[JohnR]

Always wondered what a "four-pole" cap was, finally got around to fishing about on the jensen site:



http://www.jensencapacitors.com/audio/white-papers/fig2og3.html

Not sure I understand why this difference is significant, but anyway...

[Occam]

Dan,

The local Ratte Shaque was out of the requisite power resistors to series/paralell (what else is new?) so I did the rational thing. I called the manufacturer. The regulation on that $10 transformer is 10-11%, which is workable, but probably not ideal. But I did find a more appropriate potential line. I'll coordinate through Hugh, and see if this OEM manufacturer would sell 'onesies-twosies'.

I understand your desire for something 'better' than a toroid transformer. Their concentric windings make small power ones quite acceptable as an audio output transformer, i.e. their inherent bandwidth passes the increasingly common crap that pollutes our electrical mains. And a R  or double C core, wound appropriately without interleaving might well be the ultimate for a linear supply.  But far more cost effective would be to limit the power transformers bandwith by using a far more available, and less costly dual bobbin laminated EI core, just be prepared for even more critical layout and/or some Mu metal., and forgoe any hopes for that 2RU height case.

Toroids dominate the 'high end' market for cosmetic and ecomomic reasons. They've become cheaper in larger sizes, they have lower EMI, and they make for sleek components. They've also facilitated the proliferation of hideously expensive power coditioning and voodoo. (I cringe when I see a balanced isolation transformer based on a toroid.)

While there might be pallitive methods as the ferrites mentioned  (or a 'tuned' cap accross the secondary)  each increment costs more and gains less, and ultimately leaving oneself in the embarassing activity of 'polishing a turd''. Rather that spending thousands on power conditioning tweeks,, or sourcing an expensive custom R/Double C made from unobtainium, consider an appropriate EI cored dual bobbin, and deal with the inevitable height, and annoying, but tractible, EMI. What makes sense for a commercial manufacturer may not be as clear cut to a DIYer.

Edit - and if you want to go aesthetically GainClown, with a separate enclosure(s) for tranformer(s) and amp(s), you can minimize those negatives, and with the proper design, go all Artisanal Aesthete Audio(tm) on us.

[Propstuff]

But far more cost effective would be to limit the power transformers bandwith by using a far more available, and less costly dual bobbin laminated EI core, just be prepared for even more critical layout and/or some Mu metal., and forgoe any hopes for that 2RU height case.
 

If you look at the Gallery on the Aksa site you can see my Aksa100N using EI trannys built into a 2RU case.
These are a good old-fashioned local Ferguson brand, good quality, about 4% regulation (I seem to recall) and include a copper EMI shield around the windings.

They are mounted about 80-90mm away from the input stage of the boards, and have given no hum problems at all.

So: it not only can be done, but it's a viable option. The key is finding good quality Trannys, and at the price that Chinese made Toroids are comming (into this country at any rate), you might not find that an EI is the cheapest option. (?)

cheers,
N.

[Occam]

Propsuff,

Lovely amps! While you may think your lack of hum is due to the interwinding copper shield, I believe the shield's purpose is minimized interwinding capacitance, and minimally effects dispersal of EMF.. Your lack of hum is, IMO, rather a tribute to your good design and implementation.
It would be nice to compare your amp,  head to head, with the exact same, but using bog stock toroids. But alas, AKSA builders are such a peculiar lot, with no 2 exactly alike.

Gone,

My search results are mixed. Unless willing to incur the cost of customizatiion,
stock, dual bobbin EI transformers don't reach that desired 5-7% regulation until around 175VA.. But I did find an adorable little tank, a classic 'rectifier' transformer, much like the one descibed by Propstuff. In discussing the requirements for ideal audio power transformers with an engineer at Signal Transformer, he made 2 points -

1.  Scheduled International standards (in which the US sadly lags) regarding noise put onto the powerlines is causing a reconsideration of 'old school' EI implementions due to a toroids inherently high bandwidth. Well, actually he admitted that it was more towards further development of switching power supplies.

2.  When I mentioned the preferred regulation range he interrupted with "We've been making what we call 'rectifier' transformers with regulation between 5-8% for years for capacitor input variable demand powersupplies, because this is ideal for situations where the transformers reactive component(s) can limit the caps charging which can come in massive spikes."
And then I realized that he was actually paraphrasing exactly what Hugh says in his website's discussion of power transformers... It was spooky.

http://www.belfuse.com/signaltransformer/SuppMag.asp
and download the PDF for 'rectifier power'.

its the DL36-2, a 72VA, dual 18V secondary tranny @$37ea. The slightly cheaper 36-2 lacks the dual prmaries which allow international use. Weighing in at 3.8lbs, its 1lb heavier than my dual bobbin EI cored 100VA unit, FWIW. The engineer described these trannys as 'conservatively specified'.  It might be ideal for a Nirvana++ 25 implementation, but this is speculation on my part.

