Author Topic: DIY Simple Output Transformer  (Read 44357 times)

Offline iondb

  • Trade Count: (+1)
  • AVForums Super Veteran
  • ****
  • Posts: 777
Re: DIY Simple Output Transformer
« Reply #15 on: March 16, 2010, 08:06:36 AM »
Same here.  Most (99%) of it is over my head,  but I find it interesting, especially how easy you make it look  :o  Thanks for the effort of posting all this for us mere mortals.

Offline Mars

  • Commercial Member
  • Trade Count: (+1)
  • *****
  • Posts: 1,019
  • “The path is the goal.” Mahatma Ghandi
    • Mars Tube Audio
Re: DIY Simple Output Transformer
« Reply #16 on: May 24, 2010, 10:44:21 PM »
So... onto the Mean Winding/Turn Length

The Mean Turn is half way from the base of the bobbin to the top. With a Bobbin Height of 14mm, the Mean Turn is 7mm from the bobbin base.

From the bobbin dimensions we have "A" which is an inner dimension. On the wire side/outer surface of the path length is twice the thickness of the bobbin material longer: so A+1.5+1.5=35.5mm. The Mean Turn runs 7mm above this surface, so the length of wire is 35.5+7+7=49.5

The same applies to dimension "B". The outer dimension of the " B" side is actually  32+1.5+1.5=35. The mean turn runs 7mm above this, thus 35+7+7=49.

This makes sense, since the core is square. There is a little bit more "play" on the "A" side so the laminations can slide into the bobbin.

So the Mean turn length is 49.5+49+49.5+49=197mm

The software wants a "cm" dimension, so we enter 19.7 cm into the table and click "save"

The data is now available as an option to select when doing OPT design
OEM manufacture of Tube Audio
Power and Integrated Amplifiers
Phono Pre-amplifiers
DIY Kits
DIY Workshops

Offline iondb

  • Trade Count: (+1)
  • AVForums Super Veteran
  • ****
  • Posts: 777
Re: DIY Simple Output Transformer
« Reply #17 on: May 25, 2010, 02:01:36 PM »
Thanks for keeping the thread alive and updating it Mr Mars.  But this rabbit hole is waaayy too deep for me head...... ;D

Offline Ampdog

  • Trade Count: (+1)
  • AVForums Grandmaster
  • *****
  • Posts: 7,063
Re: DIY Simple Output Transformer
« Reply #18 on: May 26, 2010, 02:13:53 AM »
Not to hi-jack Karel's thread, but how about someone stating his/her requirement, and we could then design a transformer for that. As said elsewhere I use a procedure using nomograms - the end result of the two processes should not be all that different. 
Audio must be the only branch of engineering where lack of basics' knowledge is considered a superior form of wisdom. (Anon)

Offline Mars

  • Commercial Member
  • Trade Count: (+1)
  • *****
  • Posts: 1,019
  • “The path is the goal.” Mahatma Ghandi
    • Mars Tube Audio
Re: DIY Simple Output Transformer
« Reply #19 on: June 28, 2010, 06:58:36 AM »
OK... so let's see what the software suggests we do. First we enter the data we have and click TAB

I leave Freq(Hz) setting at 30 (note this is the lowest frequency you want to pass at full power)
Watts=16.5 watts
Rp=18000
SE I unclick; this is a push-pull transformer
Primary is 5000 ohms
Current is 0.142 Ampere
Primary has 2 sections in serie and 1 in // (parallel)
Secondary is 6 ohms
and 1 section in serie and 2 in parallel
Wires // per section should be 1 (for now)

When I do that, I get this:


OEM manufacture of Tube Audio
Power and Integrated Amplifiers
Phono Pre-amplifiers
DIY Kits
DIY Workshops

Offline Mars

  • Commercial Member
  • Trade Count: (+1)
  • *****
  • Posts: 1,019
  • “The path is the goal.” Mahatma Ghandi
    • Mars Tube Audio
Re: DIY Simple Output Transformer
« Reply #20 on: June 28, 2010, 07:43:12 AM »
So... after entering the data on the light green section of the screen, I move to the gray block below, and search for my EI96 32/32 iron configuration.
Before you click, note that the Afe suggested should be 10.9 cm^2

This means the core of the iron going through the bobbin should have a section area if not less than 19.2 cm^2 to pass 30Hz at 16.5watts for these laminations.

