Now... ya made me feel bad. I posted the questions not expecting you to email AE and ask for all that info. It was a rhetorical device... hoping to show the range of performance variables and notions that goes into a design. And calling into question the myopic focus on some very small subset of these expanded parameters.

I'd bee surprised if they answered all of them... and I wouldn't expect them to... they do have a right to treat much of that information as proprietary design information.

Interestingly, you felt comfortable enough to make the purchase decision without having this info. If I may ask... and not putting you on the spot... if this info were deemed to be vital... how did you make a positive purchasing decision? On what basis did you say... hey... I spend my bucks and get a choke from this company?

In the seventeen years of building tranneys... I have never had anyone ask me for the winding capacitance of any transformer we make. And we sell to a host of professional designers who have built scores of products over the years... and to many thousands of diy end users.

Personally I think this whole issue of winding capacitance and leakage L is a tempest in a teapot. And I tried to illustrate this by posting a two part example of how focusing on any one or two parameters could lead to mischief when I put up the examples of transformer A and transformer B.

What I have always found is... the best way to get service or information of this type is to contact the manufacturer privately and ask if this information might be available. Explain to them your need for the info and your use of the info. If they say that it is proprietary... then my sense is always to respect that.

Likewise... even if AE does provide you with some info... please be sure that it's ok with them to release it publicly... if they wish to treat this info as proprietary... that is well within their rights... and, I certainly don't want to do anything (or be demanding of information) that would vitiate that legitimate stance if that is how they would choose to respond.

Though...darn... I am curious as to why they only filled up about half of the lamination stack in those bobbins.

msl

The manufacturer of this core provides complete design data for mounting brackets for their units.

www.metglas.com/products/page5_1_6_2_5.htm

actually they offer two different plans... one of which is preferred as "minimum stress" on the core assembly and then an alternative design over the preferred method.

msl

www.metglas.com/products/page5_1_6_2_6.htm

msl

:::Direct measuring with capacitance-meter gives horrible ~700pF winding capacitance (not real, I hope:-), haha...:::

If the amount of winding capacitance concerns you here is an alternative design that you might want to consider.

EI 625 x 1/2" stack

12000 turns of #40 wire

calc self capacitance of approx. 68.6pf

calc L of approx. 1170 henries

calc L sub l of approx. 1.31 henries

calc flux density at 60vrms and 20 hz = 3101 gauss

dcr of approx 3250 ohms

At 100vrms and 20 hz the calc flux density would be 518 gauss.

The inductance listed above assumes use of M6 core material.
If you substitute 50% nickel this inductance figure will conservatively be increased to approx 1720 henries which is what your unit is listed at. If you substitute in 80% nickel core... then your L will be greater than the AE-Europe design.

But no matter which core material is chosen... the winding capacitance is approx only 10% of the number you have listed above. And the winding resistance has been cut by close to 60 percent. and the flux density is even at 100vrms and 20 hz below the published sarturation induction for 80 percent nickel.

And it would be easy to house this design in a channel frame which would provide a mounting mehtod straight from the manufacturer.

Disscussion:

What is intersting also in this design... is to take a look at what two "buzzwords" bought us... many folks go oow and awe at the prospect of a c-core and amorphous as a core material.

But... first let us look at this from the vantage point of reducing Cw. the AMCC 8 core has a window length of approx 1.08". The winding length of the EI 625 bobbin is only .856". All other things being equal the shorter the winding length the less eff capacitance you will get. The EI 625 wins out here.

Now... let's look at magnetic path length. The longer the path length the less efficient the core will be as an inductance producer all other things being equal. The AMCC 8 has a magnetic path length of approx 5.2" while the EI 625 has a magnetic path length of 3.75".

Next: Gross core area comparison. The AMCC 8 has a gross core area of approx .28 sq in. The EI 625 on a 1/2" stack has a gross core area of
.3125". This is as close to equaling the gross core area btwn the two candidates as possible if we stick to widely available bobbins.

