[Occam]

a picture of some of the various caps and transformers Josh and I have evaluated -

[Occam]

Isn't more capacitance always better (up until you start rolling off the 60Hz signal)? What is your criteria for optimal?

For a single capacitor, along with higher capacitance, all things being equal, comes higher inductance which limits the efficacy of filtering action. This can be addressed with multiple caps/bypassing.
We've attempted to keep this project simple. I'll repeat myself, this project, which presently has 3 components in the filtering/balancing circuit, the 2 transformers and a single capacitor is not definitive. Presently, one other builder (to my knowledge) is experimenting with bypassing with excellent results. This will involve simply paralleling additional caps with the existing one. Frankly, if anonther, better main cap is found, I'll lose no sleep if you have to discard that $4 capacitor.

Optimal means in this context, better that the last iteration. Build it if you want, and embark on this gradient search. There is no need to justify its topology or choice of components as Felicia 'speaks' for herself.

[jdybnis]

My apologies if I came across questioning the efficacy of the design. The balance between price and series inductance is good. I was curious because I thought there might be some need to match a particular capacitor with any particular transformer.

I have found the Relcap PPFX to work very well for bypass in a similar application.

[Occam]

jdybnis - The only apology should come from me for being snarky. Your questions are perfectly reasonable and straightforward. At minimum, Felicia can be viewed as a 3rd order lowpass filter consisting of the cap and flanking inductances, and while my meter can measure capacitance, I lack instrumentation capable of measuring inductance, or a signal generator, so the development has been substantially subjective.

A quick confession about the schematic. The X2 capacitance shown across the primaries of the input transformer has lilltle to do with the efficacy of the filtering behavior of the unit which is feeding a source. It is simply there conveniently to provide a modicum of filtering (as that capacitance is in parallel with the mains circuit) for other components which don't have powerconditioning.

Welcome to AudioCirlcles. I'm looking forward to your (and others) feedback and comments that will make Felicia a better project.

Felicitations,
Paul

[ctviggen]

Projects like this always make me wish (1) that I'd taken a filters class in college and (2) that I had more time.  I think it'd be fun to program the data into an analysis tool (like Matlab or Spice; probably Spice would be better) and then determine the best combination of capacitors and transformer to use.  Unfortunately, I'll be lucky to have enough time to just build the exact version of this.

[JoshK]

The thing that I really enjoyed about this project was that we were using a great design to get good results rather than just throwing money at a big ass transformer & fancy chassis.  We (mostly Paul) scoured the surplus market looking for suitable caps and transformers that were easy to obtain and inexpensive but had the right properties (shielding, split bobbins, low bandwidth, etc) to make the most of it.

Armed with some dime store parts we got busy mocking up different designs using a DPDT center off switch to so that we could do pairwise comparisons.  We listened to many different combos playing the same couple of cuts over and over again.  Remember the cable shootout at Daniels?    This is why the project took my wife's name since she put up with all of our never ending obscessing.

The goal here was to, in the end, have an inexpensive by highly effective PC that we could share with this community.  To this end we suceeded I feel.  We already know that many who have heard it like the results.  Just how well it stacks up to the "big boys" has yet to be determined, and is somewhat moot.  Of course I'd love it to beat their pants off but even if it didn't, all is not lost because we haven't taken this to the hilt.  It is still what has been done for cheap.

Paul and I have shared this project with all of you in the hopes of starting an 'open source' project if you will.  Since it is simple enough to build and cheap enough to give it a shot, we hope that others will join in building and experimenting with us by trying other various configurations (different caps, bypassing, magic fairy dust, etc) and reporting their findings back to this forum for the benefit of all.

[Super-Gonzo]

8 transformers and various bits now on the way to my doorstep. I'll try to take lots of pictures.