When I click on the EI96 32/32 in my iron list, I get this:


Note that the iron option turns red, because the Afe of my choice is slightly (10.24 as opposed to 10.9) smaller than suggested by the software. Not a big difference; but we keep it in mind.

Next we go to the orange block on the top right. Here we enter the wire thicknesses etc. I look at the suggested copper thickness: 0.36mm diameter for the primary and 0.73mm for secondary.

Here you are going to have to look at what wire you actually have. I choose the next diameter up from the suggested I have in stock.

0.4mm for primary
0.75 for secondary

I get this screen after hitting TAB:


Note that the total Hcu is 18.8mm, a little bit more than the bobbin height (14mm) I have available. That is why the print is in red. We want to get to a configuration that fits on the bobbin, so we'll keep our eyes on this spot to check the viability of our transformer winding plan.

Also note that the suggested Amps/mm^2 is set at 2. This is a good figure for output transformers, but you can change to 3A/mm^2 (thinner copper wire) if things do not work out.

Have a look at the orange block at the computed data for the secondary (the blocks after the 0.75 figure).

It is suggested that you roll 91 windings on a bobbin that can handle a max of 54 per layer. So you need to roll two layers. I like to roll full layers, so I'll aim for a figure closer to 108 windings (two full layers) to fill the bobbin completely. This makes the transformer more efficient.

On the light blue block on the bottom right I note that the lowest frequency I pass is 20.6 HZ; a bit lower than the 30Hz I specified earlier. Go to the pink block top slider (Primary L) and adjust the position till you get a 30Hz reading on the light blue block on the bottom right.



Note that the height of the windings is still too much for the bobbin (16.7mm as opposed to 14mm). We'll have to find a way of shrinking our winding plan. This I do in my next post. Have a good day :D
OEM manufacture of Tube Audio
Power and Integrated Amplifiers
Phono Pre-amplifiers
DIY Kits
DIY Workshops

Offline Ampdog

  • Trade Count: (+1)
  • AVForums Grandmaster
  • *****
  • Posts: 7,063
Re: DIY Simple Output Transformer
« Reply #21 on: June 28, 2010, 06:53:02 PM »
Congratulations on your presentation, Karel!
Audio must be the only branch of engineering where lack of basics' knowledge is considered a superior form of wisdom. (Anon)

Offline Yves

  • Trade Count: (0)
  • AVForums Junior Member
  • *
  • Posts: 17
Re: DIY Simple Output Transformer
« Reply #22 on: June 30, 2010, 10:11:56 PM »
Hi Karel !

Just after mailing you, my password "bubbled up" in my brain  ;D
If not already done, check your mail box.  ;)

Yves.

Offline Mars

  • Commercial Member
  • Trade Count: (+1)
  • *****
  • Posts: 1,019
  • “The path is the goal.” Mahatma Ghandi
    • Mars Tube Audio
Re: DIY Simple Output Transformer
« Reply #23 on: June 30, 2010, 11:36:21 PM »
Yes Yves... I got your mail. Thanks for the suggestions. They have been useful to get to the results below

So here goes: Have a look at the previous screenshot bottom left hand block (grey colour) for the gap entry field. Change this to 0 (since a push-pull transformer does not have an air gap)

Now the numbers change significantly:



The inductance jumps up and the F Lo (-3dB) figure drops to 4.8 Hz

Now move to the Pri L slider at the top and reduce the inductance one click on the < at a time. Notice the F Lo (-3dB) starts climbing. Also note the calculated windings totals drops and the B total in the "Iron" block (grey bottom left) starts to climb. Ideally you want to stay below 1.1 Tesla (number in black print), but you can safely go up to 1.4 (number changes to red print after exceeding 1.1 Tesla).

While you do this, also note how the Total Hcu starts dropping and changes from red to black. Good... we now have a winding arrangement that fits on the bobbin.

Strictly speaking, we can reduce windings/reduce pri impedance till we get to our target F lo of 30Hz. But before we get to this, notice the point at which the secondary windings change from 2 layers to 1 layer (at 54 windings figure).

At this point we have a single layer secondary, filling the whole bobbin. An elegant configuration - easy to roll :-*

Experience has taught me, that this might still be a tight fit. So lower the impedance even more, till you get a 52 secondary winding count. This gives you bit of space on each side of the layer - it also guarantees that you get the exact calculated amount of windings onto the bobbin. An important aspect if you connect layers in parallel (as we do in this trannie).

