[FallenAngel]

Hey guys,

So I snatched up a Zhaolu D2.5A and since the AD1852 has V-Out balanced outputs, I want to put a proper balanced output stage in it. Any suggestions?

I know that there is the ZapFilter (at $275), but I would prefer something DIY if possible. I know that TwistedPearAudio is coming out with the Mesh, some new output stage but until that comes along, I'm looking for alternatives.

Now, I'm not exactly knowledgeable in terms of DAC output stages, is there anything particular about them or can I simply stick 4 JISBOS boards after this thing and call it a day?

Thanks

 

[millwood]

I don't know what a balanced output stage is: your headphone has a tied common so it cannot be powered by a "balanced" output stage.

but if you are looking for a balanced-to-single-ended output stage, there are plenty of solutions, two of them are presented in the very datasheet for ad1852.

 

[FallenAngel]

Uhm... my headphones do not have a tied common grounds, they are terminated in 4-pin XLR

 

[millwood]

in that case, take a look at those adsl driver datasheet, particularly ad815 and see the balanced to SE converter? figure 51.

you will likely need to size up those two 100ohm input resistors but those chips (cfb dsl drivers) will blow any amp out of water, spec-wise and they sound very good to my ears - but you do need to be VERY careful with decoupling and layout.

 

[FallenAngel]

As nice and simple as the AD815 looks, I already have a nice balanced amp on the workbench (Beta22 and also building a 4x JISBOS standalone buffer), mostly just looking for a balanced filter/output stage. Thanks for the recommendations though, it's a nice read into this.

 

[millwood]

sounds like you are looking for a balanced line receiver: it will take the balanced output from the dac, and then output it to a balanced (or SE?) power amplifier.

if it is to drive a SE power amp, many instrumentation amps will do, like shown in the ad1852 datasheet. But keep in your mind that they are not truly balanced the way they are configured in a typical ina134 datasheet.

if it is to drive a balanced power amp, and you want to remain truly balanced, look for a two-chip solution (as shown in the ad815 datasheet), or a three-chip solution (I think the ad815 datasheet has such examples as well, with or without the use of a transformer).

obviously, you don't have to use the ad815 and other vfb/cfb opamps will work for such an application but the concept is the same.

 

[FallenAngel]

I think I'm very thick-headed here, but I kind of do want to stick with a discrete solution (aka not opamp-based).

This will be an output stage directly attached to AD1852 capable of driving a pair of interconnects to a headphone amp or preamp.

It will be have to be balanced, not single-ended.

 

[Pars]

If it is a V-out DAC, many of the discrete stages such as Pass D-1, etc. would be a waste as they are geared towards good I/V conversion. You could build your own zapfilter:

http://www.diyaudio.com/forums/show...?threadid=20294

 

[nikongod]

The zapfilters have returned to north America at partsconexion (although they may have sold out by now...)

The DAC chip in the 2.5C is much nicer then the DAC in the "a". I would get a C before hacking up the zhaolu.

 

[millwood]

you may or may not have realized this but depending on your headphone amp, you may or may not need to have a balanced line receiver here.

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif This will be an output stage directly attached to AD1852 capable of driving a pair of interconnects to a headphone amp or preamp. It will be have to be balanced, not single-ended.

you have two decision points:

a) balanced input: sounds like that's what you want to go.
b) balanced output: sounds like that's what you want to go as well but it is not crystal clear in the discussion.

going with balanced output will require a balanced headphone amp.

the ad1852 datasheet doesn't provide much info on current output but those things don't generate much in current output so budgeting 1ma per output pin is about reasonable.

there are a few ways to buffer the output with a discrete solution.

a) emitter followers: they are the simplest and provide good performance. but they are ac coupled.

the lowest cost approach is to use capacitors. However, that may not be as desirable.

the highest cost approach is to use audio transformers. you will need two transformers per channel and that is going to set you back significantly if you go with jensen.


b) discrete diamond buffer: great ac performance and big current capabilities - to the point that you can use them to drive your headphones directly. But you may have to live with dc offset. this means nulling may have to be necessary at the headphone amps.


c) discrete opamps: they are quite a few of them. you can, again, go with a 2-opamp or 3-opamp solution, per channel. it is going to be expensive and time consuming and in the end I am not sure what you will get out of it.

the best approach is actually to NOT use any line receiver and see if you can directly hook up your headphone amp to the ad18152.