4 outputs in one 17x13x5 case... hopefully things will fit the way I have it planned in my head  

On a related note, anyone see any problem powering a LED w/ AC using the circuit shown here:

http://www.qsl.net/yo5ofh/hobby%20circuits/led_circuits.htm
(scroll down to "AC Line powered LEDs")

[tianguis]

All:
      I've got parts on order for two. In the meantime, I've been toting Occam's pirate chest around and will have it at the NY Audio Rave tomorrow.
      I've been messing with bypassing the cap and it's a very worthwhile "tweak". I've tried a bunch of film/foil caps, with quick disconnects to clip on the main cap. The combo which works the best so far (the Felicia is only powering my Transcendent Sound Grounded Grid pre) is a Bennic .1 uf with an AudioCap Theta .01 uf. The improvement in micro detail is large. Dynamics and bass response have improved, also, which I didn't expect.
       I've also outfitted a pair of Bybees with quick disconnects, and have swapped them in and out. They nicely add some smoothness on top.

Larry

[JoshK]

Larry, are you bypassing with the bybees or do you have them in series? Large or small slipstream ones?

[tianguis]

Josh:

       The Bybees are in series with the output legs. The ones I'm using are the second-generation small red ones, rated at 4 amps.
       Interestingly, I think the bypass caps are a greater improvement then the Bybees. Of course, the whole effect is cumulative.

Larry

[Christopher Witmer]

Felicia is such a good idea that I now want to build Feliciazilla. I have several interesting transformers to work with:

1) 5kVa isolation transformer from an APC Matrix 5000 UPS. I have the entire UPS but don't use it and am thinking about salvaging the transformer from it. I think this EI transformer must weigh at least 100 pounds. I still have not determined if it is capable of balanced output, but either way, I think it has the potential to be useful in this project. On the input side it has taps for 240/230/225/220/208V (and maybe some other voltages, but nothing in the 105-125V range). On the output side, it has taps for some of the input voltages as well as for 120V. I have never tried it, but the "isolation unit" housing the transformer is supposed to be capable of use independent of the rest of the UPS. (The entire UPS consists of three separate units that work together, and the isolation unit can continue to supply electricity to the load equipment while the other two units are being serviced.) The isolation unit also has EMI/RFI and surge suppression, and it weighs about 150 lbs. If I use this transformer, I am not sure whether it would be better to remove it from the isolation unit, or just use the entire isolation unit as-is.


(By the way, APC just recently discontinued this line and replaced it with a follow-up line, so these UPS units might be available for free or cheap. I got mine for less than $300 a few years ago, and recently an acquaintance got one for free that was about to be junked.)

2) Signal HPI-20 (2kVA) transformer. This 41 pound isolation transformer has dual taps both primary and secondary, and is thus capable of balanced output.

3) Union Denki MB-H-10 (2kVA) transformer. This 38 pound isolation transformer has dual taps both primary and secondary, and is thus capable of balanced output.

I don't know if this is a no-no, but I would like to use the 5kVA transformer for the input side of the Feliciazilla, and the two 2kVA transformers (independent of each other) on the output side. (If it turns out that the 5kVA transformer is capable of balanced output by itself, it may make sense to completely rethink this strategy.)

Can I have the 5kVA on the input side, "feeding" two 2kVA transformers on the output side? Or is this going to cause problems?

By the way, please remember that I intend to balance both 2kVA units, so they will actually be considered 1kVA balanced power supplies. Also, I intend to place a maximum load of 800VA on each, so the whole system will have a maximum load of less than 1.6kVA. And most of that will be in the form of two class AB 350W subwoofer amplifiers, so most of the time the load will be less than that.

This may be overkill, but I have the transformers, so I figure, why not? Besides, I hope to set up a home recording studio, and I want to be sure the noise floor is as low as possible.

And no, just in case anyone is wondering, I do not listen to "heavy metal" music . . .

Please give me your advice, suggestions, etc. Thanks!

Christopher Witmer
Tokyo

[Gordy]

Josh:

       The Bybees are in series with the output legs. The ones I'm using are the second-generation small red ones, rated at 4 amps.
       Interestingly, I think the bypass caps are a greater improvement then the Bybees. Of course, the whole effect is cumulative.

Larry

Hello Larry,

Realizing they could well be different animals, in the case of the BPT units Chris Hoff uses the Bybees immediately after the breaker, then follows the "high current" filter - transformer - caps...  just thought it might be of interest!

Gordy

[tianguis]

Gordy:
       Thanks for the info. When I was modding my Transcendent Sound Balanced Power Supply, I tried Bybees in various positions.  Placing them on the outputs seemed to work best. I'll have to give them a try on the input side. In a perfect world, Bybees on inputs and outputs would probably work best, but I only have two which aren't being used somewhere else.