Next: Net core area comparison. The AMCC 8 has a predicted net core area of approx .22 square inches. While the predicted net core area of the EI 625 by 1/2" core has a net area of .28 sq in. This is because the stacking factor of the AMCC 8 is only 79% while the stacking factor of the EI lamination is 90%. In other words you get more metal into the stack with the EI than you do the c-core made of amorphous strip. And this helps keep your flux density lower as well as increasing your inductance all other things being equal.

This is why I recommend that you always reject "buzzwords" as indisputable guides to gaining a notion of quality.

What this example also shows... is that, again, given a limited number of parameters to consider it is easy to "trump" any other design brought forth which has been designed to optimize a wider range of performance parameters.

Designing to one or two or three isolated variables is trivially easy.

msl

Maybee they have something up their sleeves (i.e., a good "trick") re: use of the bobbin with much of the core area unfilled. Like I said... it caught my eye only because after 17 years of building and being around transformers I've never seen this done before...

and... say it is something really trick... they may choose to come back and just say "we feel this is an advantage" and they may not want to offer any in depth explanation for their choice of designing on a physically larger bobbin than the core area itself requires.... if they do... again, in my opinion, we all must respect this decision if they consider their method or their design choices to be proprietary.

though by conventional theory... they have made the "plate area" of the capacitors much larger with that extra mean length of turn that is not required by the core assembly. So my guess is that using the recommended AMCC 8 bobbin and keeping everything else the same would result in less winding resistance (dcr) and less capacitance (less surface area btwn the windings)...

About mounting... your arrangement doesn't (on first blush) look like it would impart any unnecessary mechanical strain or stress on the core... when you house a unit this is one of the paramount considerations in choosing a mounting method. Your improvised mounting appears fine from this singular vantage point.

you mention "low price"... this depends on a very wide range of factors... volume being often one of the key factors... take the alternative design I offered above... have Hammond build it in sufficiently large volumes and I'd bet that on M6 they would be less than $15 each. And for your $15 you'd get a method of mounting your transformer. In small volumes I would guess that the price would be some two to three times higher than the guess I've made above.

Price also depends on a range of other factors... what materials are inside the transformer... what insulations (if any), what kind of magnet wire is being used, what kind of solder is being used to make internal connections... there are many other variables which could be juggled... to meet a low price point or a high performance level... it's the customers choice...

msl

that said... it's still a basic design... and could be tricked out much, much further though then we are likely to get into the territory of me not wanting to disclose any of our own proprietary design strategies, methods or goals.

but for a free, give away design... it might not be horrible.

msl

What tube are you using that has such a high Rout? I'm curious because I would like to run it through a program of mine and see what numbers I come up with.

14K output impedance is probably one of the worst cases.... take the same Cw numbers and check it out if you were using say a 6C45pi or a 5687 tube as just two examples.

Perhaps moreso in the case where you have an exceptionally high output impedance... you might need to concern yourself with Cw... but there are probably other factors in a trans design (i.e., a grid choke) that might be even moreso a limiting factor than the Cw.

All other things being equal.... your going to need much more L to support that output impedance of 14 or 15K... as a general rule of thumb... the more L you need the more turns you need and the larger the core size you might need (core area contributes to L)... and all of these will contribute to having greater Cw.... but what choice do you have... if you skimp on L to keep the Cw low.... then the L will come and bite you in the butt performance wise... so it is always a trade off of factors that goes into a design.... not just simply some mantra of "keep Cw low"...

As a bit of practical advice... correspond with and communicate with the magnetics provider of your choice... tell him your application... and ask him for his advice...

as an example.... if you asked me if your application would be a good mate with our CT choke... I'd say no.... and NOT because the Cw is too high (it isn't) but because our design does not have enough L to provide a good loadline for a 14 or 15K impedance. So even though I might not disclose what our Cw is... I would know enough to recommend to you that you not use the particular part that we make.

again... hopefully, this provides a richer and deeper context of circuit engineering than simply looking at just one variable and basing all notions of "goodness" on just a singular parameter.

msl

:::At 100vrms and 20 hz the calc flux density would be 518 gauss.:::

should have read 5168

msl