 

[-=Germania=-]

You can build PASSIVE OUTPUT STAGE since the Zapfilter is difficult to get a hold of and cotdt who PM'ed me introducing it to me has a Zapfilter and uses the passive output stage saying that the passive one has better soundstage and treble.

I built a temp one for testing it out and it is indeed a VERY CHEAP (I built mine from parts I had and some resistors from the lab) improvement. Even though right now I am using Electrolytics bypassed with PIO, it is considerably better than using the upgraded op-amps in detail, floor noise, and basically all other aspects (the extension of highs and lows has been greatly increased).

Resisitor Value: 1.2K ohm (Has to be this value for the load)
Capacitor: 3.3uf or above

Connector:
Red = R+
Yellow = R-
Black = Ground
White = L-
Black = L+


HERE are some PICS:

*This is Cotdt's Zhaolu*

CREDIT FOR THIS GOES ENTIRELY TO COTDT!

 

[minivan]

i assume the Vin is straight from the dac output? is it enough to drive the circuit?

 

[FallenAngel]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif The zapfilters have returned to north America at partsconexion (although they may have sold out by now...) The DAC chip in the 2.5C is much nicer then the DAC in the "a". I would get a C before hacking up the zhaolu.

Really, CS4398 is "better" than AD1852? I read quite mixed impressions here so I snatched up the A version because there really aren't that many AD1852 DACs that I've seen.

Quote:

Originally Posted by millwood /img/forum/go_quote.gif b) balanced output: sounds like that's what you want to go as well but it is not crystal clear in the discussion.

What might have given you that idea, I said I wanted balanced from the start, including in the thread title.

Quote:

Originally Posted by millwood /img/forum/go_quote.gif the lowest cost approach is to use capacitors. However, that may not be as desirable. b) discrete diamond buffer: great ac performance and big current capabilities - to the point that you can use them to drive your headphones directly. But you may have to live with dc offset. this means nulling may have to be necessary at the headphone amps.

That's the idea, but which buffer to use is the question.

Quote:

Originally Posted by millwood /img/forum/go_quote.gif the best approach is actually to NOT use any line receiver and see if you can directly hook up your headphone amp to the ad18152.

Since the AD1852 is V-Out, I'll need to get rid of that DC, that means an amp that handles this (which I don't and will never have) or removing offset at the source which brings us back to other solutions.

Quote:

Originally Posted by -=Germania=- /img/forum/go_quote.gif You can build PASSIVE OUTPUT STAGE since the Zapfilter is difficult to get a hold of and cotdt who PM'ed me introducing it to me has a Zapfilter and uses the passive output stage saying that the passive one has better soundstage and treble. Resisitor Value: 1.2K ohm (Has to be this value for the load) Capacitor: 3.3uf or above Connector: Red = R+ Yellow = R- Black = Ground White = L- Black = L+ HERE are some PICS: CREDIT FOR THIS GOES ENTIRELY TO COTDT!

Definitely a nice thing to look into and I definitely will. I don't think I have any 3.3uF, but since my balanced amps have 100K input impedance, I should be able to use 0.47uF.

 

[millwood]

Quote:

That's the idea, but which buffer to use is the question.

you wanted to have a discrete design, and you wanted to have a diamond buffer.

if you put the two together, why do you still wonder about which buffer to use?

 

[cotdt]

Quote:

Originally Posted by nikongod /img/forum/go_quote.gif The DAC chip in the 2.5C is much nicer then the DAC in the "a". I would get a C before hacking up the zhaolu.

Really? I like the A version much better than the C version, I think it is the more accurate sound, the AD1852 is what converted me to a lifelong Analog Devices fan!

Quote:

Originally Posted by FallenAngel /img/forum/go_quote.gif Definitely a nice thing to look into and I definitely will. I don't think I have any 3.3uF, but since my balanced amps have 100K input impedance, I should be able to use 0.47uF.

It's an RC highpass filter so 0.47uF would give too high a corner frequency. You can calculate the -3dB cutoff frequency by using the equation:

I used a 4.7uF capacitor, which gives a -3dB point of 28Hz to insure that all the bass comes through.

This output stage sounded great without a low pass filter, which should not be required as the DAC's spurious noise is above the audible frequency range. You could also use a transformer instead of the RC circuit. Transformers have a center tap and are thus already balanced, so you would only need two.

The Zhaolu was meant to be balanced. It has a balanced DAC and also comes with space for two balanced XLR outputs:

All just my opinion, of course.