[gonefishin]

Thanks so much for all the discussion guys!  

   Occam, thanks for the leg work!  

   The price certainly looks good on the transformers...I'll give the pdf link a read (thanks so much).  Just got done taking a quick look at the transformers...they'll even fit fine in the 3"tall enclosure I've already got

   

its the DL36-2, a 72VA, dual 18V secondary tranny @$37ea. The slightly cheaper 36-2 lacks the dual prmaries which allow international use. Weighing in at 3.8lbs, its 1lb heavier than my dual bobbin EI cored 100VA unit, FWIW. The engineer described these trannys as 'conservatively specified'. It might be ideal for a Nirvana++ 25 implementation, but this is speculation on my part.

   Hugh, any thoughts on these?


   You had mentioned that a friend of yours was comparing various transformers in an amp.  Has he come to any early conclusions?  Would it be possible to ask if he can jot down some preliminary thoughts?


  thanks again all!

[AKSA]

Paul, thanks for all the work on 'rectifier' transformers.  

I'm no expert on any of this stuff, transformers are not my game, but in summary here are the requirements for a SS Class AB power amplifier:

1.  Low EM emissions for elimination of induced hum in the signal path.
[Favors toroidals at low idling currents where hum is an issue.]

2.  Low magnetostiction for reduced mechanical hum.
[Favors EIs and C Cores which stress their cores less and are more robustly made and varnished, thus leading to less mechanical force moving the laminations.]

3.  Low weight for reduced shipping and mounting costs/difficulties.
[Favors toroidals.]

4.  Low magnetic flux density at full rated power so as to avoid core saturation and flattening of the AC waveform, which leads to high order harmonics difficult to filter out.
[Favors EIs and C Cores since most toroids are build with flux densities within 20% of the 1.6T flux limit of standard magnetic iron.]

5.  High magnetic flux overload capacity so that large current pulses during cap charging do not distort the waveform.  
[Favors large dimension iron such as EIs and C Cores.]

6.  Relatively low self-inductance in secondary windings so that current pulses during cap charging do not suffer voltage drop.
[Favors toroids which run shy on copper.]

7.  Tolerance of DC on the mains.
[Favors EIs and C Cores which have large magnetic circuits.]

8.  Low emissions of current pulses back into the mains.  
[This favors toroids, since they are limited in current pulse delivery.  This is very important since reverse emissions are becoming crucial, with the EU in particular mandating low noise injection back into the mains for future transformer designs.  This is actually driving SMPS, regrettably.]

As you can see, the benefits are distributed.  However, my experience has been that the best compromise is an EI designed for conventional M4 laminations but running Grain Oriented Silicon Steel (GOSS), which has almost double the magnetic flux capacity, around 2.8T.

You generally find that 'rectifier' transformers (so named because they are designed to drive full wave rectifiers with large capacitive smoothing, just like an audio amplifier) are EIs, built to generous dimensions.

Using the above philosophy, I specified GOSS EIs for the GK1, and these have been extremely successful.  They are dimensioned for standard M4 iron, but built with GOSS laminations.  EM emissions are very low, current pulse delivery is very high, and even constant current delivery is almost four times nominal rating.   Size is small, and mounting is easier than a toroidal.

There may well be errors in some of this, since it's all top of the head stuff in the limited time I have available to me.  I don't like making categorical statements because someone always shows me up, but I think this is pretty accurate.  Any corrections, please, let's have 'em.

A very good idea I discussed just yesterday with Andy Redwood in a pleasant cafe here in Heidelberg was constructing a 300VA toroid on a 1000VA core.  This would be physically very much larger than the normal 300VA toroid, but most of the advantages and none of the disadvantages of toroids would be available to the designer.  However, I'd rather deal with EIs, largely because they are the sort my supplier specializes in and knows so well.

I should mention that Matt has not done his definitive transformer test between toroids and Double C cores, so unfortunately I can't give any comments just yet.....

Cheers,

Hugh

[Occam]

Findings from a bit more research -

2 possible candidates for those "Retro" inclined.

1 .Hammond 167L36, 117V primary (sorry, non-American consumers) 36VCT, 72VA, 9% regulation, which might make a dandy 25Watt monobloc tranny.

2. Hammond 167P60, 117V primary, 60VCT, 300VA, 5.8% regulation which might be a treat for a 100Watt AKSA monobloc.

These are dual bobbin trannys which have the advantage of lower bandwidth than either concentricly wound and shielded EI cores or toroids. But they also have higher EMI than either, not a good thing. But, the Hammond 167 series has the advantage of being enclosed transformers, i.e. they've got those funky end bells, which add a lovely retro look, but have the advantage of substantially minimizing EMI. These are big and heavy (for their VA) suckers, substantially heavier than a toroid, and actually larger than their non end belled 165 counterparts. This is needed as the end bells, while minimizing EMI, are not ventilated as well, which requires larger EI laminations to minimize temperature rise.
They're $27 & $48 each, respectively.

