Hi,
I'm in the planning/gathering stage of building a modular DAC using DIR9001 and WM8740.
Looking at a lot of different schematics trying to pick the best parts from each.
Looking at the filter and buffer stage following the wm8740 in this schematic (pavouk) , how would I best go about swapping the filter/buffer to one using discrete parts?
I would like to know where to "cut" the schematic and if possible get some advice on a good discrete solution to replace the current filter and buffer.
I also wonder if the 75176 is essential or if it can be omitted from the design? I have two of them lying around, but often simpler is better.
I plan on powering the modules with shunt regulators, haven't tried using them yet so I am curious to see how good they are.
Thanks for any advice 🙂
I'm in the planning/gathering stage of building a modular DAC using DIR9001 and WM8740.
Looking at a lot of different schematics trying to pick the best parts from each.
Looking at the filter and buffer stage following the wm8740 in this schematic (pavouk) , how would I best go about swapping the filter/buffer to one using discrete parts?
I would like to know where to "cut" the schematic and if possible get some advice on a good discrete solution to replace the current filter and buffer.
I also wonder if the 75176 is essential or if it can be omitted from the design? I have two of them lying around, but often simpler is better.
I plan on powering the modules with shunt regulators, haven't tried using them yet so I am curious to see how good they are.
Thanks for any advice 🙂
Yes, transformer output would be an option. It's expensive though and I kind of want to go transistors on this project if possible.
I have a CS4398 dac with edcor transformers on the output.
as a Voltage Output DAC the chip already has a Switched Capacitor filter with CMOS op amp(s) built in
I might add C to agnd in the input to the multiple feedback filter shown, but even as shown the op amps in the external filter see a lot less difficult signal, lower switching, clock feed through than the internal op amps
and the external op amps are easy to spec at much higher performance
they really aren't going to be limiting compared to the built in CMOS op amps
maybe not "flavorful" enough for you? - but you can just a niceness circuit after
I might add C to agnd in the input to the multiple feedback filter shown, but even as shown the op amps in the external filter see a lot less difficult signal, lower switching, clock feed through than the internal op amps
and the external op amps are easy to spec at much higher performance
they really aren't going to be limiting compared to the built in CMOS op amps
maybe not "flavorful" enough for you? - but you can just a niceness circuit after
The internal CMOS opamps will be the major bottleneck - since you cannot bypass them the next best thing for SQ is to load them as lightly as humanly possible. Which would mean a pair of JFET buffers. Failing JFETs, bipolars biassed as lightly as possible whilst still remaining in classA.
Once you have a low impedance output from the buffers you can use the standard opamp-based differential-SE circuit with included filter as you've already shown.
Once you have a low impedance output from the buffers you can use the standard opamp-based differential-SE circuit with included filter as you've already shown.
A pair of 2sk170/2sj74 per channel(would have to get the 2sj74's)?
I also have BC547/557, BD139/140, BC549C, A42's, 2N7000 and a few other.
This is my first all veroboard project and one of my first times with discrete components, so I apologize for my lack of knowledge.
Where would I insert that in the schematic?
Before the opamps? No way to eliminate the opa2134 and do the differential - SE and filter with transistors?
Or is there no SQ benefit at all from that?
The WM8740 sounds and measures fine when loaded with NTL-1 transformers, despite the opamps allegedly made of chocolate inside. Did so with the Twisted Pear Opus. Sold it, still have all the different outputstages I made for it. Transformers performed best when compared to the discrete stages I made for it.
Last edited:
You have R1 - R4 connecting direct to the DAC at present. Break the connections between those resistors and the WM8740 and insert the buffers in place of the wire. Assuming you use high gain bipolars (I use BC817-40) WM8740 pins 12,13,16 & 17 go to the base of these transistors, their emitters are then loaded with current sources. I show the idea in this post for another DAC : http://www.diyaudio.com/forums/digi...ign-mod-not-play-music-not-3.html#post4338438
Seeing as that schematic is designed for about 8mA current and you don't need anywhere near that much you should scale all the resistors (except the 4R7 base stopper) by a factor of 20 - that'll give 400uA in the CCS.
I rather suspect that the gains will be limited compared to the substantial effort required. What I would recommend though is increasing the working impedance of all the components around the opamps - I have found this to pay SQ dividends on my active crossovers. So whereas your circuit shows 2k4 input resistors, change them and the others of the same value to 24k, 150ohm change to 1k5. Likewize divide the cap values by 10 - so 3n3 becomes 330pF and 680pF goes to 68pF.
