[tourtelot]

This is the worksheet on Tagent's TREAD site:

Power Supply Parameter Estimator

My inputs:
Power source: Raw AC
Transformer Unloaded Voltage: 21.2VAC
Rated Voltage: 15VAC
Rated Current: 1A
Allowable Line Voltage Variation: 10VAC
Regulator Dropout Voltage: 2VDC
Desired Regulated Voltage: 21VDC @ .125A

Results: unregulated voltage ≈ 24–30.4 V, ripple ≈ 0.5 V, drop across regulator ≈ 2.5–9.4 V, regulator temp ≤ 9 °C over ambient

D.

 

[tomb]

Again, 0.5V ripple is intolerable, IMHO.

Also, does the PPAV2 only draw 0.125A? There's more than that just in the buffer, I think. Are you calculating the loss across the regulator? Also, 21.2VAC is way too high for a 15VAC transformer, IMHO - even lightly loaded. Maybe a volt or two - that's it, in my experience. Perhaps those are bogus inputs/outputs.

 

[tourtelot]

Hi Tom-

I am not trying to be a jerk here, but I am simply getting more and more confused by all this. Assuming that there is a reason to use a TREAD with a PPAv2 (better sound, quieter, tighter) and not any old 24VDC wall-wart, why is it so difficult to spec an AC/AC transformer that will allow me to input between 20-24VDC out of the TREAD into the PPAv2? It shouldn't be this hard. I haven't used a scientific calculator since I was in college, and here I was, pouring over "sqrt2" formulas.

Maybe I should leave this to the experts and go buy a Cute Beyond and be done with this. Sorry, just frustrated which doesn't bode well for building the PPAv2<g>.

D.

 

[tomb]

Quote:

Originally Posted by tourtelot /img/forum/go_quote.gif Hi Tom- I am not trying to be a jerk here, but I am simply getting more and more confused by all this. Assuming that there is a reason to use a TREAD with a PPAv2 (better sound, quieter, tighter) and not any old 24VDC wall-wart, why is it so difficult to spec an AC/AC transformer that will allow me to input between 20-24VDC out of the TREAD into the PPAv2? It shouldn't be this hard. I haven't used a scientific calculator since I was in college, and here I was, pouring over "sqrt2" formulas. Maybe I should leave this to the experts and go buy a Cute Beyond and be done with this. Sorry, just frustrated which doesn't bode well for building the PPAv2<g>. D.

It sounds like you're having trouble getting accurate inputs to Tangent's calculator. I'm not an expert on it, but it's provided good info as long as you provide accurate input - but that's the hard part.

It's not that complicated. Use any old 24VAC walwart with enough current rating (500ma probably). Once rectified, it'll supply approx. (1.414x24) - 1.4 VDC. That's about 32V. It's actually more like 0.8 - 0.9 times that value, but close enough. Yes, the LM317 needs 5V to burn to provide maximum ripple rejection, but Tangent's TREAD design takes care of a lot of the ripple with the good layout and the large filter electrolytic. So, that works out to providing a maximum of about 27VDC output from the TREAD with a 24VDC walwart.

That ratio holds for any amp, as long as you size the walwart so it's about 3/4 loaded or less. IOW, it's not that complicated. If the PPAV2 draws about 300ma, then a 500ma 24VAC walwart through a TREAD will easily supply 24-27VDC to the PPA with maximum ripple rejection (ripple ~= 0.060VAC).

 

[tourtelot]

24VDC to the PPAv2 would be perfect. Thanks, Tom. Big heat sink on the regulator, right? Burning 5v sounds hot.

D.

 

[rds]

I am powering a PPA right now using a TREAD and a 24 VAC wallwart and under load (ie with the PPA running) the voltage after the diode bridge is 33.8V. So you'll need to drop 9.8V to get 24VDC.

FYI the wallwart I'm using is Triad Magnetics WAU24-750.

EDIT My wall voltage here is 125VAC, thus resulting in a higher than expected rectified voltage. 32V rectified is what you'd expect for normal 110VAC wall power.

 

[guzzler]

Quote:

Originally Posted by tourtelot /img/forum/go_quote.gif 24VDC to the PPAv2 would be perfect. Thanks, Tom. Big heat sink on the regulator, right? Burning 5v sounds hot. D.

P=IV, so assuming the 300 mA quoted above, this gives 1.5 W across the regulator. Doesn't need a massive heatsink. Heatsinks are rated (simplistically) in K / W, so pick one around 15-20 and you'll be fine.

As an aside, the problem you had with Tangent's power supply estimator is this: transformers have a higher output voltage when they're not loaded - this is the "Unloaded Voltage" value in the estimator, and you'd have to measure this for a particular transformer. This is not the rectified voltage. It doesn't matter particularly in these low current applications, but definitely has to be watched as the current increases.

 

[tangent]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif You need to burn 5V at the LM317 to gain the highest ripple rejection.

Really? The datasheet claims a maximum of 2.3 V, and then only under worst-case conditions. 2 V is more typical.

Not that you'd want to push the limit that closely, else you'd drop out of regulation during even a mild brownout. This is why the estimator defaults include 10% line voltage variation.

