PDA

View Full Version : Power Supply Question (but not what you think)



ezkcdude
2006-04-21, 11:57
So, this looks to be the first thread about power supplies, but my question is not related to the SqueezeBox, but to DIY components in general. I am in the early stages of designing my first DIY DAC, which will be used with my SB3. I'm trying to decide whether to use an external power supply, either a wall wart or open frame, or to make one using a transformer. Either way, the DAC will still need internal regulators for the digital and analog sides. Frankly, I'm a little afraid of getting electrocuted, so I'm leaning towards a lower voltage DC (i.e. using external supply) power supply. Well, if anyone has any thoughts or expertise in these matters, I'd love to hear from you. Over the next several months, I hope to be able to post some results here of my progress.

seanadams
2006-04-21, 15:19
Have you determined what regulated voltages will you need internally?

If you need a balanced supply for the output stage (eg +/- 12V) then you probably want to build an internal supply, since you'll also probably need +5 for the DAC.

However, if you can run everything off a positive input (as in SB3) then a 15VDC linear wall wart (plus internal regulation to one or two voltages) would be good choice - you won't need a lot of juice - probably less than 250mV.

ezkcdude
2006-04-21, 16:13
Have you determined what regulated voltages will you need internally?

If you need a balanced supply for the output stage (eg +/- 12V) then you probably want to build an internal supply, since you'll also probably need +5 for the DAC.



Thanks, Sean. That pretty much confirms what I had been thinking, i.e. taking the internal transformer route. I probably will be using some kind of active I/V conversion in the output stage, and some of the dual supplied op-amps run better at higher voltages (even up to +/-18V!). I'll just have to make sure I stay grounded. As far as transformers go, is there any need to spend big bucks on these? I see that some transformers sell for hundreds of dollars, while others can be had for $10-20. Well, I guess if that question could be easily answered, we wouldn't have all the PSU threads in the Audiophile forum :)

seanadams
2006-04-22, 10:56
As far as transformers go, is there any need to spend big bucks on these? I see that some transformers sell for hundreds of dollars, while others can be had for $10-20.

Cost is mostly a function of power capacity (VA or watts) and if you're just driving a DAC and line-out amplifier you can use very small transformers. The other thing that drives up the cost is when you get something non-standard with custom windings.

There is debate over toroid (donut) vs laminated (box), eg: http://www.altavistaaudio.com/plitron.html

If you have good regulators it shouldn't make any difference IMHO.

Search digikey - they have all kinds...

occam
2006-04-23, 13:02
There is debate over toroid (donut) vs laminated (box), eg: http://www.altavistaaudio.com/plitron.htmlI see no debate. What I read is that Michael Elliott has always preferred, rather enthusiastically, EI over toroidal transformers until he sourced the proprietay Plitron NBT transformers. van der Veen's patented implementation, specifically deals with transverse mode noise from the mains line with what is essentially a built in powerconditioner. Nor is that technology specific to toroidal transformers (other than by restricted licensing). The performance of Plitron NBT is comesurate with their substantial price.
The advantages of a standard toriodal transformer over a laminated transformer is a smaller radiated EM field, and with capacities over 100va, smaller size and weight, lower cost, and higher efficiency. A standard multiple bobbin laminated transformer consistently betters the standard donut when it comes to attenuation of line noise. But given your assumption that you can always throw more and more voltage regulation at the problem, I understand your indifference.



If you have good regulators it shouldn't make any difference IMHO. Is that opinion based upon empirical or thought experiments?


Search digikey - they have all kinds...But sadly, they'll find no toroidal transforrmers there whose performance begins to approach that of the Plitron NBT transformers. There are other techniques, IMO equally effective, for attenuation of transverse mode line noise.

seanadams
2006-04-23, 15:16
But given your assumption that you can always throw more and more voltage regulation at the problem, I understand your indifference.

That's a gross misstatement of my post. I said IF you have good regulation it shouldn't matter, not that one should "throw more and more voltage regulation at the problem". I doubt if you would conversely suggest that skimping on (inexpensive) filters and regulators because one has higher confidence in their (costly) transformer would be sensible.



Is that opinion based upon empirical or thought experiments?


