[Occam]

Josh - capacitive coupling takes many forms, but I believe your question is about what is coupling to what. When I refer to capacitive coupling in a transformer, I'm referring to the coupling between primary and secondary. In an ideal transformer, the coupling between primary and secondary is magnetic, largely determined by the core configuration. But anytime we have conductors in proximity to each other we have by definition a capacitor. You can see this in the capacitive measures quoted for various cables and their construction. As you know, with higher capacitance, we have lower capacitive reactance with increasing frequency, and therefore more signal couples. Which is what we don't want to happen in this application of transformers, as we want to pass our mains frequency component, but want to block/dissapate/divert higher frequencies, noise, which is exactly what capacitive coupling between primary and secondary is facilitating.

So how do we minimize this 'bad' capacitive coupling? We've two options -
1. Put a grounded shield between the primary and seconday which interrupts the dielectric field, and diverts that coupling to ground, or
2. Physically separate the primary and secondary, as we can do with dual/multiple windings/ bobbins on  EI, O, R, C and M cores.

The first works quite well, but is quite expensive as you can see by comparing prices between a typical toroidal power transformer and its 'medical grade' equivalent toroidal transformer, where the major difference in constuction is simply the existance of shielding between its overwound primary and secondary.
The second option if far easier and less costly to implement as its substantially a byproduct of straightforward automated constructon techniques.

[Christopher Witmer]

I have a choice between caps with internal bleeder resistors or caps without internal bleeder resistors. Which is better for our purposes?

Thanks!

Christopher Witmer
Tokyo

[Occam]

Christopher - without, because it gives you the option of adding one to your specific liking, or not.
If the bleeder is actually mounted like this -

rather than actually internally, as you can desolder or snip it off and choose to mount  (or not) one more to your liking.

As a side note - you can mount oil filled caps any way you want, right side up, on its side, upside down.... doesn't matter. Velcro or 'mushroom' strips that are adhesive backed to put on the cap and the mounting surface works a treat and you don't have to muck about with mounting clamps.

EDIT I just found those caps that you might have been referring to. Resistors internal to the caps are intended to drain potentially lethal voltages down below about 50v within a minute of removal of power. These are generally in the megaohm region, and (non empirically verified) they shouldn't effect objective or subjective measures. Their existance shouldn't enter into your purchase decision.

[Christopher Witmer]

Well, I just purchased some Magnetek 10.5uF/520VAC caps at $1 each. I doubt it makes much difference but one reason I chose these is that they have a rated tolerance of +/-3%.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7517133511

[Occam]

Christopher - Those magneteks appear to be an excellent choice, as they'll give you a good range for experimentation. 2 of these in series will give you 5.25uf, 1 alone 10.5uf, 2 in parallel 21uf, etc....

To date, we've evaluated various transformers with configured seconday voltages between 12 and 36v AC, all with rather excellent results with the 15uf 'betwixt and between' capacitance. (The currently specified A41 Signal transformer are effectively running the secondaries at 14vac as the their configured for 28vac due to the series configuration, but as they're fed from series connected dual 115 primaries, 230vac, but only fed 115ac the seconday puts out half voltage.)

My hesitancy in responding specifically to you questions has been largely based on my assumption that you're planning on using transformers that will put roughly 50 or 100vac on your secondaries. I'm not by any means suggesting this won't be as effective, its just that it is outside the secondary voltages I've experimented with. (and with higher secondary voltages, vigilance needs to be increased. While I lacadasically discharge my oil caps with a potential of 15v with a screwdriver accross the cap's terminals, I'd be a more vigilant when working with possible volatages of 50-100v+. Actually its potentially 1.414 X the AC voltage)
I'm currently experimenting with transformers that are going to put 80vac on the secondaries. Higher voltage secondaries are going to have higher inductance than lower voltage ones, all thing being equal. This will give me a handle on how Felicia will perform under those circumstances and what adjustments in capacitance values would be required, and I'll be able to comment more specifically.

[TimS]

Sorry if this query is a bit basic but I can't get my head around why you need dual primaries and dual secondaries (in series) for the Felicia.  
Will the Felicia work using just a single primary and secondary for each transformer?

cheers

Tim

[Occam]

Tim,

I've edited my post as Christopher's explanation below is really much better .......

[Christopher Witmer]

If you want balanced power, you need dual windings (or a center tap) on the output side.

Under the right set of design conditions, you could get away with just a single input on the primary side; i.e., if you had a transformer designed to take 120V input and split it into two 60V outputs on the secondary side. Such transformers are not all that common, but they do exist, and some of them were designed with balanced power applications in mind.

A much more common type of transformer is one that has dual 120V inputs and dual 120V outputs. This makes for a very versatile transformer. In North America, such a transformer can be used for 120V balanced power, but only by derating the kVA capacity by half. The same transformer, in the U.K., Australia, or another region where 240V is the standard household current, could deliver 240V balanced power without any derating whatsoever: just put in 240V and take out two 120V legs that can be joined to deliver balanced 240V power to the downstream equipment.

