HOME.


Winding Your Own AM I F transformers.

How Hard Can It Be?

Don't Answer That.

I'll bet you never thought you would see those words on Fun With Tubes. Frankly, neither did I. But circumstances have converged to provide the incentive so I am going to give it a try. And in the back of my mind Yoda is saying, "There is no try. Do, or Do not."

When I came to Western Kentucky University in 1968 as the most junior member of the physics department I was assigned to teach what none of the other faculty wanted to teach. Electrical measurements lab and the electronics for scientists course and lab. It was a case of "Please don't throw me in that briar patch". My degrees were in electrical engineering not physics so I felt right at home. Next to my lab room was a stock room that contained some items that must have dated back to the 1930s. One of the things was a device with a crank and a pointer and dial that increased one count with each turn of the crank. One of the older members of the department told me it was a coil winder. There was one part that moved back and forth with every turn. But after a few turns I realized that it wasn't quite every turn. There was a set of gears that cause the rotation rate to be just a little different. I tried winding a coil with it but it quickly turned into a tangled mess. I was told that the only man in the department who knew how it worked had retired a couple of years ago and moved out of the area. He had not left a forwarding address. Remember what year I said it was? The internet had not been invented yet and wouldn't be for another 20 years. I put it back on the shelf where I had found it. A couple of years later the stock room was moved while I was on summer break. The coil winder and many other items that today are considered by me and others to be precious antiques were gone. "Thrown into the dumpster" I was told. I did not have enough seniority to protest and anyway it wouldn't have done any good. It was too late.

Fast forward to late August of 2018. I've been retired for 17 years and the subject of coil winders has hardly crossed my mind since 1970. The sites I looked at where others described winding coils by hand it was in deed by hand. Pi winding, (I don't know why it's called that), was approximated by randomly moving the wire guiding hand back and forth. A member of the FWT list, Ed, screen name J Ed, began writing about winding coils for 455 kHz I F transformers using the ladies sewing bobbins wound with ordinary magnet wire. At about the same time I was contacted off the list by Bill Ward who had been doing it right with a coil winder and Litz wire. He sent me part of a coil winder and enough information that if I had possessed it in 1968 I could have successfully wound some coils and who knows where I would be today. But I didn't so I couldn't and I am where I am.

Both Ed and Bill have sent me samples of their work. Bill's look as if they were wound in a factory with litz wire and a smooth sided pi winding. Ed's look as if they were wound by hand because in deed they were. Bill's came with shield cans made from sheet copper. Ed's came with shield cans made from tin plated sheet steel. I don't want to turn this in to a contest between Ed and Bill but it seems inevitable that the reader will interpret it that way.

Here is a picture of the two I F transformers. Bill's on the left and Ed's on the right.

A necessary step is to measure the Q of the coils at the operating frequency of 455 kHz. That requires disconnecting the trimmer capacitor. The Q meter works by resonating the coil with its own internal capacitor. The external capacitor must be removed. The table below gives values of Q measured with the Boonton 260 Q meter at 455 kHz and values of inductance measured at 250 kHz. Also shown are Q values measured with a DER EE model DE 5000 digital LCR meter at a frequency of 100 kHz with the resonating capacitor connected.

Table 1. Values of Q and Inductance of Bill's I F transformer coils.
Transformer assembly is in open air as shown in photo above.
Instrument Coil Frequency
(kHz)
Q at
Frequency
L (mH) Resonating
Capacitor
Q Meter Plate 455 100 1.54 No
Q Meter Grid 455 90 1.52 No
DCLR Meter Plate 100 25.6 1.5511 Yes
DCLR Meter Grid 100 19.57 1.4087 Yes

Next I wanted to measure the inductance and Q of the coils inside the shield can. But Bill had presented me with a problem in this area. He sent me one transformer and two shield cans. One was soldered along the seam while the other was not soldered as shown in the photo below. When I took this picture I had already placed copper foil tape along the seam.

When it arrived on my bench the transformer was installed inside the unsoldered can and was jammed in so tightly I feared I would damage the transformer assembly while getting it out. I did eventually extract it by expanding the opening at the seam. There was no way the transformer would ever fit into the soldered shield.

