News:

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

Main Menu

Setting timing on a dial - A discussion of Analytical Instruments and Audacity

Started by RDub, October 28, 2016, 04:47:58 PM

Previous topic - Next topic

unbeldi

Quote from: Jim S. on February 06, 2017, 06:29:29 PM
I think the BK 1045's use the same dialing  tolerance as the phone company . If that is the case, then I don't think the tolerance is tight enough for IP.

Just a guess tho,
Jim S.


The specification of the 1045B shown earlier speaks against the first of your statements.  Also my experience dispels that notion.
The specs state, in other words, that a break ration of 30% to 85% is accepted as long as the total dial pulse cycle lasts at least 100 ms, and that the dial speed can be 8 to 20 PPS.  The lower value of the PPS is perhaps not low enough for many old telephone dials, because I have seen many between 7 and 8 PPS, but I don't recall whether I had problems on my 1045A.

But, the same specs appear to contradict themselves, because if the unit accepts 20 PPS signaling, then the pulse cycle can only be 50 ms long, violating the "Recognition Time (end of digit)" of 100 ms minimum.  I am sorry, that is not correct. I misinterpreted that statement.  That specification is the time the unit needs to evaluated the preceding pulse sequence as a complete digit, not the pulse cycle time.

dsk

Quote from: unbeldi on February 06, 2017, 06:48:15 PM

The specification of the 1045B shown earlier speaks against the first of your statements. 
You are right, and I do not know whats right, telephones collecting is not a religion, it is lots of science, but also experience of whats good enough or not.
I have 2 reasons for adjusting my own telephone dials:
1) Learning by doing.
2) Reapair it if it is not working well enough.

When I first have to do something I want it to be so accurate as reasonable. That gives a speed of 11 +/- 1 pps.
The make break ratio is something i usually do not need to adjust at all, but again if I have to I will get as close to 60% percent break as I can without risk of breaking anything, but I am satisfied up to 67% (UK std) since it never has made problems for me. I even have an old 1920ies dial never adjusted and it works at 50% No need for adjusting as long as my electromagnetical PAX understands it , so do the Dialgizmo and the Mitel smart 1. 
What I know is 15 pps is to fast for my PAX.
I have read here on the forum about problems using rotary phones on some ATA's, and I have an Ericsson exchange not accepted by the  Dialgizmo, but by the POTS line. (I should have used more time to figure out why)

I am so happy with this forum, because I can learn something day by day, and discuss such problems here.   ;)

TelePlay

Just discovered another use for Audacity. I have a Kellogg stick transmitter and wanted to test it for working condition. I attached the two alligator clips normally placed on the dial pulse contacts on the transmitter terminals, started to record in Audacity and was able to record my voice and listen to the quality of the sound on playback.

Not a precise measurement but if one were to use a constant volume sound source placed, say, one inch from the transmitter, the amplitude of the recorded sound file could be measured for a group of several transmitters, a plot drawn for weak vs strong transmitters and after that, a new transmitter could be checked not only to see if it was working but also the strength of the transmitter.
Yesterday eats you up, it eats everyone up . . .

TelePlay

     Regular Member Post

Continuing this topic, a couple of weeks ago, Ktownphoneco (Jeff Lamb) asked if I could run the data on an Audacity audio file (wave) containing the pulses from a suspect dial. The wave file he sent contained 6 full cycles (dialed 0 six times). He was in between Sage 930s at the time and needed to know if the dial on his bench was within spec.

I loaded the wave file and used the 2nd cycle, the second time he dialed 0. Using my Excel spreadsheet, found the dial was slow at 8.34 PPS and the Break/Make ratio was off at 69.64/30.36. When Jeff's improved Sage 930 showed up, he ran the dial to find the Sage had the dial at 8 pps and a Break/Make ratio of 70/30. Was nice to see the Audacity procedure was equivalent to the capabilities of a Sage 930.

When I found out both ways of testing a dial were similar, I again loaded the wave file into Audacity and analyzed all 6 of the cycles in the wave file. The purpose is to see the difference, or variation in a dial from cycle to cycle. Being a mechanical device, the gear action, governor, dirt and lubrication (or lack of) cause the dial to be slightly different in each cycle but by having the data from 6 cycles, the PPS and Break/Make ratio were found to be quite consistent, basically the same.

Having all this information and a greatly improved, over time, spread sheet, I am posting this to help anyone in the future who wants to use Audacity to check a dial's performance starting with the wave file for once cycle as seen in Audacity and how/where to read data from that screen to do the analysis.

