[feles]

For a DAC, if I want to power some 3.3V and/or 5V chips from the same elpac PS I’m using for the PPA... how to I get those voltages from 24V?
I found these:
TL750L05 for 5V and then using a TPS76133 for getting 3.3V from 5V, but I’m not sure if those are acceptable if I want to actually get clean sound out of this thing... 5V is for the DAC chip and the logic level of the S/PDIF receiver, and the 3.3V is for the analog and digital levels of the S/PDIF receiver. I have pretty much no experience with this whole 'get one voltage from another' thing so any educational material/suggested reading would also help.

 

[Wodgy]

Since you're using wall power, there's no need to use an LDO, unless you're worried about heat dissipation. A standard regulator will do fine and will usually have lower noise. Consider the UA78M33 for 3.3v and the TL780-05 for 5.0v. You won't be able to feed 24v into either of these, though, if I remember correctly, but they can handle 12v inputs, so put them after the rail splitter.

 

[aos]

Amongst TPS79xxx are some of the highest spec'd integrated regulators available. Classic 3-pins don't hold a candle to some of these. However they are mostly SOT-23 package which is VERY small - I hope you know how to solder them.

I am not familiar with the particular chip you're using though. I use TPS79133 in my latest DAC. But unless you are willing to use such a small package you'd be better off with some classic regulators or at least look into SOIC ones, like ADP3303-3.3 from Analog Devices, which I used before. Standard regulators like 78L05 (and there should be 78L3.3 or something like that as well) will do fine, though look around because specifications vary wildly for those "generic" chips. Get them in TO-92 and you'll have a nice and easy supply.

 

[jeffreyj]

Quote:

Originally posted by feles For a DAC, if I want to power some 3.3V and/or 5V chips from the same elpac PS I’m using for the PPA... how to I get those voltages from 24V?...

This is a big drop in voltage, and if your DA draws even a small amount of current your regulator(s) are going to get into trouble quickly because voltage dropped X current drawn = heat dissipated. For example, if your DAC draws 100mA at 5V (a reasonable estimate) then your regulator will have to dissipate 1.9W. This doesn't sound like much but it is enough to require using the TO-220 package and a small heatsink.

There's no need to use Low Drop Out (LDO) regulators in this application at all! In fact, this is the worst place to use those cranky things! You want to be using standard NPN pass transistor-type regulators here, like the 78xx and LM34x series.

One thing to consider would be to cascade two regulators in series. Say a 12V regulator feeding the 5V/3.3V regulator(s). This splits the dissipation up between two (or more) devices and improves the ripple rejection and regulator figures at the final output. That's just icing on the cake, though, as the real benefit is simply the division of dissipation.

As far as suggestions as to what to read to become more familiar with voltage regulators.... well, I'm afraid I don't personally know of any online sources suitable for a beginner. The Art of Electronics (Horowitz and Hill) is an excellent book on electronics in general, though, and highly recommended if you wish to do this as a serious hobby (and virtually every engineer I know owns a copy, so it's good for the not-beginner as well).

 

[aos]

Quote:

One thing to consider would be to cascade two regulators in series. Say a 12V regulator feeding the 5V/3.3V regulator(s). This splits the dissipation up between two (or more) devices and improves the ripple rejection and regulator figures at the final output. That's just icing on the cake, though, as the real benefit is simply the division of dissipation.

That is actually a good suggestion, but there is more to it. No modern regulator that I'm aware of can tolerate more than 20V, and even those are very few. So you need a first stage regulator that will tolerate 24V and will drop voltage enough for perhaps more esoteric units. Your 7805 could be the first stage but the dissipation might be too much. Digital section will typically draw say 40mA, (24-5) * 0.04 = 0.76W. That's quite a lot, you'd have to use 7805 instead of 78L05. Since your ELPAC is already regulated, second stage is not needed for the benefit of ripple rejection, but splitting dissipation is not a bad idea.

I don't know why jeffreyj doesn't like LDO's, most of the stuff made this days is LDO (probably because the batteries are order of the day) and I never had any stability problems with any that I used.

 

[jeffreyj]

Quote:

Originally posted by aos That is actually a good suggestion, but there is more to it. No modern regulator that I'm aware of can tolerate more than 20V, and even those are very few....

Err... I know the LM340-x and 78xx regulators are a bit long in the tooth, so may not qualify as modern, in the way you see it, but they certainly will accept more than 20V input. Very few LDO regs will tolerate more than 20V, true, but plain NPN pass transistor types will do so easily!?!

Here's the link to National Semiconductor's web site for the LM340 Series Regulators .


Quote:

I don't know why jeffreyj doesn't like LDO's, most of the stuff made this days is LDO (probably because the batteries are order of the day) and I never had any stability problems with any that I used.

I don't dislike LDO's, I dislike the misapplication of LDO's. You use LDO's when you still need a regulated output with only 1V or less of overhead. Besides the fact that their very raison d'etre is negated when much more than 1V overhead is available, they also have some peculiarities that argue against their use unless LDO operation is actually needed, like a much steeper slope of rising output impedance with frequency because of the PNP or synthetic PNP output stage. Stability issues can be a problem, despite your experience to the contrary, aos, especially when thousands of uF of rail capacitance are added after the regulator - LDO's usually require a rather narrow range of capacitance present at their outputs to remain stable; more or less and they can really start to kick and scream.

 

[Chipko]

I found a nice +/-12V +5V transformer I'm using for a project that has some logic circuits and relays.

Not the cheapest way but since relays are pretty hungry it was the best way by far.

 

[aos]

Quote:

I know the LM340-x and 78xx regulators are a bit long in the tooth

When I talk about "modern" regulators, it is exactly 78xx and LM317/337 and modern clones kind of stuff that I'm trying to make look old. Not that they shouldn't be used, on the contrary. Their widespread use is what's keeping their prices way down, and they are available in many packages and a lot of people know how to apply them. But the world moves on.

Today's regulators don't seem to suffer from problems you're mentioning quite that much because manufacturers were focusing on them for many years. For example, there are many that aren't that sensitive o the "stability region in regards to output capacitor" phenomenon that you mention. ADP330x is advertised as "anyCap" because as long as the cap is over 1uF it can be anything - ceramic, electrolytic, tantalum, whatever, and there's no upper limit (or you could use smaller 0.47uF ceramic). TPS79133 also specifies minimum requirements but doesn't have upper limit and yet it is a very fast regulator which PSRR holds into several MHz (most are gone after few 10's of kHz). Some other regulators in the TPS7xxxx series do suffer from stability region so they have upper limit on capacitance.

In this case, if he's powering DAC electronics, he's not going to be hanging thousands of uF after the regulator anyway.

 

[alsq]

What would be wrong with dropping voltage very simply,
like with a power transistor biased with a zener? The
real regulation happens afterwards anyway, and a
cheap NPN in TO220 or TO3 package can dissipate a
lot of power in the safe zone (well, it's abot 20V-21V at
worst Vce, so we're talking ~2A for a TIP31 @ $0.30
and I'm sure there are cheaper solutions). Add one zener
and one resistor, and a heat sink perhaps. Do I
say $1.00 altogether? Why would this not work?