My reason for wanting to measure inductance and Q with the coils inside the shields is because the shield affects a change in both of those parameters of a coil. The why is, that the shield is one big shorted turn surrounding the coils. You know what a shorted turn will do to a power transformer. It would have an analogous effect on an I F transformer if coupled too tightly to the coils. That's why I F transformers of this style of construction are always so large. The open seam in the shield would negate the "shorted turn" effect but it would also render the shield next to useless by permitting magnetic fields to enter and leave the shield cans. It was impossible to bring the two sides of the seam into electrical contact with the transformer inside. So I used copper foil tape on the outside and inside of the shield to electrically join it into a single continuous conductor. It was very difficult to disconnect the coils from the resonating capacitors and still make connections to the coils. Consequently the Q meter was not used to measure inductance and Q at 455 kHz with the transformer in the can.

Table 2. Values of Q and Inductance of Bill's I F transformer coils.
Transformer assembly is in the shield can.
Instrument Coil Frequency
(kHz)
Q at
Frequency
L (mH) Resonating
Capacitor
DCLR Meter Plate 100 19.45 1.4868 Yes
DCLR Meter Grid 100 17.50 1.3778 Yes

Now let's have a look at the frequency response of the transformer. Here is the diagram of the test circuit.

Figure 1 Test circuit for I F transformer.

For a verbal description click here.

The value of the resistor effects the Q of the plate (primary) coil. The first scope picture was taken with a value of 470 k ohms. The second with 100 k ohms.

For a verbal description click here.

Bill told me that he had read somewhere that Professionally designed I F transformers were made with a higher Q in the plate winding than in the grid winding. At first I was skeptical but after some thought I understood the reasoning behind it. In an operating radio the secondary, (grid winding) has no load except a capacitive one from the grid of the following tube and wiring capacitance. This becomes indistinguishable from the adjustable trimmer which makes up most of the capacitance connected in parallel with the coil. On the other hand the primary, (plate winding) is connected in the plate circuit of a pentode tube. The plate resistance is admittedly high, 250 k and up, mostly the latter, which will tend to lower the Q of the primary coil. Bill designed his transformer in this way and he did the right thing but he over did it a bit. As will be covered below the coils in his design are over coupled which leads to the double humped response. The asymmetry is caused by unequal values of Q. No amount of tuning would equalize the humps. In the picture below the Q values are still unequal but it doesn't matter because the two humps have merged into a single one.

For a verbal description click here.

To understand the cursors you probably need a little help. The Y1 cursor is set to the -3 dB point on the curve. The vertical range is set to 700 mV / div. The curve comes up 4 div from the center so the height of the curve is 700 mV/div x 4 div = 2.8 v. -3 dB is 2.8/sqrt(2) = 1.98 V. The Y1 cursor is actually set to 1.95 v but what's 3 hundreds of a volt between friends. The X1 and X2 cursors are set to the -3 dB points on the curve. X1, X2, and ΔX are read out in milliseconds so we must translate them to kHz. The time per division is set to 20 ms/div and the frequency scale is 10 kHz/div. Dividing 10 kHz/div by 20 ms/div causes div to cancel leaving 0.5 kHz/ms. So the ΔX reading of 42.4 ms translates to 21.2 kHz. That's wide enough for high fi sound but too wide for night time D exing.

Coupling, Over, Under, and Just Right.

When two resonant circuits that are tuned to the same frequency are coupled together the band pass they produce depends on the amount of coupling. If they are over coupled the response is a double humped curve as shown by the upper scope picture above. If under coupled the insertion loss goes up but the bandwidth gets narrower. Sometimes loss is tolerated to obtain narrow bandwidth. But when the coupling is just right…as shown by the lower scope picture…ah, that is just right. The amount of insertion loss is minimum and the bandwidth has a nice single humped response. It doesn't matter if the method of coupling is magnetic as is used in tube type I F transformers or capacitive as in the transformers shown on this page. For more details on why over coupling produces the double hump follow this link. At the end of that page you will be given a choice of 3 links. The "Winding Your Own IF Transformers" link will bring you right back here.

Adjusting the Coupling.

The equation for critical coupling is;

KCRIT = 1/sqrt(Q1 Q2);

Where Q1 is the Q of coil 1, Q2 is the Q of coil 2, and KCRIT is the critical value of the coupling factor K. The most obvious way to adjust coupling is to move the coils. But if glue prevents that there is still another way. That is to adjust the value of Q of 1 or both coils. In the top scope picture the resistor that connects from the output of the signal generator to the primary winding is 470 k ohms. In the bottom picture the resistor has been changed to 100 k ohms. As Q is decreased the value of K increases. The original value of K which was too high for the conditions of the top picture becomes just right for the conditions of the bottom picture.