The first image below is what is seen on a computer screen expanding the horizontal wave form to cover just one 10 digit cycle and pulling the bottom of the wave form bar downward to get a full screen view of the cycle.

The yellow circle in the upper left shows the start of a dial that will send 10 pulses, dial 0. The yellow circle below that shows the end of the first cycle, one pulse sent. The vertical change between this is the point at which the dial closed the pulse dials to go from a break to the make status. The lower yellow circle is also the first part of the next pulse, the opening of the contacts to produce the second digit "break." This continues for all 10 digits.

In the center of the wave form there is a vertical yellow line, this is the cursor as it looks within the wave form. The top bar shows the time in seconds and the position of the cursor is shown below in a box to the 1/1000th of a second. Placing the cursor exactly on a pulse spike will turn the pulse spike yellow and that exact point in time is shown in the box at the bottom. This is the number to be used in the Excel spreadsheet to analyze or test the dial.

Since there is no "break" after the last pulse, the "make" for the last digit dialed, the wave form trails off and this happens on every dial. There is no need to estimate or calculate the last, missing spike in that the time from the first "break" spike to "break" spike of the 9th pulse can be used and the time between those points divided by 9 to get the PPS rate of the dial. The information for the last digit is not needed for a PPS measurement.

A dial in spec will have that last "break" spike (the end of the 9th "make") exactly 0.9 seconds from the first "break" spike. Simply divide 9 (digits) by 0.9 seconds (the time it took to create the 9 full pulses) to get the speed of the dial in PPS, pulses per second. 9 divided by 0.9 seconds is 10 or the dial is running at 10 pulses per second (10 PPS).

A dial taking MORE than 0.9 seconds between the first "break" spike and the beginning of the 10th digit "break" spike will be running slow, less than 10 PPS. 

A dial taking LESS than 0.9 seconds between the first "break" spike and the beginning of the 10th digit "break" spike will be running fast, more than 10 PPS.

=============================

The data from the 2nd of the 6 cycles was plugged into my spreadsheet, second image below.

The green box in the upper left shows the dial speed for that cycle. 8.35 PPS, too slow.

The red box just below that shows the Break/Make ratio for the first pulse, 69.03%/30.07% with the spec normal at 60/40 ratio.

Since there is no 11th Break, the 10th Make trails off and the length of the 10th Make cycle can not be read. This last value was calculated from by adding the length of the first 9 Make times, finding the standard deviation of those values, dividing the standard deviation in half and adding it to the length of the 9th Make cycle length, the yellow box. In this case, 0.001 second was added to 0.039 seconds to estimate the 10th Make cycle length to be 0.040 seconds. This value is used to determine the 10 Break/Make cycle.

The graph below the yellow box show the time of each pulse. This plot always slopes up and that shows the dial spring winding down, less tension so slower speed. The slope itself is an indication of how dirty the dial is. A clean dial will have a rather flat slope and a dirty dial upward and even curved upward during the last couple of digits. It's an interesting graph when the dial is so dirty that it does not return to its stop or rest position by itself. In a perfect world, this graph would be flat at 0.1 seconds per pulse or 1.0 seconds for 10 pulses, the spec speed or 10 PPS. The dial speed of the first pulse was 0.115 seconds and the final pulse took 0.132 seconds showing the slowing of the dial about 15% over 10 digits.

The tall red box on the right is used to determine the Break/Make ratio of each pulse to get an average Break/Make ratio for the dial. In this 2nd cycle of 6, the average Break/Make ratio was 69.18%/30.82%.

The above data was obtained for each of the 6 cycles available for this dial. The 3rd image below shows the speed graph for each cycle. There is some variation for each pulse in each cycle but they are basically the same. The points above or below the calculated slope line can be a small piece of dirt getting in the gears for one pulse in any one cycle. That's why those outliers are not consistent from cycle to cycle. The slopes of each 6 graphs are consistent.

The last image below shows the data gathered from each of the 6 cycles. The top line is PPS and it is quite consistent from cycle to cycle with the average being 8.37 PPS +/- 0.04 PPS. The Break/Make ratio for the first pulse and the average of all 10 pulses for each of the 6 cycles is also quite consistent 69.3%/30.7%.

Again, the Sage 930 stated the dial had a PPS speed of 8 PPS and a Break/Make ratio of 70/30. Assuming the Sage 930 rounds, 8.37 PPS would round to 8 PPS and 69.3/30.7 would round to 69/31.

All this work was done to prove to myself that Audacity can be used to quite accurately determine both the dial speed and its Break/Make ratio using just one cycle and the first pulse Break/Make ratio (no need to do all ten pulses or multiple cycles to get an average - one cycle of 10 digits recorded by Audacity and plugged into a spread sheet will provide the same information as a Sage 930, but for just these two specs).

