Inexpensive toroidal transformer works as a modulation transformer!

[WBear2GCR]

Yeah Stu!

FB!

I think I'd prefer a resistive load for spec testing, but the ultimate test is still in the rig.

The only caveat I have is that it is unlikely - not impossible, just unlikely - that larger toroidal power transformers will exhibit such good results.

          _-_-bear


PS. I forgot to mention that there is a company called Plitron that makes toroidal output transformers for hi-fi. These would be very good run as a reverse mod iron, using a Heising type connection only. If you read the fine print on their specs you find that the DC offset permitted is extremely low - many hi-fi hobbyists seem to miss this little bit of fine print that they try to hide! (ha!)

[WA1GFZ]

A tape core will work fine at almost 10 times the flux density. Do the math anything else needs a lot more turns or core area. Also the right tape core will work out to around 20 KHz. Ferrite cores start getting useful above around 40 KHz.
E-I core does the best job of moving heat out of the core area.
Hypersyl is easier to build because you just have to strap the two halves rather than building up the E-I stack while you worry about hypot issues. Round cores are the hardest to make because you have to transfer the wire to a machine to wind it. A round core concentrates the heat under the wire but is also best at concentrating the field inside the winding. The metal of the core can be the same in all configurations but I have never seen E-I core with 3 or 10 mil thick slices.
 

[KD5MHQ]

Ok, my turn...
Been reading this thread with rapt attention...way cool! I'm gonna purchase 1 of these torroids for my 60W AM rig. I've been really interested in torroid xfrmrs for power and have only imagined them for modulation...now I'm convinced. Thanx for the R&D and for posting it here.

Bob, KD5MHQ

[AB2EZ]

Bob et. al.,

Please remember:

1. No DC on any of the windings (must use a Heising configuration)

2. Select the turns ratio to be close to what you need (it doesn't have to be exact, because your audio amplifier should be chosen to have more than enough audio power to modulate the transmitter).

For example, suppose you are driving a 5000 ohm transmitter load from an audio amplifier whose output is optimized for use with an 8 ohm load (i.e., 8 ohms gives the maximum power out of the amplifier). If you use a 19.2:1  transformer (230 volts => 12 volts), like I used, then the load seen by the amplifier will be 5000 / (19.2 x 19.2) ~ 13.6 ohms.

This is fine... but the amplifier may not deliver its peak rated output power into a 13.6 ohm load. So it is good to have an amplifier whose rated output power is somewhat higher than 50% of the input power of the r.f. amplifier you are planning to modulate.

3. The "secondary" of the transformer should be rated to deliver the peak voltage you need to modulate the transmitter. For example, the transformer I am using has a 230 volt (rms) rating for its "secondary". This corresponds to 230 x 1.414 ~ 325 volts peak. The Ranger operates at around 550 volts B+ when I run it at full power. Thus, it would have been better if the transformer were rated for a higher voltage than 230 volts (rms). Fortunately, the transformer seems to work fine, when I run the Ranger at full power. Part of the reason for this is because frequencies higher than 50Hz (the frequency at which the transformer is rated) produce proportionately less saturation of the core (which is the limiting factor with respect to the 230 volt (rms) rating at 50 Hz), and therefore the transformer can produce higher voltages than 230 volts (rms) provided I don't try to produce those voltages with a very low frequency sine wave.

Of note: I plan to try this approach in a scaled up experiment. I ordered a custom transformer from Antek with the following ratings:

1300 volt : 57.5 volts (57.5 volts is 1/2 of 115 volts, so John will have to cut the standard 115 volt primaries in half)

Turns ratio: 1300/57.5 ~ 22.6

Core VA rating: 800 VA

Maximum rated output voltage at 50 Hz: 1300 volts (rms) x 1.414 ~ 1838 volts peak

Application: Modified Heising modulation of my existing homebrew legal limit amplifier (which uses a Russian GS-35b triode), re-biased to operate in Class C, with the B+ reset to 1500 volts. Power output at carrier: 375 watts (1500 volts x ~400 mA on the input). Audio power required for 100% modulation (delivered to the r.f. amplifier): ~ 300 watts.