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Again, 0.5V ripple is intolerable, IMHO.

The estimator calculates pre-regulator ripple only, because it doesn't assume any particular regulator design. It's general-purpose, useful with any linear regulator.

You can expect 80-90 dB of ripple rejection with a TREAD, turning that half volt of pre-regulator ripple into about a quarter millivolt of post-regulator ripple.

I've changed the estimator to say "pre-regulator" before "ripple", to avoid this confusion in the future.

Quote:

does the PPAV2 only draw 0.125A?

Yes, "only".

This is actually a conservative estimation value. In practice, a PPA "only" has to draw about 0.1 A to sound its best.

(Yes, I get that many other designs are even more wasteful. I just shake my head when I see things like this...the cans only need about a tenth that to lay waste to your stereocilia. < 10% efficiency is wasteful enough in my book.)

Quote:

Originally Posted by tourtelot I am simply getting more and more confused by all this

If you want easy, buy an Elpac WM080 and be done with it. It's a perfectly reasonable choice.

Curse the estimator now if you like, but if you'd tried to build a TREAD and a randomly-chosen wall wart without calculating anything, you might be cursing your burning TREAD later, not understanding that you'd chosen the wrong power source to go with it.

DIY is always harder than off-the-shelf. There are compensating benefits, else we wouldn't DIY.

 

[tourtelot]

First off, thanks to all who have spent time to try and help me with my question. I REALLY do appreciate if even through my frustration.

Quote:

Curse the estimator now if you like, but if you'd tried to build a TREAD and a randomly-chosen wall wart without calculating anything, you might be cursing your burning TREAD later, not understanding that you'd chosen the wrong power source to go with it. DIY is always harder than off-the-shelf. There are compensating benefits, else we wouldn't DIY.

Jeez, all I said was I was confused and frustrated. If I didn't want to do it right, I wouldn't be here. What I am having trouble with is not only the learning curve on the process but some obviously cross-purpose advice. That is to be expected in a forum setting but it makes for some crazy-making.

Tangent: rds is running his PPA as below and it seems to be working fine. Does his set-up bother you on aesthetic grounds for being wasteful or is it gonna blow up?

Here's my deal: I need a good sounding headphone amp the WILL NOT blow up for use in the professional arena. I am an experienced bench tech, although limited in my theory. Part of the problem was in my expectation of designers giving more specific advice. Fine for me, probably really hated by the hard-core DIY to whom this forum is targeted. Tangent; you do that very well with the PPA parts choices, and my sense of what it will take to build the basic PPA is pretty clear. But the TREAD AC input was really cloudy for me. It seemed that the estimator worked backwards, the builder needing a specific final AC output which resulted in a transformer value that couldn't be purchased. Also, to ask for a critical value that can only be measured on an already purchased unit and not spec'd on a data sheet seems a bit of a problem. I am not a tweeker, so in a way, a simple kit is probably more up my alley. So I get your recommendation for the "merely okay" store-bought supply, but I would like to do better and so, I wander on.

I am sure that I will end up with a nice amp and it will probably serve me well. And, I am learning a lot of new stuff. But maybe I need to be more specific in my requests. All I really needed was a referral for a proper AC transformer. rds did that, and thanks. Now, if I can only find caps of the proper value that are the correct size to fit on the board. One of the problem with better technology is shrinking size. Bank to the search.

D.

 

[tourtelot]

Ooops! Sorry. Just re-read the spec sheet on the Triad 24VAC transformer and guess what? There is the unloaded voltage value. MY BAD!

D.

 

[tomb]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif Really? The datasheet claims a maximum of 2.3 V, and then only under worst-case conditions. 2 V is more typical. Not that you'd want to push the limit that closely, else you'd drop out of regulation during even a mild brownout. This is why the estimator defaults include 10% line voltage variation.

The National Semi data sheet I have doesn't really specify the minimum volts for maximum ripple rejection. However, every spec listed is based on "Vin-Vout=5V". This includes the three graphs showing ripple rejection. While those three graphs are used to display differences in Cadj vs. output voltage, output current, and output frequency - everyone one of them is based on "Vin-Vout=5V." Perhaps I shouldn't have made the conclusion that the LM317's maximum performance of 80dB ripple rejection required 5V, but the implication seemed pretty clear:

I would agree that actual regulation is achieved with about 2V, it's just that I thought the stand alone ripple rejection (minus a nice layout with ripple-filtering caps) was greatest with the 5V difference that the data sheet references. It's a moot point anyway, since we know that the high ripple rejection in the TREAD is achieved with less than that.

EDIT: To the OP: based on Tangent's estimate of the PPA's current draw, I'd go with a WAU24-450 (500ma) - should be plenty and is fairly common. There's a WAU24-200, but it's only 62 cents cheaper at Mouser, but a lot closer than comfortable to the actual load IMHO.

There's also the option of a linear-regulated walwart from Jameco. They have a 24VDC, 500ma linear regulated walwart here: http://www.jameco.com/webapp/wcs/sto...ductId=174861&
It's only $12.95. I have a couple of them and the measured ripple is in the same ballpark. The difference with one of these or the Elpac is that you can't eke out that extra voltage to get ~27VDC with that ripple.