So the proposed design ends up here:



Next step is to roll the proposed design and build it into an amplifier. This we leave for another day ;)
OEM manufacture of Tube Audio
Power and Integrated Amplifiers
Phono Pre-amplifiers
DIY Kits
DIY Workshops

Offline Ampdog

  • Trade Count: (+1)
  • AVForums Grandmaster
  • *****
  • Posts: 7,063
Re: DIY Simple Output Transformer
« Reply #24 on: July 01, 2010, 04:56:29 AM »
Hi Karel,

I hope I might add something. (I did not reread the whole thread so my apology if you did include this somewhere.)

The figure for primary inductance (as with the h.f. leakage inductance) figures in NFB stability conditions. (I did not get the impression that you are afraid of using NFB; you are capable of designing with that - good!) In that respect it might help to have a higher primary inductance than is required for frequency coverage, if you have calculated for minimum (low power) inductance - again I might owe an apology depending on what excitation you designed for.

[Brief explanation for the uninitiated: Inductance increases with excitation, so the low frequency pole moves downwards and that might upset NFB stability if calculated for inductance measured at the customary 5V 50Hz. If that pole was low enough (inductance high enough) to begin with, increase in inductance can only move it further down (and out of the way), improving l.f. stability. This is also one of the reasons for the high inductance (100 Hy) used by Willliamson.]

With this I am not criticising your design procedure - in fact I think it was a clearly written and neatly illustrated 'document' - it might be copied and kept. I am simply saying that it might be prudent to keep a high inductance despite the somewhat higher loss from more copper. (In my cases, using mostly C-cores with their higher µ's, that task is relatively easy.)
Audio must be the only branch of engineering where lack of basics' knowledge is considered a superior form of wisdom. (Anon)

Offline Yves

  • Trade Count: (0)
  • AVForums Junior Member
  • *
  • Posts: 17
Re: DIY Simple Output Transformer
« Reply #25 on: July 01, 2010, 07:56:02 AM »
. . .
 Ideally you want to stay below 1.1 Tesla (number in black print), but you can safely go up to 1.4 (number changes to red print after exceeding 1.1 Tesla).
Yes, 1.1 Tesla is allowable for standard laminations and 1.4 for GOSS ones.
Choose the right one double clicking in the list at the right of the grey block (Std or M6X) ;)

I second Ampdog about the primary inductance !

Also look at copper losses wich are below 7%.
Of course, nobody wants to build a loosy OPT, but you may safely accept up to 10%, that means you may reduce wire dia, at primary and/or at secondary and choose it to optimize the filling of the bobin (full-but not so-layers).
You will also note a reduction of the leakage inductance (good for Gnfb stability at hi end !)

Yves.

Offline Mars

  • Commercial Member
  • Trade Count: (+1)
  • *****
  • Posts: 1,019
  • “The path is the goal.” Mahatma Ghandi
    • Mars Tube Audio
Re: DIY Simple Output Transformer
« Reply #26 on: October 29, 2010, 02:00:09 PM »
Thanks for the replies Ampdog and Yves. To recap: We need a primary inductance that is higher than 100Henries and we can safely accept copper losses up 10%. So let's play a bit with the wire diameter options. A thinner wire will allow more windings (more primary inductance - good), but has more internal resistance and dissipates more heat (up to 10% is OK).

If we decrease the primary wire to 0.3mm and the secondary wire to 0.6mm things get better. Enter the suggested wire dimensions in the orange block in the "actual" fields and hit TAB. Now increase the Pri L on the top slider while keeping our eye on the total secondary windings. We aim for the point before the secondary winding total exceeds the amount allowed for a single layer. If you exceed this point, just decrease the Pri L slider till the actual turns per layer is the same as the max turns per layer. (Nothing wrong with double layers; I just like the idea of a neat single layer of secondary wire. A full layer is smoother on the profile than a bumpy partly filled layer)



As you can see, we do not reach the target of 100 Henries primary inductance. So let's decrease the secondary wire diameter even more to 0.5mm and slide the Pri L slider to the right till the secondary layer is full with one layer (just before it jumps to 2 layers and the Total Hcu jumps up a digit).



This is better. 146 Henries is inside the target. But our losses is still under 10%. We can increase the Pri L, because it allows us to lower the flux in the core (B total in grey block). Lower flux is more linear ( and therefore less distortion in the music)

So let's try 0.4mm secondary wire diameter. Enter the data, press TAB, slide the slider till layer is full and check the Cu losses:



This shows 9.8% Cu losses - within the limit of 10%! Pri L is 230 Henries - more than double the target of 100H - so we can expect to hear good bass in the music and no instability with large amounts of negative feedback. Any comments? Maybe we should roll this one?