=============================

And once the dial is cleaned and adjusted, Audacity or a Sage 930 can be used to see if the work brought the dial into spec or if further cleaning/adjustment is needed.
Yesterday eats you up, it eats everyone up . . .

HarrySmith

Harry Smith
ATCA 4434
TCI

"There is no try,
there is only
do or do not"

dsk

A good analysis of the picture, and an excellent explanation of the process!
I have done it simpler, and got pretty close numbers.

I played a little with the picture to get closer readings, and found the values written under the boxes marking out break no 1,2,9 and 10.

Those numbers are putted in to my online spreadsheet: https://goo.gl/9Pzxt5

and I got Speed 8.33 pps and a break/make ratio of 69.23/30.77 So I hereby confirm that you are right!

Do you need help? take a look here: https://goo.gl/w2apmG


dsk

TelePlay

I installed one of there inexpensive external USB sound cards for dial speed testing.



https://www.ebay.com/itm/313578863726

I was working on a WE 7D dial that to my surprise tested at 11.26 PPS as its "as purchased" dial speed.

The process of adjusting the dial speed is to remove the spring from the center of the governor (using a fine pointed tweezers to unhook one of the ears (red circles) from the governor wings and removing the spring.



Adjustment is to either 1) reduce the spring tension to slow down the dial (with less tension the wings open easier or sooner and slow down the governor and slow down the dial speed) or, 2) increase the spring tension to speed up the the dial (with more tension, the wings open harder or later and speed up the governor and speed up the dial speed).

It's a trial and error process. In this case, the dial was fast so the first attempt was to loosen the tension. There is no way to know how much the adjusted spring will change the dial speed.

In my first attempt to slow down the dial by decreasing spring tension (A), the dial dropped from its original 11.26 PPS to 9.20 PPS, in spec but too far from the ideal 10 PPS.

The second attempt (B) was to put more tension in the spring and the speed changed to 11.04 PPS.

The third attempt (C) was to put less tension to slow down the dial speed and it dropped to 9.51 PPS. This could be acceptable to some but being on the low side of 10 means that if the dial slows down in the years going forward, it could drop closer to 9 PPS. I prefer to adjust dials so they are a bit more than 10 PPS.

The fourth attempt (D) put more tension into the spring and it tested at an acceptable 10.24 PPS.

I got lucky with just 4 adjustments in that I have had to adjust dials 5 or 6 times to get them in my spec. I've also been lucky a few times in hitting the "just a bit more" than 10 PPS on the first adjustment. In bending the spring by hand, there is no way to know how much tension was put in or taken out of a spring until the spring is re-installed and the dial speed tested.

The Break/Make ratio for this dial tested at 61/39 so no need to adjust that dial mechanism.

I normally don't keep each Audacity wave form for each adjustment but this time I did to show how the trial and error method looks on paper. The thin yellow line shows the speed swings from slow to fast to slow to the final 10.24 PPS dial speed.

The red vertical line on the left is the start, the red vertical line on the right is where the 9th pulse would occur if the dial was returning at the perfect 10.0 PPS. When testing dials using this hardware and software, only the first 9 digits when dialing 0 (ten) is performed so a perfect 10.0 PPS dial speed would have the 9th pulse occurring at 0.9 seconds from the start of the dial return. In this diagram, the dial starts to return at 0.2 on the chart and would stop 0.9 seconds latter (1.1 seconds on the chart). You can see how the trial and error process moves the dial speed from one side of the perfect line to the other.



PS: The top scale of 0.2 to 1.1 is arbitrary, could be 1.3 to 2.2, and came to be by cropping the wave forms during post recording processing.


Yesterday eats you up, it eats everyone up . . .

SUnset2

I apologize for hijacking this thread, but I can't find a better spot.  I have a 50AL that gives correct dialing on the BK tester, but I get dialing errors when I connect the phone to the fiber adapter.  I measured the dialing speed using an oscilloscope.  It turns out to be running a little over 14 pps.

So, I am trying to figure out how to adjust the governor on a WE No. 2 dial.  I found a thread on here that referred to BSP 501-162-100, which shows how to adjust the governor on a No. 5 dial using a 260 tool.  I don't know if this applies to a No. 2 dial, and I need to know how to do it without a 260 tool. 

Any advice?

TelePlay

A WE #2 dial has the same governor as a #5 dial.

The governor has an arm the pivots on one end and has an adjustment screw at the other end. Just above the screw are adjustment markers.