The Felicias use oddball (from the perspective of balanced AC power systems) transformers, because although they have the requisite dual taps, the secondaries are totally wrong voltage for standard AC powered equipment. The genius of the design focuses on their low price and says:

'What the heck, for this price we'll just use a symmetrical pair and boost the voltage back up to the proper level. And on top of that, we'll insert a hefty capacitor in between the two transformers, as that will give both power factor correction and also help convert normal mode noise into common mode noise that is more effectively eliminated. And because the capacitor is "trapped" in between two isolation transformers, any garbage resulting from its operation is "blockaded" on both the upstream and downstream sides.'

So, to recap with regard to your question, there is a theoretical (and actually encountered, but not all that commonly) transformer design in which a single primary side input is all that required to achieve balanced power. But in all cases, dual secondary side outputs are required for balanced power, as required voltage "X" is achieved by summing of two "1/2X" legs. If the secondary side lacks dual outputs, it may still be useful as an isolation transformer, but not for power balancing.

[tianguis]

All:
      I've been playing about with the Felicia since I picked up the original "Treasure Box" from Occam several weeks ago. I toted that one to a NY Audio Rave meet shortly after, where it was warmly received, providing nice power to a Modwright pre. The Felicia stayed in the system, with various tweaks, for the next seven hours, pulling duty with several components.
       Last week, Occam and I got together and tried a number of cap configurations in the Felicia, powering my Transcendent GG pre (Lowther ML TL's, CD-Pro2M source, modded T-Amp). We both preferred the sound of an Arcotronics 15 uf, 450 v. , bypassed with a Dayton .1 uf 400 v. and .01 uf 600 v. AudioCap Theta. We also tried an Auricap 1.0 uf by itself, which seemed to lose a little too much "sparkle".
      During the week, I built another Felicia, using VH Audio cryoed 12 AWG wire for the tranny/cap runs. Yesterday, we got together again and swapped in an Aerovox 14 uf. 180 v. oil cap. After a few minutes it was plain that we were on to something. We tried the same bypass configurations and, in this case, the Auricap was clearly superior. This combo has taken the Felicia to a new level in my system, which previously had power supplied by a modded Transcendent Balanced Power Supply. Of course, the usual caveats apply: YMMV, power down and drain the cap when fiddling, don't get zapped.
       I'm using the Felicia with small Bybees on both output legs. The Bybees are a nice touch, but don't nearly have the impact of proper caps   and bypassing. I have a few more bypass configurations to try. I'm awaiting arrival of a few parts for a second Felicia, to power the CD-Pro2M and/or Hagerman Cornet phono pre. The Felicia provided a nice boost to the CD-Pro when I tried it. Multiple Felicias will be cumulative in effect.
       The next NY Audio Rave is going to focus on power conditioning. A goodly number of commercial conditioners will be tested. Should be revealing.  

Regards,
Larry Welsh

[Occam]

Folks-

As a result of Larry's great efforts, I'm updating the first post of this thread to reflect his findings. We've come up with 2 alernative caps to the previously recommended Amrad. The first is a Arcotronics 15uf motor run cap (plastic bodied, not 'protected', but rated for accross the line use @ 450 VAC.) This is available for $2 at BGmicro, the same place that you obtain the recommended transformers from. The second is an oval metal protected motor run cap, a Aerovox (Now Parallax ne Magnetek) 180 VAC 14uf cap, also $2, but from Hosfelt.
IMO, both are substantial upgrades over 15uf Amrad 370VAC oil caps. I'd hypothosised that the higher voltage rating of the Arcotronics was the reason for my preferrence. And then we evaluated the 180VAC 14uf Aerovox.... go figure. At this point, I don't have a clue.
Both Larry and I prefer the Aerovox over the Arcotronics, both bypassed and unbypassed. But we're both old farts and have a long history with tube equipment. Others of the Spectral/MIT 'school' might well prefer the Arcotronics from BG. YMMV.
We evaluated a number of bypass caps, a copper leaded 30nf silver mica from my private stash, a combo of a .1uf Dayton foil with a .01uf AudioCap Theta, and a 1uf 450v Auricap and various combinations. We both preferred the Auricap on the Aerovox by a large margin, as it made everything mo' betta.
Again, I want to emphasise that none of these selections are definitive. Why did Larry only evaluate a 1uf Auricap instead of a .47, .22 or .1? Because thats all he had at hand. Same goes for the main cap. Larry, Josh and I have evaluated about 5 different main caps, but this hasn't been exhastive. Felicia is an 'open source' project and its improvement largely depends on the input and efforts of the DIY audio community. I see lots of caps in the 12-20uf range that I'd love to try out as the main cap. There are regularly scads of them up on Ebay. But given that I've now about 18 transformers laying about, whose only purpose is now doorstop, Mrs. Occam is understandably annoyed, so I'm trying to minimize the deliveries from the UPS man.

I just got a call from Larry. He told me that the orientation of the Auricap makes a significant difference. I trust Larry's ears. I've no problem with cap orientation with regards to signal direction or  as bypass to ground, but in this particular situation where the caps are accross out of phase floating AC, I'm at a total loss for an explanation.
EDIT Actually, it is quite logical. The main cap  is also a metallized polyprop cap, and the efficacy of the bypass would be impacted by the orientation of both with respect to each other.

