News:

"The phone is a remarkably complex, simple device,
and very rarely ever needs repairs, once you fix them." - Dan/Panther

Main Menu

Western Electric Ringer Impedance Measurements

Started by unbeldi, May 25, 2014, 01:20:00 PM

Previous topic - Next topic

unbeldi

The ringers made by Western Electric, as well as those of other manufacturers, are often characterized as having either low impedance or high impedance. Catalogs usually list the direct current (DC) resistance in ohms (Ω). However, ringers are operated by either alternating current (AC) or, in the early days, by rapidly pulsed DC current, which is an approximation to AC.

DC resistance and AC impedance are both measured in ohms. However, their numerical values can be quite different and the AC impedance depends on the frequency of the AC current.  In the limit of zero frequency, meaning when the frequency of AC approaches zero, the frequency dependence vanishes and resistance is equal to impedance.

So what does this impedance really look like for typical ringers of the Bell System?
Here are some measurements that I performed in 2013 of the impedance of typical ringers as a function of frequency.

Essentially, this entailed an application of Ohm's law and measuring both the AC voltage (in volts) across a ringer and simultaneously measuring the current (in amperes) flowing through the ringer.

E = Z x I
voltage = impedance multipled by current

Measuring this for a variety of AC frequencies gives us a graph of the frequency dependence.  It tells us what the resonance frequency is of the ringer. This is the frequency of best operation and shows up in the graphs as the point of lowest impedance, or largest current flow.

The performance of the ringer could be measured quantitatively with a microphone. Instead, I simply listened to the ringer and used an arbitrary scale of 1 to 10 to judge ringing loudness.

Each graph contains two curves: the blue curve is the measured impedance and corresponds to the vertical axis tick mark values in ohms.  The red curves are the subjective loudness at each point measured. The values 1 to 10 are scaled so that 1 scales to 1000, and 10 scales to 10000 on the vertical axis.  The horizontal axis (y-axis) is always the ringing current frequency, which was varied between 10 Hz and 60 Hz in several steps. So, for example, at a value of 20 Hz on the horizontal axis or x-axis, the 8A ringer has an impedance of about 4000 Ω on the y-axis.

The measurements were performed for three generations of WECo ringers:
  • A 8A ringer as can be found in a 534A subset of the 1920s used with the candlesticks, or the B1 or D1 sets later. It has a DC resistance of 2x 700 Ω, and is often referred to as "low-impedance" ringer.
  • A B2A ringer that is typical of the 304-type desk telephone. It performs electrically the same as the B1A ringer in a 302 and has about 4600 ohms DC resistance, and is therefore often called a "high-impedance" ringer.
  • A C4A ringer that is contained in the 500-series telephones. This is also a high-impedance ringer.

What we learn from these curves is that all three ringers operate optimally at the nominal ringing frequency of 20 Hz (cycles per second), that is standard for straight-line ringing. The blue curves (impedance) have a valley and the red curves (performance) are at their peaks.  In each case, the range of good operation extends a little further upward (to up to 30 Hz) as well as a little lower.

Especially the C4A ringer has good flat response from just under 20 Hz to a little over 30 Hz, which was necessary to operate them in the 1A2 key systems that had 30 Hz ringing supplies.

The ringer in the 534 subset shows a surprise, it also performed quite well at twice its principal frequency, a harmonic response.

The curves also justify the designation low and high impedance as used in the catalogs. At optimal response, the 8A ringer shows about 4000 Ω, the B-type shows 8000 Ω, and the C-type is just above the B-type.

Outside of the optimal response region, the impedance of all sets increases drastically. On the low end of the frequency range, i.e. below 20 toward zero hertz, this is partly due to the fact that the ringing capacitor represents an increasing resistance to the varying voltage and at 0 Hz, which is the same as direct current, it completely blocks the flow of DC current. If the circuit only contained the ringer itself, the impedance would converge toward the DC resistance value as stated in the catalogs.

The measurements were all performed with properly configured telephone sets for bridged ringing, including the ringing condenser (capacitor) in the circuit. In detail the sets were: a 1930 102-type/B1/534A combo, a 1947 304-type set, and a late 1950s 500D set.

PS: I added a summary graph of the three measurements of impedance on one graph. Blue is the 8A ringer of the 534A, black is the B2A ringer, and green is the C4A.

