Using Modulation Transformers "Backwards"

[KE6DF]

I'm doing some planning for building a modulator around a transformer I posted about previously -- a UTC LS-67.

I have a question:

What are the issues/problems with using a modulation transformer backwards -- ie using the secondary as the primary and using the primary as the secondary?

This transformer is specified with a primary impedance of 9000 tapped at 6900.

There are two secondary windings, which, when connected in parallel give and impedance of 2500 and when connected in series give an impedance of 10000.

It would be slightly easier (I realize there isn't much difference) to select tubes to work with the 10K secondary side for the modulator, and the 9K primary side connect to the final amplifier.

I intend to use this transformer conservatively with a 1300 volt supply for both the primary and secondary.

I'll probably get out a carrier of about 150 watts or even slightly less.

The transformer is rated at 260 Watts in the UTC catalog, but I figure when working with a transformer that is older than I am, and is pretty much irreplaceable, it makes sense to be conservative.

I can always build a linear at some point to get more power if I need to.

I know that the UTC CVM series specs call for using those units either way depending on what impedances you want to match.

But what about the LS series? Am I giving up audio quality?

Thanks
Dave

[Steve - WB3HUZ]

I would imagine that the primary is set up for push-pull operation. If you use it as a secondary in series connection, it probably won't deal well with the unbalanced DC current and will saturate. If you use a blocking cap and mod reactor you'll be OK.

[KE6DF]

I was planning to use a blocking capacitor and a mod reactor in any case as this
 transformer does not have a gapped core and was not intended to have DC
 on the secondary.

[k4kyv]

The secondary apparently is  divided into two identical sections to support parallel or series connection,  for 2500/10K ohms load.  By using the series connection, the common point would serve as midtap, precisely the same as in the case of the multi-match mod xfmrs.  I don't see why connecting it backwards wouldn't work equally well.

That xfmr does not have a very useful range of impedance transformations with its intended connections.

As I recall, Timtron used one as a driver transformer in his 4-1000 rig, taking advantage of that midtapped secondary.
 
I would run a response check with a scope and see how well it works, and maybe connect it the other way to compare, to find out if the backwards connection degrades performance in any way.  There shouldn't be a tremendous difference between 9K to 10K, vs 10K to 9K.

[KE6DF]

That xfmr does not have a very useful range of impedance transformations with its intended connections.

It seems like the best set of ratios to use would be to reverse the transformer and use the 10K secondary on the modulator side, and then use the 6.9K Ohm primary taps on the final RF amp side.

That gives a ratio of 10 : 6.9 which is a lot more useful.

[WA1GFZ]

When using a secondary as a primary the coupling to the core is not as efficient because the new primary is now further from the core. It the case of an audio transformer it may help avoid saturation.  I would think the new primary Z will be a little lower.

[WBear2GCR]

The LS series, Linear Standard, was the best that UTC made.

It was intended to be very high-fidelity and was typically spec'd flat at full power from 20/30Hz to 20kHz.
Should work fine.

Forwards, backwards, backwards forwards... shouldn't matter.

Afaik, the windings are very interleaved, which is how they get the performance. Probably won't matter which way it is run at all, as the windings are not the usual single wind on top of single wind - afaik.

Plus you'll be running it at >3dB down from its max power rating...

                _-_-bear

               _-_-bear

[WA1GFZ]

Really, I wonder how they get the voltage isolation. Sounds like a cool idea. I need to find a dead one and take it apart. I've never done audio transformers.

[k4kyv]

I once had a very large Thordarson mod xfmr designed to go in a ship-to-shore radiotelephone transformer.  Got it new in the box, along with a blueprint of its physical design.  The primary and secondary windings were split into several sections and interleaved.  Also, each layer of the windings was split down the middle so that the two sides were symmetrical mirror images of each other.

Class B driver transformers and high quality AF output transformers usually have interleaved windings, sometimes divided into up to a dozen or more sections.  El cheapo AC-DC broadcast radio output transformers and "amateur radio quality" audio transformers are not built this way.  Not critical when all you are running is space shuttle quality audio.

The sections do have to be carefully insulated from each other, which would make the windings take up more space.  The main reason for the interleaving is to increase the coupling between windings and thus reduce the leakage inductance.  This is very important when the internal resistance of the source impedance must be kept as low as possible, as in the case of a transformer that drives a wide range high fidelity loudspeaker or class B or AB2 amplifier.