The least expensive place to souce them in small quatities is actually  the Partsconnexion -
http://www.partsconnexion.com/catalog/transformers.html
and click on' Filament (enclosed 117v)'

Edit - I've changed the regulation % reported to more conservative values which reflect the data I've recieved directly from Hammond....

[gonefishin]

Well...it looks like I've finally gone and done it     Hugh, you've got an e-mail waiting for ya



  I do have some more questions on a few things I just want to double check before ordering.

    You had mentioned that for a 25Watt AKSA55 I would need a transformer rated at

The transformers, of which there are two, one for each channel, should be rated at approximately three times the power rating of the amp. For a 25W AKSA, this would be 24V rails, with an AC secondary of 18-0-18V. Rating should be 80VA, a commonly available transformer.

   I was looking at one of the rectifier transformers that occam had recommended (and linked) earlier in this post.  The transformers are 18VCT and 72VA and not to be pushed above 75VA.  Here is a link to the transformer in pdf format.  Note that you need to scroll down until you see the 36-2 transformer.  The 36-4 transformer is 144VA.  I also asked belfuse if these transformers had GOSS laminations, and they did not reply back to this part of my question...Is this critical?  Do you suggest looking at other transformers (perhaps of other types) , or does this seem to be a good match for sonic bliss with the AKSA 25


  thanks,
  dan



   I gotta say...I'm a bit nervous about ordering a SS amp after being a tube guy     ugh...I suppose only time will tell.

[AKSA]

Hi Dan,

Thanks for our post.  You wrote:

I was looking at one of the rectifier transformers that occam had recommended (and linked) earlier in this post. The transformers are 18VCT and 72VA and not to be pushed above 75VA. Here is a link to the transformer in pdf format. Note that you need to scroll down until you see the 36-2 transformer. The 36-4 transformer is 144VA. I also asked belfuse if these transformers had GOSS laminations, and they did not reply back to this part of my question...Is this critical? Do you suggest looking at other transformers (perhaps of other types) , or does this seem to be a good match for sonic bliss with the AKSA 25

My thanks to Paul (Occam) for this information too, he's been wonderful with netsearching, and has a gimlet eye at component choice.

The DL 36-2 is fine for a 25W version of an AKSA 55W at 72VA, 36VCT, 2A.  It should yield a little over 24 volt rails.  As a rectifier transformer, it is designed specifically to drive huge capacitive loads through bridge rectifiers, so it is difficult to imagine a better choice as it will deliver short, sharp current pulses of high magnitude to keep the filter caps fully charged, and its regulation will be exceptional.  Just be sure to angle one transformer slightly, say 30 degrees, against the other so that their external magnetic fields do not link and reinforce, causing hum.

A rectifier transformer is not designed with Grain Oriented Silicon Steel (GOSS), just the normal M4 silicon steel, generously dimensioned.  I use GOSS in the custom transformers on the GK1 because it gives a much bigger magnetic flux density for the same size item, permitting high charge currents into rectifiers and cap loading.

I would suggest this will be an excellent transformer for the task, and cannot see any advantage in going to the double rated 144VA version.

Cheers,

Hugh

[gonefishin]

Quote:
Can the gain be changed some, while keeping the integrity and stability of the amp?  


Yes, within prescribed limits. You can drop gain from the present 38 to 26 without appreciable change to the sonics.

   Hi Hugh...a few more questions.   I know...I know.  You probably thought the questions were finally done  


    What parts (values and brand) would be needed to change the gain of the lower voltage AKSA55 from the designed 38db down to 26db?

   I am still certainly interested in maintaining the original designed 38db when I get an AKSA100...so I certainly want to build it as intended with all intended parts.

 But, In the mean time...I've got to match this mid/tweet amp with my woofer amp.  My woofer amp has 26db of gain.  I'd be interested in finding out what parts would need to be changed...and to what value (but still keeping the recommended brand also).  Once I find out the values, type and brands...I'll go ahead and order them so they'll be here when my amp arrives

    thanks,
   dan

[AKSA]

Dan,

For reasons of stability/phase margin, I would never suggest dropping the gain of any AKSA - even the lower powered variants - below about 26, which is 28dB.

To lower gain to 28dB, you merely increase the value of R8 from 2K2 to 3K3.  No other changes - at least, superficially.  YMMV, of course, and here's why....