Insertion Loss.

Figure 1 above shows how to measure the insertion loss. The trace that resulted is shown below.

For a verbal description click here.

The worst case insertion loss is -1.41 dB. That's not bad. The failure of the input voltage to show symmetry around the center frequency is due to the difference in Q between primary and secondary.

Ed's I F transformer.

Scramble wound instead of pi wound, magnet wire instead of litz wire, Ed's transformer seems to have everything going against it. At least that's what the theory says. Let's do some practical measurements and see what we get.

Table 3. Values of Q and Inductance of Ed's I F transformer coils.
Transformer assembly is in open air as shown in photo above.
Instrument Coil Frequency
(kHz)
Q at
Frequency
L (mH) Resonating
Capacitor
Q Meter 1 & 3 455 9 1.20 No
Q Meter 2 & 4 455 10 1.24 No
DCLR Meter 1&3 100 30.1 1.1931 Yes
DCLR Meter Grid 100 30.0 1.2301 Yes

Table 4. Values of Q and Inductance of Ed's I F transformer coils.
Transformer Assembly is in Shield Can.
Instrument Coil Frequency
(kHz)
Q at
Frequency
L (mH) Resonating
Capacitor
Q Meter 1 & 3 455 8 1.18 No
Q Meter 2 & 4 455 8 1.21 No
DCLR Meter 1&3 100 23.4 1.1668 Yes
DCLR Meter Grid 100 23.1 1.2020 Yes

For a verbal descriptionclick here

The cursors show a -3 dB bandwidth of 31.6 kHz. At this point I will remind you that Radio engineers use the -6 dB points for specifying bandwidth. In the usual AM broadcast radio there are two if transformers so -3 dB in one transformer adds up to 6 dB in two of them.

insertion Loss.

For a verbal description click here.

The measurement values near the bottom of the screen can be used for the calculation. The P-P (peak to peak) readout is the maximum value on the screen. From this we have Loss = 20 Log(0.318 V/1.46 V) = -13.24 dB. That's a lot.

Conclusions.

Although the hand wound transformer looked promising on the DCLR meter, when it was read by the Q meter at its operating frequency the results were worse than disappointing. The combination of scramble winding and magnet wire was too much against it and it didn't stand a chance.

Some may question how the DLCR meter could read essentially the same Q values at 100 kHz and the Q meter readings be so different at 455 kHz. As a matter of fact I wondered about this myself. To test this I measured the Q of one of Ed's transformers at a series of frequencies moving down to 100 kHz. As a reminder at 455 kHz the Q of the coil connected to terminals 1 and 3 read 8 on the Q meter. On the DCLR meter the Q read 23.4 at 100 kHz.
When I lowered the Q meter frequency to 250 kHz the reading was 13.5.
At 200 kHz the reading was 16,
at 150 kHz the Q was 19, and
at 100 kHz the Q meter read 24.
Nuf said?
Those of you who know how a Q meter works will know that I had to connect fixed capacitors to the C terminals of the Q meter to achieve resonance at these low frequencies. I used High Q silver mica capacitors.

On the other hand a properly wound coil using litz wire gave a good account of itself. I think the inductance was a little high. A few years ago one of the members of the Fun With Tubes list disassembled the IF transformers from an AA5 and measured the inductance. He found that for the interstage transformer the inductance was about 1 mH and for the output IF transformer, (not to be confused with the audio output transformer), the inductance was about 600 μH.(Note: The output IF transformer couples signal from the last IF amplifier to the detector.)

For those who don't want to go through all the difficulties involved with winding your own coils I recommend using commercially available RF chokes. Their Q is a little on the low side but is sufficient for AM broadcast radio applications. I have used them for many years and they provide selectivity that is good enough with little insertion loss. If you want to make an I F strip for SSB reception you may have to resort to winding your own coils.

So Have I Given up on Winding Coils?

No. I have posted the page so far because it seemed to be a good breakpoint. Additional adventures with coil winding will appear in the daily activities log as they occur and after another natural breakpoint is reached they will be moved here.

Stay Tuned.


HOME.