I have a collection of UTC LS series transformers.  They are very good, but still not quite what they are cracked up to be.  I find phase shift distortion and dips and valleys in the response curve in nearly every transformer once you get above about 11 kHz, even though the specs may say flat to within a dB or so well beyond 20 kHz.  I have found this in new old stock virgin transformers that I unpacked from the factory sealed box. Some of the older ones begin to go spastic above about 5 kHz.  This must be a fault in the design, because I can't conceive of what could go wrong with a transformer to alter its frequency response, except maybe shorted turns, and that would affect its performance throughout its range, not just at the higher frequencies.

[WA1GFZ]

Don,
We have an old mac 200W amp with a pair of 8005s. We use it for inducing low frequency noise on power lines to try and mess up power supplies. We couple with a transformer.  Last week I did a sweep and it was flat out to almost 80 KHz.
So interleaving is in sections not every other layer. I can imagine that was hard to get voltage isolation.

[k4kyv]

I never heard of one being interleaved every other layer.  As far as I know, is always done in sections.


Layers split in the middle and assembled on core as separate coils,
interleaved something like this:

      1/2 winding              1/2 winding

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  S
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  S
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

--------------------------------------
--------------------------------------  Core
--------------------------------------

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  S
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  S
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx  P
xxxxxxxxxxxxxxxxxx   xxxxxxxxxxxxxxxxxx

[Steve - WB3HUZ]

Many McIntosh amps used bifilar and trifilar windings on the primaries and interleaving of primary and secondary windings. Some of those windings were connected to the cathodes of the output tubes. These were used for various forms of feedback. McIntosh claimed response well past 100 kHz for many of their amps.

[KE6DF]

I was prompted by another thread on modulation reactors to make a couple measurement on my modulation transformers. Someone on that thread pointed out that the secondary of the modulation transformer is in parallel with the modulation reactor as far as the audio signal is concerned. So, you need to consider the inductance of the modulation transformer's secondary in parallel with the modulation reactor when figuring low freq response.

So I measured the following:

Inductance of 1/2 of the secondary of a CVM-3 = 5.7H
Inductance of the full secondary of the CVM-3 = 22.6H

The reason these values are not additive (like resistors) is because coupling between the two halves of the secondary creates mutual inductance.

I tried the same thing on the LS-67

Inductance of 1/2 of the secondary of the LS-67 = 16.2H
Inductance of the full secondary of the LS-67 = >50H

My inductance meter only goes up to 50H, so I couldn't measure the inductance of the full secondary of the LS-67. Figuring about the same level of coupling between halves of the secondary on both transformers, and therefore proportional inductance, the full secondary inductance of the LS-67 should be somewhere north of 60H.

Conclusion: While a modulation reactor works well to get DC off the secondary with a communications quality gapped core transformer like the CVM unit, there is a point of diminishing returns in making the modulation reactor really large because of the relatively low inductance of the modulation transformer secondary.

Another interesting comparison:

Weight:

CVM-3, 125Watt,  15 lbs
CVM-4, 300Watt,  22 lbs

LS-67, 260Watt,  37 lbs

So the LS unit is a lot heavier for it's power level -- probably it has a bigger core, more highly segmented windings (hence more insulation layers), and probably more turns of wire.

[WBear2GCR]

One of the reasons that the McIntosh output transformer works as well as it does is that it is a special case. All of them use a winding on the plate and the cathode. A quick thought about having equal resistances on the plate and cathode brings to mind the classic phase splitter circuit - unity gain. The output stage of the MacIntosh has no voltage gain to speak of. This arrangement permits the primary of the transformer, each winding to have a much much lower impedance than the usual impedance required for a given tube, like 1/4 iirc.In the 200watt version with the 8005 tubes the driver has to be bootstrapped to swing the required monkey to drive the outputs. 

(I think I have a jpeg or pdf of that amp, if you don't have it already)

There is a very good explanation of the McIntosh output transformer in the Radiotron Designer's Handbook. There is a version of it available for download, iirc on Pete Millet's site, FYI.

I works like a champ because the very parameters that limit typical transformer HF response are effectively (magically) reduced in value with this method. A good friend of mine had some 100watt class audio outputs wound up for him that we tested (15 yrs back now - geez) and they spec'd fabulously, with response out to 100khz for them. Most of the McIntosh amps like the Mc30 and Mc60 are "only" good out to 80kHz. Contrast that with the typical output iron that barely makes 20kHz, and you can start to see why that method is so good. We were doing this with a transformer that was rated at 100watts, very very conservatively and most typical output iron in that class has so much - whatchmacallit - is it leakage inductance and interwinding capacitance - that it barely makes it up to 20kHz.