The reason is like anything in the physical world.  There is a myriad of carefully balanced parameters, particularly on the N and N+ variants, to achieve the sonics delivered, and any change to this fine balance starts to skew the design choices, almost always to the cost of stability.  The AKSA is designed with maximum speed in mind, so it can skip nimbly across all musical genres.  Some are pretty fast-moving, as we all know - music is a tight race circuit!  This means the handling must be fast, and it won't take too much to make it skittish.....

This will not always be apparent to you in anything but a finely controlled AB, or traumatic outcome with a difficult load, but generally I'm very diffident about changing things like gain, lag compensation, phase lead, operating points, stage current, etc unless I know precisely and exactly the environment in which the amp is operating.

That said, the AKSA is more stable than many amps out there, and will even handle electrostatic speakers with some grace, though not at very high levels (which would anger Zeus anyway, and likely damage that precious membrane!).

I guess I'm trying to say this amp is a Ferrari, not a Chevy, and while I love both marques, I cannot pretend that the AKSA in its highly tweaked form is something easily messed with.  It's not;  it's quite racy!

Cheers,

Hugh

[gonefishin]

Hugh, thanks for being you.  Descriptive and to the point.  If I thought you would give your customers anything else...I wouldn't have asked.

   Your comments are appreciated.  There are certainly other avenues I can go to first for level matching...and I will.  I'll most likely leave the system much as it is now.  Which would mean I wouldn't have to change a thing inside the amplifier.

  thanks again,
  dan

[gonefishin]

Well, just thought I'd let people know that I'm joining the AKSA family .  I've got my AKSA 55 (25Watt version) Nirvana plus ordered with two 36-2 36VCT Rectifier transformers ordered from Signal Transformer.

   I'd just like to thank everyone who offered advice while I was making my decision.  

   thanks so much,
   dan

[AKSA]

Dan,

Thanks for your post, and your decision to join our club!

Now, this Signal Transformer worries me;  it should be 36VCT, that is, 18-0-18Vac.  Just check that's so.  This will give you around 25V rails with the slightly overvoltage mains we all use, and this is sufficient for around 25W nominal, actually closer to 30W peak.

This is a very good transformer from what I can make out.

Cheers,

Hugh

[gonefishin]

Hugh, thanks for double checkin' my post.  It is 36VCT.  I went ahead and changed it in the post above too.  Just to avoid confusion.

   thanks  just imagine, I may be a SS guy after all  

[gonefishin]

Well looky looky what I got in the mail yesterday!  


   Not only did I receive the SignalTransformers for my AKSA project...I also received (much earlier than expected) the ASpen Amplifier kit on the same day!

   I may just put together a temporary enclosure and get this thing up and running.  Of course, the 2 year old will take precedence to amp building  hey, ya gotta prioritize.

  can't wait to get goin'!

[gonefishin]

Hi Hugh,

   I've got a couple questions regarding transformer selection on the recently ordered AKSA100W N+.  This amp is going to power my woofers while the 25Watt version that I built will power the midrange tweeter.  I would like to stay with the similar rectifier transformer that I have in my low power AKSA55, but the place I got the transformer doesn't offer this selection.   The nearest they have for the AKSA100 is a 40V  80VCT  336VA.  

   How much will this drop the power in the AKSA100W?  Should I just go with a torrid?

  The transformer can be found scrolling near the bottom of the page.  Look for
DL-80-4
.

   Thanks,

         dan

[AKSA]

Dan,

Get the DL-68-4, NOT the 80-4, which is 40-0-40Vac, too much voltage.  The DL-68-4 is actually 34-0-34Vac, or 68CT, and about ideal for this application.  It will give you 4A continuous, much more peak because of the caps, and while it's a little shy at 275VA it's close enough to 300VA for government work......

Dan, check your email!  I sent you one just now....

Cheers,

Hugh

[gonefishin]

The DL-68-4 (34-0-34Vac)...ok.  I'll do that for the AKSA100N+


   What would you recommend for my brothers AKSA55n+?  Going with something like the dl56-4 or dl56-6?  Or would the 28-0-28 be pushing it with some ComEd over voltage?

  e-mail sent back to you...thanks
   
 dan

[andyr]

Hi Dan,

Oh dear - quite a few questions!! ... EIs and C Cores have a less stressed iron core which is less susceptible to saturation and thus clipping, and their frequency response is very poor over about 10KHz. ...

Sorry, Hugh,

I picked up this quote of yours and I couldn't understand what you were implying.  Earlier on in your reply to Dan you said you liked double-C cored transformers (whatever the F*$%  they are !!   ) yet your comment, above, that "their frequency response is very poor over about 10KHz" seems to imply that they are no good??  

Are you saying that because an amp has to produce frequencies over 10Khz, it's no good using a C-core (or double C-core) as the power transformer?

Or are you saying because it can't reproduce frequencies over 10Khz AND it's "less susceptible to saturation and thus clipping" it must be a good thing?

Can U elucidate?

Thanks,

Andy