C'mon folks!!! We need more input from builders.

[Christopher Witmer]

Sorry, can anyone give me a quick rundown on the theory behind having tinier bypass caps in conjunction with the big sucker?

Thanks!

Chris Witmer
Tokyo

[LanceL]

I hope this wont be considered a temporary hijack of the thread as I have the Transformers on order for the felicia.....in the mean time I pulled out a Oneac CS1115 I had picked up used from someone in the past and decided to clean up the execution to have as a fun comparison when the balanced power project is complete

The oneac has a typical filter config(Pi?) composed of 2 inductors and multiple caps before the transformer Primary and 2 metal can 5uf (polyprop in oil?) in parallel after the secondaries.

Those will/have been removed, I plan on hardwiring a PC to the primary (through the existing switch/breaker and probably some MOV's and a .047uf Xrated cap across line to neutral)) and hardwire a cord to the secondary as well with a female IEC.

My question is both the primary and secondary have a white/green/black wiring(actually the primary also has 2 additional wires Red+Brown that run to a terminal barrier but are unconnected to anything...I assume for different voltage selects)...........the primary side green was Grounded to chasis
the secondary green was connected in parallel with the white (neutral) on the output caps..........additionally the original output duplex outlets had the neutral connected to the gnd screw and then on to the chasis gnd.
I'm assuming this is some kind of technical gnd?

My question is are both of these Green wire from the Elect shield and should just be run to chasis gnd along with the PC's green/Gnd

Hope someone can clarify and thanks for the time
lance

[Occam]

Christopher - simply do a search on diyaudio.com or audioasylum.com in the 'tweaks' or 'tubediy' forums on 'capacitor' and 'bypassing'. You'll see much argument on how to do it, but basic agreement as to the why, to provide a parallel low impedance path for high frequencies, as the construction of large capacitors leads to high inductive reactance at higher frequencies, which the bypass capacitor seeks to address.

[Occam]

Lance - I believe you are correct, the separate shield on the secondary winding allows the connection to a separate 'techincal ground'. The equitech site has far better info than I could provide.
http://www.equitech.com/support/techgrnd.html
But if you'd go to the bother of a technical ground, why not just run a dedicated line with a separate isolated ground which does much the same...

For your purposes just connect all the grounds (green wires) to a single 'star ground' connection on your chassis. This would be the grounds from your cords (both input and output) and the primary and secondary shields.
On the output  the connection of the secondaries white (neutral) line to the green (ground) is reestablishing the neutral-ground bonding as required for a new service point by the NEC. The exception to this is isolated power as required where exposed to flamable anesthetics and 'wet' conditions in hospital environments.
Whether you do this neutral-ground bonding in a star to that central grounding point, or as a daisy chain, depends on how obscessional you are. I'm not going there....

If you need to pursue this, PM me, and I'll move your question and my response to a new thread.

[LanceL]

Thanks Occam for your help...I have PM'd you for one brief follow-up question

[Christopher Witmer]

Does anyone know the lowdown on a UTC HIT-3 transformer? (TF4R03YY) This is a milspec transformer, weighs 15kg (33 pounds), is double shielded, and has a single set each of primary and secondary taps, 1:1 115V / 3.5A. I am a bit amazed that this 115V 3.5A transformer is so heavy. It is being sold locally for the equivalent of US$25; I'm wondering if I should pick it up or pass on it . . .

[Gordy]

One more Felicia up and running!  I'm using the Aero 14uF caps and  1.0 + .01 bypasses, because that's what I had already. It's one of two going into the same box, the other only lacks the fuses and output cord... kinda got excited and had to hear it in action.  It's now driving a transport, tomorrow I'll pull it back out and finish the second one... to drive a tube pre.  

   so far...

[tianguis]

All:
      I've had a second Felicia running my CD-Pro2M and Hagerman Cornet for a while and the results were great. The second one used a single Auricap 1.0 uf 450 v. bypass cap. Both use the Aerovox 15 uf oil cap.
      Today, I got in a bunch of Auricaps to mess with. Occam came over and we tried a bunch of combos. Not mentioning the combos we felt didn't work, I made up a .47, .1 and .01 uf Auricap combo for my first Felicia (powering a Transcendent Grounded Grid pre) and the benefits were almost immediately evident: greater space and separation, better image focus, greater detail over the 1.0, .1, .01 combo.
      I then stuck the old bypass combo on the CD's Felicia. Wow! Space and detail increased in spades. Paul liked the increase in PRAT and musicality. I appreciated the ability to distinguish every voice (instrument OR vocal) in complex music. Bass is way tighter and better controlled. Over-run and ringing on cymbals and piano notes are gone. This combo is the clear winner so far, by a wide margin. Of course, YMMV, though I don't think it will.

Regards,
Larry Welsh

[JoshK]

So your saying the .47uf was the ticket?

[tianguis]

Josh:
       Not just the .47, but the .47, .1, .01 combo. The .47 alone doesn't cut it.

Larry