ABB03C00-F54A-4DF8-B4B6-E0954DCF6C6A.jpeg

There was a tool but it can be adjusted without it by using a small screwdriver.

Simply loosen the screw a bit so the screw end of the arm can be moved left and right.

I forgot which way slows down the speed so look at your governor and if the arm is off to one side, move it to the middle, tighten the screw and retest the speed (if the arm is near the middle already, someone may have gotten oil inside the governor raceway and that prevents the governor braking - in that case you will have to remove the oil with lacquer thinner and when clean and dry, retest the speed).

Keep doing that, moving it a bit each time left or right, until you get it close to 10 PPS.

Here's a topic with a picture of the governor removed from its race way. You do not have to take anything apart to do this adjustment.

http://www.classicrotaryphones.com/forum/index.php?topic=2918.msg39423#msg39423

Yesterday eats you up, it eats everyone up . . .

TelePlay

Another thing to check first is movement of the 2 wings in the governor.

Use a toothpick to see how easily they expand. If they seem sticky (a very, very rare occurrence), the pins they pivot on (one pin on each wing, the rose heads seen from the top), need lubrication but be careful not to get the lubrication inside the race.

If the wings open easily, the next step is to clean any lubrication out of the race with lacquer thinner and then put a small amount of lubricant on each pin, between the metal adjustment arm and the wing.

Then use the set screw to adjust the speed.

Your governor could need any or all of the 3 steps.

Yesterday eats you up, it eats everyone up . . .

countryman

Quote from: TelePlay on June 14, 2022, 08:53:47 AM... clean any lubrication out of the race with lacquer thinner ...


You might also want to try brake cleaner as used in auto shops, it comes in spray cans. It evaporates clean and very fast and is safe on many materials, yet not on all kinds of plastic. It's made to remove gummed up grease easily.

SUnset2

I cleaned the governor as well as I could using naphtha in case it was oily.  I verified that the wings pivoted smoothly.  I retested, and the speed was the same.  I could see that the screw was almost at one end of the slot.  I jammed the governor in place with two toothpicks, loosened the screw, and moved it most of the way to the other end of the slot.  I tightened it and re-measured the speed.  9.94 PPS.  Close enough.  I connected the phone, and was able to dial out without error. (Note that the picture is a different scale, both horizontally and vertically).

Thank you for the excellent advice.

TelePlay

Excellent!

Strange that it was set so fast but glad you were able to get it close to 10.

Yesterday eats you up, it eats everyone up . . .

TelePlay

#58
Sometimes this old stuff amazes me. Got a 1972 WE Princess some 8 to 10 years ago, put it into a box then, opened it yesterday and tested the dial. Untouched since built, the dial tested at 9.92 PPS with a 60.4/39.6 Break/Make ratio.

I tested the dial by connecting my USB external sound card directly to the red and green prongs of the 4 prong line cord connector still on the original round line cord. The phone was on my bench and I used a 50' long modular cord to get the pulses from the phone to my USB mic in connecting harness.

The set up here is annotated. The external USB sound card required a stereo mic plug so cut the ear buds of off a set of earphones, soldered the red and green stereo connectors together and use those along with the black ground wire to connect to the phone dial.

The connection can be made on the dial pulse terminals, on the network where the pulse wires are connected, the red and green wires of a modular Jack used as an adapter for a modular line cord, the red and green prongs of a 4 prong plug or on the dial pulse wires found on newer dials.  I used a loose WE 5H dial I had laying on my bench just to show the set up. This was not the dial in the Princess (a #8 IIRC, never took it out of the phone).

I simply open Audacity, select the input as the USB card, start record and dial 0 six times. I select the 3rd or 4th set of data points for analysis.

In this dial's case, the audacity waveform was clean and clear. Inputting the Audacity times into my Excel spread sheet showed the dial was working at 9.92 PPS with a 60.4/39.6 Break/Make ratio.

Rarely can I get a dial close to perfect spec after cleaning and adjustment. To find one as removed from service probably 30 to 40 years ago still factory perfect if remarkable.

This reply is basically to show how easy the interface from dial to computer is, how easily the data can be collected and how accurate the results are using a simple Excel spreadsheet (note, the last "make" point tails off and can not be read so 9 full break/make cycles are input into the spreadsheet and the speed normalized to 10 cycles).

No need for anything other than a computer with Audacity (free software), clip wires, an external USB sound card (to protect the computer internal sound card) and a simple harness to connect the USB mic in to the wires coming from the dial.

Yesterday eats you up, it eats everyone up . . .