The modulation impedance of the r.f. amplifier will be 1500 volts / 400 ma = 3750 ohms. The load impedance seen by the audio amplifier will be 3750 / (22.6 x 22.6) ~ 7.3 ohms. I plan to use an audio amplifier that is rated to drive loads between 4 ohms and 16 ohms... with a peak output power capability of 400 watts (hopefully, borrowed from somewhere).

Best regards
Stu
 

[WBear2GCR]

I look forward to your experiment Stu.

But I'll be very surprised if you manage to get any decent frequency response from the larger transformer. Sure, if you wrap feedback around it, you'll force a response - but that's not "good practice" imho, since that solution is never satisfactory. But, I guess we'll see.

I sent an email to that fellow via his website, but never got a response!
Perhaps he only responds to ebay inquiries??

                          _-_-bear

[AB2EZ]

Bear

Here is a "thought experiment" for you (although I might actually try this on a smaller scale). The purpose of this thought experiment is to demonstrate how a single large transformer can provide 16x the audio power, with the same performance, as a single transformer.

I have already demonstrated that one of the small Antek 25 VA, 230VAC (rms) : 12VAC (rms) toroidal transformers does a great job as a modulation transformer for my Johnson Ranger (i.e. flat frequency response between 30Hz and 10kHz, linearity excellent down to 30Hz, 100% modulation of the Ranger when it has up to 550-600 volts on the plate of the 6146).

Now, consider the arrangement... shown on the attached file... of sixteen (16) of those same transformers in a 4 x 4 array ( 16 x $8.00 per transformer = $128.00). This is, in principle, a quick and dirty way of doing the scaled up experiment... without having to wait for a custom transformer.

Each column (top to bottom) has four (4) transformers in series. The primaries are in series, and the secondaries are in series. So, each column represents a transformer with 48 volts in and 920 volts out. The turns ratio is still 19.2:1, just as in a single transformer (230/12 = 920/48 = 19.2). Each column, therefore, still represents an impedance transformation of 19.2 x 19.2 = 367 (just as for a single transformer).

Next, consider the equivalent circuit of a single column (see the attachment). The parallel capacitance of a single column is 1/4 of the parallel capacitance of a single transformer. [Four identical capacitors in series => 1/4 the capacitance of a single capacitor]  The magnetizing inductance of a single column is 4x the magnetizing inductance of a single transformer.

Next consider 4 columns in parallel (a total of 16 transformers). The turns ratio of four columns (or any number of columns) in parallel is the same as for a single column. Likewise, the impedance transformation of four columns in parallel is the same as for a single column.

The equivalent circuit of four columns in parallel has 4x the parallel capacitance of a single column, and 1/4 the magnetizing inductance of a single column.

Therefore, the complete array of 16 transformers has exactly the same parallel capacitance (1/4 x 4 = 1) and exactly the same magnetizing inductance (4 x 1/4 =1) as a single transformer.

The maximum voltage that the 4 x 4 array can produce at its secondary (which is limited by core saturation effects) is 4 times the voltage that a single transformer can produce (because a column consists of four transformers in series)

The maximum current that the array can deliver (limited by a combination of several effects) is 4 times the current that a single transformer can deliver (because there are four columns in parallel to share the total current).

Thus the array of 16 transformers has the same characteristics (turns ratio, frequency response, equivalent parallel capacitance across the output winding, equivalent magnetizing inductance across the output winding) as a single transformer... except the maximum power it can deliver will be 16 times as large
(4x the voltage and 4x the current) as a single transformer.

One large transformer (which doesn't cost all that much less than 16 small transformers) will perform at least as well as the array. It will weigh the same as the array, and it will be similar in physical volume (actually a little smaller) as the array.

Stu

[WA1GFZ]

Stu,
Check out the flat matrix transformer. The guy who held the patent Ed Herbert was my boss years ago. I think you can still find data on line.

[WBear2GCR]

Stu,

If your assumptions were correct for this situation, then there would have been no need to design Modulation Transformers differently than power transformers, eh?

Since there is/was a need to design transformers intended for wideband audio using different methods than power transformers, something must be missing from your analysis, one might surmise?

                         _-_-bear

[Steve - WB3HUZ]

wideband audio

This may be the flaw in your analysis. It seems you are thinking in terms of hi-fi audio transformers with high-end response out to 50k. Stu is looking to get response to 5 or 10k, all that's really needed to amateur radio voice work.