I have been told by an an audio "enthusiast" that has dissected and wound new McIntosh outputs that they are typically scramble wound (although the wire is bifilar or trifilar)!!

I always thought that when I heard the LS series as outputs in tube amps that they sounded "funny", but never having owned any, I never got to test them... the other series of UTC transformers, like the PVM and LS-845 (a PVM type made for the 845 tubes) are spectacularly not particularly good. I presume they are very similar to the CVM series...

           WBear2GCR

[Steve - WB3HUZ]

And to think, McIntosh came up with this beautiful approach in 1947!

One of the reasons that the McIntosh output transformer works as well as it does is that it is a special case. All of them use a winding on the plate and the cathode. A quick thought about having equal resistances on the plate and cathode brings to mind the classic phase splitter circuit - unity gain. The output stage of the MacIntosh has no voltage gain to speak of. This arrangement permits the primary of the transformer, each winding to have a much much lower impedance than the usual impedance required for a given tube, like 1/4 iirc.In the 200watt version with the 8005 tubes the driver has to be bootstrapped to swing the required monkey to drive the outputs. 

(I think I have a jpeg or pdf of that amp, if you don't have it already)

There is a very good explanation of the McIntosh output transformer in the Radiotron Designer's Handbook. There is a version of it available for download, iirc on Pete Millet's site, FYI.

I works like a champ because the very parameters that limit typical transformer HF response are effectively (magically) reduced in value with this method. A good friend of mine had some 100watt class audio outputs wound up for him that we tested (15 yrs back now - geez) and they spec'd fabulously, with response out to 100khz for them. Most of the McIntosh amps like the Mc30 and Mc60 are "only" good out to 80kHz. Contrast that with the typical output iron that barely makes 20kHz, and you can start to see why that method is so good. We were doing this with a transformer that was rated at 100watts, very very conservatively and most typical output iron in that class has so much - whatchmacallit - is it leakage inductance and interwinding capacitance - that it barely makes it up to 20kHz.

I have been told by an an audio "enthusiast" that has dissected and wound new McIntosh outputs that they are typically scramble wound (although the wire is bifilar or trifilar)!!

I always thought that when I heard the LS series as outputs in tube amps that they sounded "funny", but never having owned any, I never got to test them... the other series of UTC transformers, like the PVM and LS-845 (a PVM type made for the 845 tubes) are spectacularly not particularly good. I presume they are very similar to the CVM series...

           WBear2GCR

[k4kyv]

I use a UTC LS-49 class-B driver transformer, to match a quad of 2A3's to class-B 810 grids.  It works excellently in the transmitter, but if I really wanted to transmit true hi-fi audio it would be mediocre at best.  The response curve begins to show dips and peaks above about 12 or 13 kHz, even though the manufacturer's specs rate it flat to above 20 kHz.

When I first got the transformer n.o.s. in factory sealed box, I tested it by feeding a signal from a generator across the primary, and connecting a scope to the secondary so that 1/2 the secondary winding, from midtap to one end fed the horizontal plates, and the other half fed the vertical plates.  From the low end of the audio spectrum up to about 11kHz, the scope pattern showed a diagonal  straight line, as it should.  But above that point, the straight line began to separate into a narrow ellipse, which opened up more widely with increased frequency, indicating phase shift error at the high end of the audible spectrum.  I normally use a low pass filter to cut off everything above about 7 kHz, so the phase shift is of no consequence in my transmitter, and I'm  sure it would still pass an FCC broadcast audio proof of performance, but it would fall short of audiophile quality for a studio or theatre sound system.

Most of the LS transformers in my collection fall similarly short at the high end of the spectrum, and some of the older ones from pre-WW2 begin to show a degradation in performance above about 5 kHz, even though the catalogues from that era claimed flat response at least to 15 kHz for all the LS series transformers.

[KD6VXI]

One of the reasons that the McIntosh output transformer works as well as it does is that it is a special case. All of them use a winding on the plate and the cathode. A quick thought about having equal resistances on the plate and cathode brings to mind the classic phase splitter circuit - unity gain. The output stage of the MacIntosh has no voltage gain to speak of. This arrangement permits the primary of the transformer, each winding to have a much much lower impedance than the usual impedance required for a given tube, like 1/4 iirc.In the 200watt version with the 8005 tubes the driver has to be bootstrapped to swing the required monkey to drive the outputs. 

