No, I think I get the point. With a passive RC output, the DC on both legs is the same, i.e., there's no differential mode DC and the output is balanced. Now, we put an input transformer on the far end, which we can do because there's no differential DC. So we have the same galvanic isolation, the same removal of the common mode DC, but we now have 40-60dB better CMR than with an output transformer on the sending end.
There's still a transformer, but we've gotten better performance. What am I missing?
There's still a transformer, but we've gotten better performance. What am I missing?
Of there's nothing else between the O/P pins of the DAC and the I/P of the amps, why not place the Tx's there and get the CMR improvement?
But you do. Anti-imaging filter for proper reconstruction.
edit:I may have been unclear about the RCs. They're not a high pass filter, they're a low pass filter. See Jan's excellent article on how the networks are set up.
Where's the reconstruction filter then? Do you use a DAC with digital reconstruction filter?
jan
Really?
Jan, it's the DCX2496 that Pano and I are using. So anti-imaging would SEEM to be necessary, unless there's something clever Pano is doing that I'm not getting.
I have not found it necessary with the AKM, ESS, or CS chips. The transformer is all that is needed. These chips all have digital low pass filters built in, AKAIK. The external anti-imaging (smoothing) filter does not seem to be needed when using iron on the output of these chips.
Perhaps the transformer is working as an anti-alias filter, I don't know. It would take spectrum analysis at frequencies above the capabilities I now how to find out. If the aliased signals are in phase, then the transformer would kill them. But I don't know that they are in phase.
Bottom line: If the out of band harmonics are there, they don't seem to be causing any trouble.
Perhaps the transformer is working as an anti-alias filter, I don't know. It would take spectrum analysis at frequencies above the capabilities I now how to find out. If the aliased signals are in phase, then the transformer would kill them. But I don't know that they are in phase.
Bottom line: If the out of band harmonics are there, they don't seem to be causing any trouble.
high internal upsamping and digital filtering puts the images up at the upsampling frequency - even 8x 44.1 KHz upsampling will be attenuated a good bit by most xfmr - and more recent chips use much higher upsampling ratios
but I would advocate explicit anti-imaging filtering
but I would advocate explicit anti-imaging filtering
I think you mean "anti-imaging." In any case, you said earlier that the transformers are very wideband, so they're probably not doing much to filter. Some equipment will be sensitive to that, other equipment won't be. If I had a very wideband power amp on the far end, I'd be more worried than with, say, a transformer coupled tube amp.
My point remains, though- if you start on an even footing, that is, you don't care about imaging artifacts (in the digital sense, not "localization"), for the same parts count (a transformer only), changing from a balanced output transformer on the sending end to a balanced input transformer on the receiving end gets you 100-1000 times better noise rejection.
Basically the same thing, it's just that "anti-imaging" is the term used at the output. A low pass filter to get rid of out of band energy.
While technically I have nothing against what you say, in practice I have not found it to matter. You know me, I'm a simple guy with a simple mind and simple tastes. The transformers work. None of the many I've sold or demoed over the years has ever caused the least problem on anything downstream.
The caps are not needed because the transfo blocks the DC. The LP filter shouldn't hurt and may be good engineering practice, but I've not found it necessary. See the post by JCX above.
And yes, you could put the transformer at the receiving end, if you want - I've done it. That gives you rejection of cable borne noise as well. But in a practical sense, I've never found it to be needed. If it were a long run, like 100' or more, a transformer at each end might be merited.
Just to be clear again, the caps aren't blocking DC in Jan's circuit. They're shunted between + and - output legs. I'm even a simpler guy than you. 😀
Oh, gottcha. Don't think I've seen the circuit, or don't remember it. Yeah, I can see a LP filter in there, tho I haven't needed it.
So how DO you block the DC?
So how DO you block the DC?
OK, out of curiosity what follows the DAC out in yours?
I'm designing the PCBs for my DCXs now, a job I loathe.
Oh, Stuart {blushes}, you are so kind.
Edit: Forgot the new avatar. I have a weakness for blonde Russian/Ukrainian pianists of prodigious talent. I'm listening to her now.
OK, we're talking about the same thing really. You just put the transformer near the next stage, that's all. Probably better, but sometimes not practical. With 3 to 6 foot cables, not a big deal. But at least you've got your cables DC polarized!
I did try the PLLXO trick, but found it didn't help me any. Didn't seem to hurt, either.
Brett, you are looking right sexy these days. I hardly knew you. Blonds and pianos, what a combo.
I did try the PLLXO trick, but found it didn't help me any. Didn't seem to hurt, either.
Brett, you are looking right sexy these days. I hardly knew you. Blonds and pianos, what a combo.
Going back to the main idea of this thread, IMO the new records (new pressings) do not sound as big as the originals because all the info has been digitally remastered. There are records that sound exactly like in CD format (Bjork).
There are others that even with it´s digital path sound really good (Keb Mo).
The best pressings are from the 80´s.
Bass can be quite good on vinyl but digital is hard to beat.
The overall picture is better on Vinyl because there is no 20khz firewall and the resulting soundstage is much bigger. IMO the high freq information is very important for the reproduction of the soundstage and imaging.
- Status
- Not open for further replies.