Just a thought....

[Tom WA3KLR]

Bear,

I think you are trying to (unknowingly?) equate the old iron-laminated transformer technology to the new powered iron (or ferrite) toroidal transformer Stu is using.  The toroidal core material must have a higher frequency response than the iron laminations, no surprise that this could be, I would think.  And we are comparing incidental use of power transformers as audio transformers.

I'd have to dig into my books to see if there is anything else governing the top end limit, but the core material, leakage inductance, and winding distributed capacitance are the main 3 factors.  I think all of these are inherently better with the winding you do on a toroidal core.  The E-I core really packs the wires together.

The toroid has more surface area for the windings so that all can be close to the core and spread out - wonderful.

[WA1GFZ]

Better core material does have higher frequency response but it takes a lot more wire to get the same reactance so the leakage inductance monster will get you.
A good tape core will give you high permability like lron and good frequency response.

[AB2EZ]

Frank

I very much appreciate your inputs and comments. I am however puzzled by your last comment.

"Better core material does have higher frequency response but it takes a lot more wire to get the same reactance so the leakage inductance monster will get you."

Here's why I am puzzled:

A. The ferrite core material has a much higher permeability than any of the iron core materials. This is the most important difference between ferrite core transformers and iron alloy (e.g., hypersil) core transformers (tape wound or otherwise).

B. The much higher permeability of the ferrite core material means (by definition) that the B field is much larger in the ferrite core (vs. an iron alloy core), for a given H field. Therefore, to get the same output voltage (# of secondary turns x area of 1 turn x dB/dt) you can use fewer turns on the primary (i.e., less H field for the same primary current) and fewer turns on the secondary (even with less H field, the permeability of the material is so high that you still have a larger B field).

Example:

If the permeability of the ferrite core material is 81 x the permeability of a given iron alloy core material... then you can have 1/9th as many turns on the primary and 1/9th as many turns on the secondary (i.e., the same turns ratio). I.e.,

1/9th as many turns on the primary => 1/9th the H field (for the same current in the primary winding)

A permeability of 81 x that of iron alloy => the B field in the ferrite core = 1/9 (H field) x 81(permeability) =  9 x the B field in the iron alloy core (even though you have 1/9th the H field)

9x the B field => you only need 1/9th as many turns on the secondary to get the same output voltage. It also implies (of course) that the voltage on the input winding is the same (i.e. 1/9 x 81 x 1/9 = 1).

C. Because the permeability of the ferrite material is much higher than the permeability of iron allow core materials... you have less leakage inductance (not more)... because the (cross sectional area) fraction of the total B field that is in the core (where the permeability is very high) will be much even greater (vs an iron core) than the (cross sectional area) fraction of the total B field that is in the air outside of the core.

Leakage = Cross sectional area of the H field that is not inside the core x 1 (i.e., the permeability of air) / [(Cross sectional area of the H field in the core x the permeability of the core) + Cross sectional area of the H field that is not inside the core x 1]


D. The only downside, which I have pointed out in each of my posts in this thread, is that the ferrite transformer cannot accommodate any significant amount of unbalanced dc in any of its windings (because the high permeability ferrite core material will saturate)

Best regards

Stu

[AB2EZ]

Mack

Thanks for the kind words....

All:

Check out this commercial audio output transformer (rated at 70 watts). It is an audiophile transformer... with extremely tight frequency response specifications... and, at around $400.00, I don't think it addresses our needs. However, looking at the diagrams, it doesn't appear to be any more sophisticated than the filament transformers that Antek sells for a small fraction of that price.

http://www.plitron.com/audio_4004.asp

Note: see reference 5 in the notes shown under the specification table

Best regards
Stu

[AB2EZ]

Here is a free version of the complete paper:

http://www.next-tube.com/articles/Veen/VeenEN.pdf

You might want to skip to Section 5 of the paper to get closer to the punch line (section 5.6). For those who are interested in the theory, the entire paper makes a good tutorial review of the issues. I skimmed it rapidly... but I plan to go back and read it more carefully in the future.