Larry

[Gordy]

That certainly makes sense as the Bybees' were far more noticable, in my experience, directly connected to speaker drivers than before the x-overs. Thanks!

[Occam]

Christopher,

I'll be building my own version of Feliciazilla (with appropriate apologies to JoshK's wife for the terminology) with some 600va transformers this week. To be honest, I'd feel more comfortable with your endeavor if you'd build a smaller version for source components first. This would allow you to evaluate the concept first, without a large financial outlay. Furthermore (although I've no idea as to the typical power quality in Japan), this would allow you to evaluate the potential improvements available accross the whole audio spectrum, rather than those specific to a subwoofer.
Those Union Denki transformers look absolutely gorgeous! Tape wound M-cores with dual bobbins to boot, all shielded!

The idea of using a large isolation transformer to feed smaller balancing transformers certainly makes sense. My previous efforts at a single input transformer and multiple output transformers was not not particularly successful, but I really did do it in a half assed way. As I understand it, you presently have a Signal HPI-20 which you could wire the input in series, and take the 2 secondaries and feed each to the separate, single  primary of your 2 UnionDeneki transformers, and get balanced out from each of their series connected dual secondaries (grounding the center connection of the secondaries).
But prior to doing this, I'd strongly suggest using the 2 Union Denekis wired primary to primary for a single balanced output limited to 200va to see what you get. Dunno what to suggest for the capacitance betwixt and between, but I'd start with 15uf of motor run cap and work upwards.

Hopefully, folks will start to post their experiences with various bypass capacitors on this thread over the next few days.

[Christopher Witmer]

Questions, questions! I would be ever so grateful if someone could give me some guidance.

What is the method or reasoning behind the choice of motor run capacitor in any given "Felicia-type" balanced power system? Is it dependent on the specifications of the transformers being used, or on the specifications of the downstream devices, or is it a combination of both?

How does one know when one has the right capacitance value?

Is there a standard test that can be performed to check, or does one simply use one's ears? (Is there a safe way to induce noisy, crappy line conditions to see how well Felicia-type balanced power systems are working?)

If value "X" seems to work well, what would be the results of "2X, "10X," or "100X" values?

Can you create a dangerous situation through the wrong capacitance value?

Would it be okay to use a bank of multiple capacitors, or should only single capacitors be used? (It occurs to me that if a bunch of good capacitors of relatively small values could be found, the total capacitance value could be adjusted by adding in or removing capacitors from the bank.)

Is there an ideal voltage range, or is any voltage above a certain minimum voltage fine?

Thanks!

Christopher Witmer
Tokyo

[Occam]

Christopher - Excellent questions.... Dang!

The choice of motor run capacitors for the Felicia was based on a number of reasons, both techincal and practical.

They're 'protected' and self healing, like 'X2' capacitors. Capacitors specified for 'accross the line' use failure mode is an open circuit rather than a short. When subjected to electrical conditions that would pierce the metallized polyethelyne they don't short. When subjected by a surge that would typically cause other capacitors to blow up, they purposely disconnect electrically, permenantly. You loose the capacitor, but aren't showered by shrapnel. These capacitors are found throughout our appliances wherever there is an AC motor, air conditioners, refrigerators, etc.... They don't heat up under constant AC conditions like many lesser quality capacitors.

While these are rather 'common', don't underestimate their quality in audio applications -
http://home.versatel.nl/geenius/Cap.html
and when it comes to Japanese built ASC motor run capacitors which you should find relatively easy to source -
http://db.audioasylum.com/cgi/search.mpl?forum=tubediy&searchtext=asc

While most builders will be using the specified Signal A41 transformers running with a secondary voltage of 28vac, which would allow the use of caps specified far lower than the 370VAC of the recomended Amrad caps (with the fused input 'hot' line providing requisite protection), I believe your intent is to use isolation transformers which would use the Japanese mains voltage of 100vac on the secondaries where the cap is placed. I'd not use anything less than a 180VAC rated run cap.

