an output buffer to J1-3 with bias about 4mA so R3 = 150ohm, R2 = 68k for your schematic shown in post #22. AC couple its output with a capacitor 220uF/16V and 220k to GND on the output side. Positive side of capacitor towards the buffer.
Something like this? (only filter and output buffer)
I'm very new to Spice, so I'm learning as I go...
I'm very new to Spice, so I'm learning as I go...
That looks the ticket yep - except you have three .op statements (presumably from cutting and pasting).
You can get a plot of the differential output by choosing Rload1 output as your 'reference' (right click on the net and go to the bottom of the menu 'mark reference') then plot the signal on Rload2.
You can get a plot of the differential output by choosing Rload1 output as your 'reference' (right click on the net and go to the bottom of the menu 'mark reference') then plot the signal on Rload2.
Thanks alot 🙂
I'll look into that later or tomorrow as I have to get our doberman male to the physiotherapist soon.
I'm not sure why that would happen, but its possible to work around that. Place a VCVS (component 'e' in the components interface) and wire its two inputs (+ and -) between your two outputs, then connect its negative output to AGND. Probe its positive output for the difference signal.
<edit> I forgot to mention - you need to set its 'value' (default shown as 'E') to 1. Right click and edit the 'E' to '1' (the second 'E' that is, on the line 'value' not the first instance).
<edit> I forgot to mention - you need to set its 'value' (default shown as 'E') to 1. Right click and edit the 'E' to '1' (the second 'E' that is, on the line 'value' not the first instance).
Last edited:
Like this? Tried without the GND's for Rload1 & Rload2 but got very different values after C11 & C12 when I did that.
Considering adding a SRC to my modular dac project as I have an AD1896 laying around.
Would that be on any benefit? From what I've read a oscillator of 22.5792Mhz and 256Fs, I'd get all formats upsampled to 88.2Khz?
I'd place the SRC on it's own board as that would give my the option to use it/not use it as I see fit.
Would that be on any benefit? From what I've read a oscillator of 22.5792Mhz and 256Fs, I'd get all formats upsampled to 88.2Khz?
I'd place the SRC on it's own board as that would give my the option to use it/not use it as I see fit.
I don't have much experience of ASRCs so can't really comment either way.
You look to have connected your 'E1' to GND both sides (inputs) - surely you want it to go to the other side of those load resistors?
You look to have connected your 'E1' to GND both sides (inputs) - surely you want it to go to the other side of those load resistors?
Thanks, I'll try that as soon as my pc's intake filters are dry and I can boot it up again.
Did some maintenance on my watercooling as I planned on doing some work on the pre-regs I'm thinking of putting in front of the shunt regs anyway.
Figured I'd drop voltage from around 11-12Vdc at the capacitor bank to about 8Vdc for the 5V shunts and to around 5V for the 3.3V shunts.
No - now you have C11, C12 shorting your output to AGND and Rloads in series with the inputs to the VCVS.
You want the inputs of the VCVS (E1) to go to the negative terminals of the caps C11,C12. But you don't want those terminals connected to AGND.
You want the inputs of the VCVS (E1) to go to the negative terminals of the caps C11,C12. But you don't want those terminals connected to AGND.
- Status
- Not open for further replies.