Phonesrfun

Interesting information.  I think I need to digest it for a bit.

-Bill G

unbeldi

#2
Perhaps it is worthwhile to mention how one can obtain a variable frequency ringing supply for these kind of measurements.

Ringing supplies are available from many telecom suppliers, such as Tellabs, of course, but often circuit diagrams are unavailable.

One of the cheapest and easiest methods I found was to modify a Panasonic KX-T61610 EasaPhone electronic switching system, the same that are so popular among us collectors for making rotary dial telephones work on modern lines.  These can be obtained on eBay for $20 to $50 with just a little patience. Instead of the 16-port 616, a 308 (8 stations, KX-T30810) can be modified as well. The advantage of using these units is the power supply is already integrated in the system to ring any connected telephone through any station port.

The ringing supply in these Panasonic systems consists of a signal amplifier and a ringing voltage transformer. The signal amplifier is driven by a 20 Hz sine wave signal that is synthesized by the system CPU on one output pin of the chip through a low pass filter circuit.

Having located this, one can cut a convenient wire jumper on the power supply board, the brown board above the transformers in the picture, which can be easily removed, and insert a single pole toggle switch to direct an external driver signal into the unit. The external signal may come from any variable oscillator, or signal generator, that outputs 15 to 70 Hz and about 4 to 6 V RMS signal level.

The little modular junction box on top of the PBX (see picture) provides a BNC input for the external signal, and a toggle switch to select either the external signal or the internal signal from the CPU.  It also provides an output of the ringing voltage on the modular jack for use in external circuits, to drive a manual PBX, for experiments, whatever.

Of course, this permits to ring any telephone with a frequency ringer through the PBX.  It is a one-stop solution to make rotary phones work, as well as ring old phones with frequency ringers.  Simply dial the right frequency on the signal generator to ring any phone.

All in all, a very elegant solution, I felt.

The first picture shows an internal view of a 616 version 2 unit, highlighting the wiring modification with red arrows. The second image is an exterior view of the added interface box with signal input (BNC), source toggle switch, and RJ11 ringing output on opposite side, not visible.  The details of this implementation would likely be a subject of a separate discussion.


==See also==
Frequency Ringer Designations Codes: http://www.classicrotaryphones.com/forum/index.php?topic=11938.msg126866#msg126866

unbeldi

#3
Here is an update to this topic. I made more measurements this year, and compared more ringers, which was a goal after my initial experiments in 2013/4.

A second goal was to find an answer to a question that came up in the first set of experiments, namely that the impedance curves for the C4A and the B-type ringer showed a curious looking bump around the 30 Hz frequency mark.

And finally I wanted to compare some of the specially marked B1A ringers that we sometimes find in telephone sets, namely the red-striped ringers and those that have a date stamp on them suffixed with an –I and –A, or prefixed with S–.   I didn't get to the S and A ringers this time. Actually, I am not sure I have an S–ringer, but I do have several –A types.

I decided to use a different ringing generator this time, and settled on the feature card and Option installed in my Sage 930A Communication Test Set.  The ringing option on the set is very convenient to operate; one can simply dial a value for the ringing frequency and for the voltage on the key pad.  Another advantage, I found, was that the voltage is better regulated with varying loads than it was on my home-modified key system.

I performed ringing current measurements at 5 Hz intervals from 15 Hz to 65 Hz.

The ringers chosen in this experiment were:
A) C4A ringer dated 1982 in a WECo 2500
B) B1A red stripe ringer dated 1943
C) B1A ringer stamped 12-49-I in a 302
D) Northern Electric B1A in a NE 684BX subset dated 1951
E) 8A ringer in 534A subscriber set of ca. 1930, the same 102-type telephone as measured previously, for direct comparison.

The ringing voltage was set on the test set to 96 VRMS, but varied slight in the low frequencies below 20 Hz, probably due to saturation of the ringing transformer.  In addition the wave form was distorted quite a bit at 15 Hz from the nice sine wave that was produced at higher frequencies.

The ringing current varies in all measurements between ca. 3 mA and ca. 22 mA, the latter being observed on the low-impedance 8A ringer, of course.

Here is a summary result of all measurements on one graph.  Plotted is the overall impedance of the telephone set while ringing versus ringing frequency on the abscissa.


TBC ... with some discussion of results.