OEM manufacture of Tube Audio
Power and Integrated Amplifiers
Phono Pre-amplifiers
DIY Kits
DIY Workshops

Offline Yves

  • Trade Count: (0)
  • AVForums Junior Member
  • *
  • Posts: 17
Re: DIY Simple Output Transformer
« Reply #27 on: November 08, 2010, 01:53:59 PM »
Hi Mars !
Many comments to come  :o

- The Ip0 is for EACH tube, not the sum of both.
- The current density in an OPT does not need to be so low (2Amp per square mm).
- 9.8 is the TOTAL copper loss in Watts, not the percentage wich is indicated in parenthesis on the same line  :'(
- Keep an eye on Leakage inductance  :-\

Somewhat in a hurry, I'll be back soon.

Yves.

Offline Ampdog

  • Trade Count: (+1)
  • AVForums Grandmaster
  • *****
  • Posts: 7,063
Re: DIY Simple Output Transformer
« Reply #28 on: November 09, 2010, 02:56:23 AM »
How nice to have CAD programs today for design! But not complaining.  I mentioned earlier (on another thread) that I am still using the Crowhurst graphs, as I have set out with them about 5 decades ago, and do not design OPTs that often. Karel, you then asked me to compare the two, which I could not then do without knowing more about your program.

As far as moving things up and down and noting the difference, as in 'zeroing in' on the final specs, the programme is undoubtedly convenient! For interest sake (now that you have a final design), I will do the same using the graphs - interesting to see how the results will compare. I must mention that if it does, a bit of my experience comes into it, in the sense that I corrected e.g. the metal µ (I have used the same core material over the years) to fit measurements. I sometimes got a practical primary inductance well over what was used in a design (which usually did not worry me as it was advantageous, as you stated). Leakage was usually spot-on - which is surprising, because it seriously depends on how tight the windings are; spacing between sections etc.

From me, congratulations on a method well-explained.     
Audio must be the only branch of engineering where lack of basics' knowledge is considered a superior form of wisdom. (Anon)

Offline Yves

  • Trade Count: (0)
  • AVForums Junior Member
  • *
  • Posts: 17
Re: DIY Simple Output Transformer
« Reply #29 on: November 09, 2010, 12:39:08 PM »
How nice to have CAD programs today for design! But not complaining.
. . .
Hi Ampdog !
Perhaps you already know that joke:

"Give one hour to a programmer to solve a problem and he will spend 59 minutes in writing the program that give the solution in one minute"

It's exactly what I did with this CAD program . . . except that it took me muuuuuuch more than one hour because I didn't know anything about transformers when I started (10 years ago) and that the programming itself was some form of "auto learning".

Must I add that the chapter 5 of RDH4 (co signed by Langford Smith and N.H.Crowhurst) is a real gold diamond mine  :)

That said, our friend Mars is still on the learning curve for both the goal to reach and the way to use the program to obtain it.

Quote
I mentioned earlier (on another thread) that I am still using the Crowhurst graphs, as I have set out with them about 5 decades ago, and do not design OPTs that often. Karel, you then asked me to compare the two, which I could not then do without knowing more about your program.

As far as moving things up and down and noting the difference, as in 'zeroing in' on the final specs, the programme is undoubtedly convenient! For interest sake (now that you have a final design), I will do the same using the graphs - interesting to see how the results will compare. I must mention that if it does, a bit of my experience comes into it, in the sense that I corrected e.g. the metal µ (I have used the same core material over the years) to fit measurements. I sometimes got a practical primary inductance well over what was used in a design (which usually did not worry me as it was advantageous, as you stated). Leakage was usually spot-on - which is surprising, because it seriously depends on how tight the windings are; spacing between sections etc.

Full agreement ! Many "empirical tweak constants" was added along years as the result of real mesurements !
. . . and I'm still "tweaking"  ;D

Mars,

try to load the latest version:
http://www.dissident-audio.com/AutoIndex/index.php?dir=OPT_da/&file=PST-and-OPT.tar.gz

This archive contains ALL the necessary files and need not to be installed, just extracted in a folder.
Neither your system files nor your registry wil be altered.
Works in a native Winxxx machine as well as under WINE.

Yves.
« Last Edit: November 09, 2010, 01:05:13 PM by Yves »