Stu

[WA1GFZ]

Stu,
What permeability values are you using for iron core. I'm thinking hi silicon steel is up around 12,000 but have not looked at specs in years. I think 3CB, F etc are much lower. Also I think 3C8 and F can't handle the flux density of grain oriented   high silicon steel. I could be wrong

[AB2EZ]

Frank

I've been searching on-line, and I also looked in:

Lee, Reuben "Electronic Transformers and Circuits", 2nd Edition, John Wiley and Sons, Inc., 1955

[The entire book is available for download from

http://www.pmillett.com/technical_books_online.htm ]

On line, I find values of relative permeability (permeability / the permeability of vacuum) of around 4000 for "transformer iron" in:

http://en.wikipedia.org/wiki/Permeability_(electromagnetism)

i.e. The actual permeability of "transformer iron" = 5 x 10**-3 Newtons / Ampere squared.  The actual permeability of free space is 1.26 x 10**-6 Newtons / Ampere squared. The ratio is ~4000.

In Lee's book (e.g. on page 30, Fig. 27; in Section 15. Core Materials) I find "grain-oriented" steel shown to have a relative permeability that varies from 2000 (at a Flux Density of 20 Gauss) to a maximum of 35,000 (at a Flux Density of 10,000 Gauss). This figure also shows a relative permeability for "silicon" steel that varies from about 1000 (at a Flux Density of 20 Gauss) to a maximum of 10,000 (at a Flux Density of 3000 Gauss).

[See the copy of Figure 27, attached below]

It could be that the strong dependence of the permeability of the above materials on the Flux Density is a source of non-linearity in transformers made from these materials. This would lead one (perhaps) to design audio transformers from these materials to work at low Flux Densities (where the permeability doesn't vary as much with the Flux Density)... which also correspond to low permeabilities. [E.g., by including an air "gap" in the core]. On the other hand, power transformers made from these materials could be designed to work at much higher Flux Densities (and, therefore, higher permeabilities) because non-linearities would be much less critical for power transformers.

In Chapter 5 (starting on page 140) of Lee's book: "Amplifier Transformers", he talks about Harmonic Distortion (section 67 page 153). On page 167, in Table XIII, he shows how harmonic distortion caused by saturation of the magnetizing inductance of an audio transformer increases rapidly as the Flux Density rises above 500 Gauss in a transformer employing a silicon-steel core. Note, from Figure 27 (below) the relative permeability of silicon steel is about 6000 at a Flux Density of 500 Gauss

Meanwhile, the Ferrite transformers probably provide a more constant (and also very high) permeability over the range of Flux Densities at which they are operated.

[WA1GFZ]

Very interesting. I seem to remember we did 60 hz. power transformers 12,000 to 14,000 Gauss. E,I core is pretty unstable and maybe that is why it was never used for switchers. Tape cores were used for low frequency switchers and I remember thinner tapes gave the best performance. So the gap must even things out and allow the DC offset. All this stuff is a real art.

[WBear2GCR]

Bear,

I think you are trying to (unknowingly?) equate the old iron-laminated transformer technology to the new powered iron (or ferrite) toroidal transformer Stu is using.  The toroidal core material must have a higher frequency response than the iron laminations, no surprise that this could be, I would think.  And we are comparing incidental use of power transformers as audio transformers.

I'd have to dig into my books to see if there is anything else governing the top end limit, but the core material, leakage inductance, and winding distributed capacitance are the main 3 factors.  I think all of these are inherently better with the winding you do on a toroidal core.  The E-I core really packs the wires together.

The toroid has more surface area for the windings so that all can be close to the core and spread out - wonderful.

Tom,

I am unaware of anything other than tape wound cores being used in power iron. If there is some new core material being used for this, I'd like to read about it online, got an URL?

I wrote an email to Antek asking them about the core material, and no answer came back. Maybe I will try asking them via their ebay address...

------------------

Pliotron does make toroidal "audio output transformers" but if you read the fine print they can handle nil DC offset, making the idea of using them for PP service a joke! (even though people do...). Obviously without DC on the core, it's a different matter.

For a good overview of the issues in transformers, always refer to the "bible" - Radiotron Designer's Handbook.

                _-_-bear

[WA1GFZ]

magnetics makes tape cores. They have been around for many years.