(I think I have a jpeg or pdf of that amp, if you don't have it already)

There is a very good explanation of the McIntosh output transformer in the Radiotron Designer's Handbook. There is a version of it available for download, iirc on Pete Millet's site, FYI.

I works like a champ because the very parameters that limit typical transformer HF response are effectively (magically) reduced in value with this method. A good friend of mine had some 100watt class audio outputs wound up for him that we tested (15 yrs back now - geez) and they spec'd fabulously, with response out to 100khz for them. Most of the McIntosh amps like the Mc30 and Mc60 are "only" good out to 80kHz. Contrast that with the typical output iron that barely makes 20kHz, and you can start to see why that method is so good. We were doing this with a transformer that was rated at 100watts, very very conservatively and most typical output iron in that class has so much - whatchmacallit - is it leakage inductance and interwinding capacitance - that it barely makes it up to 20kHz.
           WBear2GCR

Years ago I had a pair of MC500 tubed amps.  Won them from an idiot.

I had NO idea the passband of them was so high, but that sure does explain the harmonic rich sound they gave out....  I used a fairly simple set of speakers, but replaced the tweeters with ones that "spec'ed" out to around 60 or more Khz, IIRC.

Magic amplifiers, the McInTrash are.  Loved mine, and a mono-bridged MC500 will REALLY crank some power out to a speaker.

Then it was off to Klipsch Corner Horns.  They only 'required' 100 watts of exciter.  And at that level, they required a waiver to be signed before they would be dropped off.  Most efficient set of speakers I've EVER heard (and to be fair, the largest cabinets, as well).

--Shane

[WBear2GCR]

Steve,

It is a great design from 1947... but if you look back at the time there were many, and I mean many competing variations upon that same theme. They tried almost every conceivable variation of connection for the output tubes.

Most were not terribly sucessful, and few were commercial products.

One that was was the Quad (England) amp that used partially cathode coupled primary windings.

I think the thing that kept more people away from attempting the McIntosh design, I'll bet someone considered it, was the fact that it is unity gain, and that means trouble for the driver circuit for a high power version, and more expense no matter what.

I have a nice model 50w, the original amp that McIntosh made... or one of the original, and it used an IT which they abandoned later... and the 200w version with the 8005s required that bootstrapped driver circuit to get enough swing. None of it made for a "cheap" amplifier.

But the performance of the OT probably makes it worth the extra bux, and certainly made their reputation. Otoh, the rest of the circuitry left a bit to be desired by today's standards...there is a definite "McIntosh sound" because of that - not the output iron.

An MC500 is likely not a tube amp...would have to look it up... probably one of their solid state amps with an autotransformer at the output... different beast. Interesting idea though.

                  _-_-bear

[KD6VXI]

Steve,

It is a great design from 1947... but if you look back at the time there were many, and I mean many competing variations upon that same theme. They tried almost every conceivable variation of connection for the output tubes.

Most were not terribly sucessful, and few were commercial products.

One that was was the Quad (England) amp that used partially cathode coupled primary windings.

I think the thing that kept more people away from attempting the McIntosh design, I'll bet someone considered it, was the fact that it is unity gain, and that means trouble for the driver circuit for a high power version, and more expense no matter what.

I have a nice model 50w, the original amp that McIntosh made... or one of the original, and it used an IT which they abandoned later... and the 200w version with the 8005s required that bootstrapped driver circuit to get enough swing. None of it made for a "cheap" amplifier.

But the performance of the OT probably makes it worth the extra bux, and certainly made their reputation. Otoh, the rest of the circuitry left a bit to be desired by today's standards...there is a definite "McIntosh sound" because of that - not the output iron.

An MC500 is likely not a tube amp...would have to look it up... probably one of their solid state amps with an autotransformer at the output... different beast. Interesting idea though.

                  _-_-bear

Oh, an extra 0... What's an 0h between friends

You're right, though.  The 50 was what I had.  Still cranked out the power bridged.

I used to have some mono-bridging adapters from my car stereo days, they would allow you to bridge nearly any amplifier (within reason).

Really, just an inverter for one channel.

--Shane

[KE6DF]

I use a UTC LS-49 class-B driver transformer, to match a quad of 2A3's to class-B 810 grids.  It works excellently in the transmitter, but if I really wanted to transmit true hi-fi audio it would be mediocre at best.  The response curve begins to show dips and peaks above about 12 or 13 kHz, even though the manufacturer's specs rate it flat to above 20 kHz.