 

[tangent]

Quote:

Originally Posted by tourtelot /img/forum/go_quote.gif Jeez, all I said was I was confused and frustrated.

Don't be mad, I was just trying to be sure your mind is in the right place for this work. It seems it is.

Quote:

obviously cross-purpose advice. That is to be expected in a forum setting but it makes for some crazy-making.

True, but it's also one of the best things about a forum. You could have emailed any one of us directly, and gotten one answer, and been on your way. But is the answer right, and if there are a range of "right" answers, is it the best? How would you know without getting a second opinion? Here, we get to argue and debate our way towards some middle ground answer, which I find is usually more valuable in the end. It's messy, but better than the alternatives.

Quote:

Tangent: rds is running his PPA as below and it seems to be working fine. Does his set-up bother you on aesthetic grounds for being wasteful or is it gonna blow up?

My comments above about waste are just me going off on a Tangent. (I do that.) They're nothing to do with your TREAD configuration choices.

To get closer to the ideal of 100% efficiency, you'd have to switch to a class D amp (or similar), which the PPA most certainly is not. So, moot here.

Quote:

It seemed that the estimator worked backwards, the builder needing a specific final AC output which resulted in a transformer value that couldn't be purchased.

Propose another way it could work.

A goal-seeker would be ideal, but there's no way it's going to know about the types of transformers and wall warts on the market. There are no standards, and I'm not about to go downloading datasheets for weeks to build up parameter tables for everything available. Even if I did, people would still be upset with it because it wouldn't recommend parts from a distributor in your country, or parts currently in stock, or parts from your favorite distributor or whatever. No matter how you slice it, much of the part selection is up to the person driving the calculator.

Given that, what is there to improve?

Quote:

Also, to ask for a critical value that can only be measured on an already purchased unit and not spec'd on a data sheet seems a bit of a problem.

As you've found, power sources from electronics distributors generally do have this info in their datasheets. You don't get it for surplus stock, or wall warts purchased at retail, etc., but that's just the nature of such sources.

Also, most of this problem is just due to your inexperience. You quickly start gaining an intuitive sense for what should work even with incomplete information. The best way to cross that gap is to start building things.

 

[Voodoochile]

Quote:

Originally Posted by tangent /img/forum/go_quote.gif The best way to cross that gap is to start building things.

I was just thinking the same thing.
After you build a PSU, especially from scratch, many of these questions will be worked out via the process. It is good to ask questions up front, too, but the build is a memorable growth experience.

Once you start looking deeper into data sheets, you will note things like the regulation % value for unloaded toroidal transformers, and how that relates to loading. For example, bigger units have tighter regulation, resulting in lower overvoltage when unloaded, but conversely they require more load to bring down.

Hammond also has a few great datasheets on their website regarding specs and what to expect based on different rectifier configurations and filter configurations. The more of this information you absorb, the clearer the relationships between the various values become.

By knowing what your end needs are, one of the main benefits of building a specific PSU for a project is that it can be sized to work optimally, increasing efficiency and with better performance than a generically spec'd unit that will work, but may not be optimized. That is, you can achieve as good or better regulation with better regulator efficiency with careful planning, resulting in a quieter and more capable PSU that generates less heat and uses less power, for example.

Finally, be sure to examine the datasheets for a given regulator from more than one manufacturer, too. Although they are often sold as compatible replacements, some companies are much more prolific with their data. Linear, National, and TI all have very different "application" material for a given device, with the same general guidelines up front.

 

[tourtelot]

Now I have some empirical evidence to run by the experts. I built up a TREAD PS today, a very pleasant little chore. It fired right up and got about to doing its business without a fuss.

But here's the data, as read by my Fluke DVM and my 'scope.

Wall wart is a nominal 18VAC. Unloaded voltage is 27.03VAC. Also, BTW, the loaded voltage is the same as measured with it attached to the TREAD. The minimum VDC that I can achieve is 22.1 with the trimmer turned all the way down, far higher than the "sane" voltage of 19VDC as recommended by Tangent's estimator. I have kept the VDC below 25VDC because I don't need any more volts but it seems as if there is still a lot more "up" to go. I have rolled the voltage up to 23VDC which should be perfect for running my PPAv2 and the regulator is cool as the proverbial cucumber; maybe a degree or two hotter than it is when it is powered down. I will install a heat sink but I think that the .125A draw of a PPAv2 shouldn't tax it.

Why does this real-world experience differ so much from the estimated experience? By the way, I see zero noise component on my scope; it is a very stable and clean DC.

I'll order the parts for my PPAv2 tomorrow, but I may hold off on the 24VAC power supply if the 18VAC that I already own will do the job. Any thoughts?

D.

 

[grenert]

What's the voltage from your home AC outlet? It can vary quite a bit from the expected value. Higher voltage on the transformer primary will give you higher-than-rated voltage on the secondary.

EDIT: Also, if the TREAD isn't supplying anything, it's not going to draw much current. So, your transformer is still essentially unloaded. Do you have a dummy load or something to supply with the TREAD?