[AB2EZ]

Bear

You wrote:

"Pliotron does make toroidal "audio output transformers" but if you read the fine print they can handle nil DC offset, making the idea of using them for PP service a joke! (even though people do...). Obviously without DC on the core, it's a different matter."

I believe the requirement is for nil unbalanced DC.

I agree that this is still a challenge, but in push-pull, the average (a.k.a. DC) currents going to each of the tubes flow in opposite directions in each half of the associated winding. While, traditionally, one might not worry too much about the balance of the average current flowing in each half of the push-pull circuit, with modern technology it would be relatively (compared to decades ago) easy to implement a simple feedback circuit to balance out the average plate currents in each tube (e.g., using Hall-effect devices). One could adjust the grid bias of one tube relative to the other to achieve this.

As an aside, I wouldn't mind having a way to separately measure the plate currents of the 810 modulator tubes in my KW-1... but that hasn't reached the top of my "to do" list.

You also wrote:

"I wrote an email to Antek asking them about the core material, and no answer came back. Maybe I will try asking them via their ebay address..."

I think that John, of Antek, being a small entrepreneur, is focused on delivering off-the-shelf products to folks like us, and on answering questions from prospective customers who are likely to place large orders. He sometimes answers my E-mails (asking to place an order for a custom transformer) and sometimes I get no response. I don't mind, because my orders are of little significance to his bottom line.

There is some information on Ferrite materials here. (See, for example, materials: W and H):

http://www.bytemark.com/products/ferrmat.htm

There is also a tutorial on the use of ferrite materials in transformers here. (See, in particular, the discussion of the use of ferrite materials in wideband transformers):

http://users.catchnet.com.au/~rjandusimports/tut_2a.html

Also, please refer to the table at this url:

http://users.catchnet.com.au/~rjandusimports/ft_mat_1.html

Note that the difference between the initial permeability (low values of Flux Density) and the maximum permeability (high values of Flux density) is fairly modest in ferrite material H... compared to the variation in the permeability vs. Flux Density in the grain-oriented steel and nickel steel curves I posted as a JPEG attachment in my prior post. This confirms that ferrite core materials can be used over a wide range of Flux Densities, where their permeabilities are very high, without the transformer suffering from the nonlinearities that might result from the use of (for example) grain-oriented steel or nickel steel (when used at Flux Densities where their permeabilities are high).

Best regards
Stu

[w3jn]

BUMP for WB4BFS

[N4LTA]

Several months ago I asked a toriod supplier about making a modulation transformer and was told  that they could not do it due to DC current. I didn't discuss modified Heising.

I am just starting to test a 60 watt transmitter using a Hammond 123JSE as a traditional modulation transformer. It is physically large and I suspect it would do 100 watts in a modified heising arrangement.

The problem that I have run up against with the modified Heising,  is the chokes are harder to find than modulation transformers. Hammond has 10h at 500 ma but 4 of these is large and heavy and not cheap. I sell Hammond products and it is still expensive even at dealer cost. (If you need some Hammond stuff at a discount - let me know)

I may try 4 of them on my 4-400 rig that is under construction.

Pat
N4LTA

[WQ9E]

Pat,

I had planned to use the Hammond chokes in series for a project last year but I came up with a traditional choke first.  But please let us know how it works out as the Hammond units are nice, readily available, and not that expensive. 

It is nice to know that there is a Hammond dealer on board!

Rodger WQ9E

[KK4RF]

Stu,
     Could you put your experomental AM rig on the air this weekend? How about checking into the Old Military Radio net this weekend on 3885 early Saturday morning? I would love to hear what it sounds like.
                       ---Marty, KK4RF---

[WA1GFZ]

Stu,
A better answer for your question. Looking at the tape itself. The thinner the tape the higher the frequency response. I have seen tapes as thin as 3 mils. This reduces eddy current losses and will blow the doors off an EI transformer core for audio response. Many years ago I worked on early switchers and we could make a very nice 15 KHz square wave using tape cores. I bet a tape core will make a good audio transformer but not cheap to wind. Magnetics Co. has some good application notes that go back 30 years on selecting the right tape core. I build a CD ignition that would throw a 1/2 inch spark. It had a 300 watt inverter wound on a tape core. It wiped out a stock coil in a couple weeks.