When I first got the transformer n.o.s. in factory sealed box, I tested it by feeding a signal from a generator across the primary, and connecting a scope to the secondary so that 1/2 the secondary winding, from midtap to one end fed the horizontal plates, and the other half fed the vertical plates.  From the low end of the audio spectrum up to about 11kHz, the scope pattern showed a diagonal  straight line, as it should.  But above that point, the straight line began to separate into a narrow ellipse, which opened up more widely with increased frequency, indicating phase shift error at the high end of the audible spectrum.  I normally use a low pass filter to cut off everything above about 7 kHz, so the phase shift is of no consequence in my transmitter, and I'm  sure it would still pass an FCC broadcast audio proof of performance, but it would fall short of audiophile quality for a studio or theatre sound system.

Most of the LS transformers in my collection fall similarly short at the high end of the spectrum, and some of the older ones from pre-WW2 begin to show a degradation in performance above about 5 kHz, even though the catalogues from that era claimed flat response at least to 15 kHz for all the LS series transformers.

Better not tell this to the audio types on ebay -- the ones who pay $300 - $400 or more for LS output transformers. We wouldn't want the old transformer market to crash the way the stock market has.

Every once in a while a Thordarson Tru-Fidelity, Chicago, or Kenyon K or P series broadcast quality transformer comes up and usually they go for much less that the LS series equivalent. Probably they are just as good (or bad), but I guess UTC dominated the market back in the tube days, and they still seem to dominate the mind share of the audio hobbyists.

[k4kyv]

If anyone has an LS-51 (or its equivalent) they are not using, I have plenty of other LS and similar quality types I would be willing to swap for a good one.

FWIW, I once picked up a n.o.s. Chicago broadcast quality p-p plates to p-p grids interstage transformer and found that it suffered similar deficiencies in frequency response to the UTC's.

[Gito]

  I'm a new comer here,reading this article,a question cross my mind, would you used this Transformer for modulation  Transformer in a transmitter,or using for an output for audio amplifier like home sound system ,when you used it for a modulator Transformer the fidelity  needs not be to high (audio frequency response) ,because there' s restriction for the bandwidth of your transmitter, if the audio frequency is to high the bandwidth of your transmitter will be to wide.
And the radio we have has limited bandwidth  ,especially tellecomunication radio.And it's again the rules.
Looking at your transmitter .it has 9 k ohm and 10 k ohm,in my opinion the turns ratio of the winding of the transformer is the most important thing ,for instance you have 16 ohm Primary (impedance) and 4 0hm (secondary },the turns ratio is th root of (16 : 4) =2 , so the turn ratio of your transformer is the root of (10 : 9) = 1.05 ,So when you know the turns ,you can you used an audio amplifier with has 6 k ohm impedance Plate to plate as a modulator and used a tube for the RF Final 1.05 x1.05 x 6 k ohm = 6.6 k ohm,You can get by loading the RF final ,for instance you can load 813 with 1300 volt and load it to 200 ma, the impedance you get 6.5 k ohm,So you can used your Transformer. 
I think you still will have a good quality audio/sound,


 Gito.N

[WBear2GCR]

FYI, here in the USA we do not have a bandwidth restriction.
Generally, for practical reasons most AM stations limit their effective audio to 5-6kHz.

You didn't state your call sign, but I am guessing ur in europe, maybe Italy. Or maybe ur in Bayonne NJ... ??

                  _-_-WBear2GCR

[Gito]

 
 Bear

I lived in Indonesia,I'm sorry I did not know ,that there are no restriction in America,I have no call sign .actually I made small Medium wave broadcast transmitters (400 to 600 watt output),by looking/serving the Internet I have found this web,I'm very interested in this forum.and gain a lot inputs from here.
In my country ,there are such restriction,Maybe because there  is a time when all the  radio channel which used for amateur radio were to "crowded " .

Thanks

Gito.N

[WBear2GCR]

Welcome to the Amfone forum, Gito.

Perhaps if it is not too difficult to get a license in Indonesia you should try?

Many other areas of the world are more restrictive of amateur radio than we are in the USA. I can speculate as to why that is, but I suspect that it has nothing whatsoever to do with "crowding" of the bands.

No matter, glad you can learn things here, and hope that you will become an Amateur Radio Operator soon.

           _-_-bear