Yes, I have tested transformers and am well aware of the general differences. However, you're taking my comment wildly out of context. I think you should read this thread again from the beginning to understand where the original question was coming from. I was offering a practical suggestion: 1) there is debate, here's a a starting point if you're interested 2) in THIS situation I wouldn't worry about it. Perhaps I should have said focus on getting the circuit up and running with whatever transformer you have lying around and THEN try opimizing. Surely you'd agree there are much more important considerations to be thinking about right now, such as what chips to use?

ezkcdude
2006-04-23, 16:10
Surely you'd agree there are much more important considerations to be thinking about right now, such as what chips to use?

I didn't intend THIS power supply thread to start yet another flame war, and I hope it doesn't go further than Sean's post. It seems to me that Occam's post was a blatant attack, and not necessary, at least not for my benefit.

As to the point Sean raised in the end, actually, I have thought about the chips: CS8416 -> AD1896 (upsampling to 96 kHz) -> PCM1794;

I will also use an external 24.576 MHz XO to clock the AD1896 and PCM1794. I am going to design the PCB using Express PCB and hopefully, have a 4-layer board made. The issues remaining are A) the above-mentioned transformer vs. external PSU question and B) how to do the I/V conversion. Other than these two things, I need to figure out how to connect all the pins, and how to make sure the board is laid out properly, with respect to minimizing noise, obviously. I'm not sure if I said this in my first post, but I am not going to be done with this any time soon. It's basically a long-term pet project for me.

aberdeencomponents
2006-04-28, 11:33
I have thought about the chips: CS8416 -> AD1896 (upsampling to 96 kHz) ->


But I see no reason for the Ad 1896 chip .....as the CS8420 can do the whole job alone.

This is what I got in front of me now.... the CS8420 is a better choice for the aplication.

ezkcdude
2006-04-28, 12:44
I just took a look at the data sheet, and maybe it's just me, but it seems way more complicated to use than the other two chips combined (CS8416+AD1896), which I've seen specific schematics for. The other thing that jumped out at me is the dynamic range for the 8420 ASRC is not as good as the AD1896 (128 dB vs. 144 dB). Also, I don't see a MCLK input, so do you have to rely on the internal clock, or is there a way to re-clock externally? What D/A are you planning to use with it, if I may ask? From what I've read on various forums, most DIYers are going with AD1896.

EDIT: My bad, it looks like RMCK is the input for an external clock. I'll have to drudge through the other 98 pages and see what else I can figure out!

aberdeencomponents
2006-04-28, 13:31
I just took a look at the data sheet, and maybe it's just me, but it seems way more complicated to use than the other two chips combined (CS8416+AD1896), which I've seen specific schematics for. The other thing that jumped out at me is the dynamic range for the 8420 ASRC is not as good as the AD1896 (128 dB vs. 144 dB). Also, I don't see a MCLK input, so do you have to rely on the internal clock, or is there a way to re-clock externally? What D/A are you planning to use with it, if I may ask? From what I've read on various forums, most DIYers are going with AD1896.

EDIT: My bad, it looks like RMCK is the input for an external clock. I'll have to drudge through the other 98 pages and see what else I can figure out!

Don't worry about the dynamic range spec... there be limiting factors in the circuit, that you will never see 144db, such as a Dac or optamp. I am very familiar with the 1896a chip as it is used heavily in my TacT Mods... (see: www.mauimods.com)
the 8420 chip is used in my Northstar Transport mods (Join my Yahoo group "the real tact hackers" and see photo section) In both my full SB mods,and Northstar Mods, the Northstar can, and will blow a SB3 out the water in sound quality. And there is a good reason for this... the Northstar uses the SC8420 chip, that upsamples the signal to 24/96. It uses a triangular dithering function, that seems to give a digital recording that Slam, Bam, Details mam that is lost during the recording. I know that a high end digital source company,Name withheld) who contacted Sean Adams about his technology, uses that same Dithering technology I mentioned. And Sean can EASILY implement this chip in his SB design, that would yield tremendous results in sound quality. So, Here I am, doing this as I type this post..
In my work, we use hacksaw to remove the entire analog section,
where the SC8420 chip can reside, or better yet,,,, remove all analog I/c's and will have all the proper voltages needed to run the Sc8420, as a added bonus.
Anthony
www.aberdeencomponents.com
www.mauimods.com