Finally if you're driving any length of cable from the output, don't use the opamp directly rather put another one of the 3 transistor buffers after it and bias it with 4mA (so double the resistor values in the schematic I've linked to).
If you have plenty of time you can wind your own, using RM14 ferrite cores available from Mouser. That works out cheaper than off-the-shelf normally and they do give very valuable isolation from common-mode noise in the case of a DAC's output.
That's an interesting idea.
I will look at that.
I'm in Sweden though and shipping might be very steep, otherwise Edcor has reasonably priced transformers.
I wonder if that type of cores or others that works as well might be available on eBay or similar sites?
I had a look on eBay, plenty of sellers of cores but none included the bobbins. Over on Aliexpress though I found these - rm14 ferrite core and bobbin 6 6 pin -in Cable Winder from Consumer Electronics on Aliexpress.com | Alibaba Group
Note that a full set of hardware should also include clips to hold the two sides of the core tightly together. In their absence though you could use tape, wrap a few turns of that yellow 'transformer tape' around the finished trafo.
Note that a full set of hardware should also include clips to hold the two sides of the core tightly together. In their absence though you could use tape, wrap a few turns of that yellow 'transformer tape' around the finished trafo.
Hi,
Ordered a bunch of bc817-40 and sop-23-3 adapters (would need it for the MCP120T-300 on the dir9001 board anyway).
The "L" in the schematic you linked to, will a regular ferrite bead work? Got 20pcs cheap, "wired" beads, ie looks like thick resistors in lack of a better description.
"Scale by a factor of 20", what did you mean by that?
English is not my first language and I'm self taught in electronics, so many terms that are natural for most of you guys...I have to look up lol
Oh, what supply voltages can this buffer work properly with?
If you're not so comfortable with SMT then those transistors can be TO92 types (BC337).
Yes a regular ferrite bead is fine - connect them as close as you can to the base/emitter terminals of the transistor.
By 'scale by a factor of 20' I mean multiply the resistor values by 20.
As for the supply voltages it really depends what levels of signal you need to handle. Assuming that the DAC puts out 3V p-p then having +5/-5V supplies would be the minimum I'd recommend and the max would be +20/-20V.
Yes a regular ferrite bead is fine - connect them as close as you can to the base/emitter terminals of the transistor.
By 'scale by a factor of 20' I mean multiply the resistor values by 20.
As for the supply voltages it really depends what levels of signal you need to handle. Assuming that the DAC puts out 3V p-p then having +5/-5V supplies would be the minimum I'd recommend and the max would be +20/-20V.
Thanks for your reply.
I'm ok with smd's, I have done a few ssop28, 0805 etc.
Thanks for explaining about the resistors.
I'll run the buffer off the same supply as the opamp then, about 14-15Vdc dual rail.
Now there's just the wait for all parts for this modular DAC to arrive.
First time veroboarding an entire build, but it's fun.
Looking forward to hearing how it sounds.... Don't use veroboard though, use some of that double sided PTH board that I linked to on the DAC thread. With veroboard I've gotten errors from forgetting to break tracks. Its also single sided which is considerably less flexible when prototyping.
I'm using single island solder point boards (FR4)
Trying to similate the circuit in spice, don't have the values for the ferrite beads though...
They are: "Inductor magic beads 3.5*6*0.8 AXIAL FERRITE BEADs"
I get a "singular matrix check node n005". N005 is the wire between the top BC817-40 and the bead.
Any hints on what to set as load and input for the sim?
Unfortunately I sold my etching equipment a few years ago. Thought I was done with my stereo system and hence DIY. Too bad as I actually got good at making nice PCB's.
Don't do enough DIY these days to justify getting a new etching setup.
Managed to do a sim of the buffer, no idea if I got it right though.
Set Vout from WM8740 to 2V as an arbitrary number just to see if I could sim it. Not sure about the load either.
Any way, here it is:
I used these ferrite beads in the sim, I have no clue about the values of the ones I ordered.
Does it look even close to accurate? Any hints, tips, and suggestions are appreciated.
Set Vout from WM8740 to 2V as an arbitrary number just to see if I could sim it. Not sure about the load either.
Any way, here it is:
I used these ferrite beads in the sim, I have no clue about the values of the ones I ordered.
Does it look even close to accurate? Any hints, tips, and suggestions are appreciated.
- Status
- Not open for further replies.