From a practical perspective, these very high quality caps are industrial commodity products, far less expensive than bespoke audiophile capacitors. They're even less expensive in the surplus market. I do not recommend used caps as the potential savings are simply not worth the uncertainty. They are terminated in multiple .25" quick connects allowing easy and convenient connection of bypass caps, etc....

The choice of the 15uf cap was done by JoshK and myself, when we found that the paralleling of a 5uf polyethelyne oil and foil cap ($140 new) added no perceptible improvement. Not overly scientific, but given our lack of instumentation, the best we could (and were willing to) do. This choice was specific to that Signal A41 transformer.
In your specific case with different larger, transformers, I'd start with 15-20uf and work upwards by paralleling additional caps. Twist all connections to and from transformers and caps.
Until I'm shown otherwise, I don't see any reasons not to use these overspecified motor run caps.

As Tianguis has pointed out, Felicia does benefit from capacitive bypassing. I defer to his experience and expertise, and am looking forward to his update.

[Christopher Witmer]

Is a 100uF/370V can *too large* for what we are talking about?

[Occam]

Dunno, try it. It really would depend on the characteristics of the transformers you're flanking it with, and some other things......

Realize that (assuming you're running your windings feeding that capacitor a 100v AC you're going to be running around 600VAR (voltage amps reactive) and you're actually doing some serious power factor correction, passively. Which is a whole 'nuther discussion. Nor is this necessarily a bad thing. Just substantially beyond what I'm willing (or presently feel competent) to discuss. I'm simply going to say, do a google search on 'power factor correction' and point out that both Furman,  who uses 80uf of line voltage passive power factor correction capacitance, and Audience similarly has some unknown amout of capacitance doing the same thing within the bowels of their new raved about Adept.
Two things are going on here.
1. That capacitor is functioning as an element of, at minimum, and 18db/octave low pass filter. This is great, as long as the attenuation at your 50/60hz line frequency is not material. As capacitors are not perfect, the inductance of any 100uf capacitor I'd use is going to require bypassing to compensate (its that 'in parallel' thing....) for its loss of efficacy at higher frequencies.
2. If you look the waveform of our mains power (especially in urban and/or industrial areas, the voltage and amperage waveforms are not coincident, in phase, nor do they necessarily look like each other. If you don't follow, go back and actually read what you found in the google search on 'power factor correction'. That bulk capacitance is 'correcting' that misalignment due to inductive loads, motors, transformers, etc.... The higher the ac voltage that capacitor is accross, the higher the VAR. This is why industries prefer to do their bulk passive power factor correction at 14,400 V AC, at their main distribution transformer.

Now I've thoroughly not answered your question, I'll summarize by saying, dunno, try it.
You're not building the Felicia, but rather her testosterone laden elder brother. I can't be more specific without knowing a whole lot more about the specific transformers, about the Tokyo power grid, and having the time and inclination. But I really would suggest starting with that 15uf and working upwards.
Prior to blazing new ground, start with, if not those specific Signal transformers, a similar Japanese sourced equivalent, 2  100-200va split bobbin EI cored transformers with dual primaries, and 15uf, to power a single source componet. I realize you've 'other fish to fry', but I really can't help you go directly there  

[JoshK]

OK,

I am going to ask the newbie question, partially because I am not quite certain myself and partially for the benefit of other newbies who might not know either.  You mentioned a few times why EI core transformers & interwinding shieldings are good because they minimize capactive coupling.  

My naive understanding of cap. coupling is that it "couples" the ground to the hot line sending "stuff" to the ground.  Can you explain why this is a bad thing and why it might sound bad?

From www.atis.org

capacitive coupling: The transfer of energy from one circuit to another by means of the mutual capacitance between the circuits. Note 1: The coupling may be deliberate or inadvertent. Note 2: Capacitive coupling favors transfer of the higher frequency components of a signal, whereas inductive coupling favors lower frequency components, and conductive coupling favors neither higher nor lower frequency components.

I guess my own confusion lies with why passing HF crap to ground would be bad.  The only reason I can see is maybe this modulates the ground or contaminates other circuits in the same circuit.  This is done intentionally in a PSU after the rectifier, but here it is ok because of the transformer behind it?  If the capacitive coupling is inside the tx then why is it then different?