Precise dial speed setting

[BakeliteBear]

Can anyone offer some clarification on setting exact dial speeds and how this relates to the various lines offered by BT, Virgin and Telewest?

To clarify, I have serviced dozens of dials, which I have set by eye and tested on my home Virgin (formally) NTL line. I have never had any problems, but I am told that this is not good enough for some exchanges, particularly Telewest.

I have also been told about a ‘Lab-line’ tester. Does anyone know how useful/necessary it is, where I might source one and what sort of costs are involved.

[Station X]

I can't give any exact figures for the various Telcos and they may vary from exchange to exchange depending on who manufactured the exchange equipment.

Detecting dial pulses electronically is much more difficult than doing it electro-mechanically and of course it never was intended that they'd need to be detected electronically.

You may find that having the correct mark/space ratio is just as important as having the correct speed.

[russell_w_b]

Originally Posted by BakeliteBear

Can anyone offer some clarification on setting exact dial speeds and how this relates to the various lines offered by BT, Virgin and Telewest?

I'm guessing that they'll all work from the old GPO standard, which is as follows:

Minimum inter-digital pause: 400ms.
Permissible range of impulse: 7 - 12 Impulses-per-second (nominally 10).
Standard inpulse ratio: 66.66r% break (tolerance 63% - 72%).

Considering a call can be tapped out on the cradle, and a small child can alert the Pan-EU emergency services by idly tapping out a sequence of thumps equating to 112, the tolerance - at least on BT - is pretty wide!

[Station X]

When Post Office Telephones started to design their digital exchanges (System X) in the late 1970s they knew that they'd have to work with dial telephones with a wide range of tolerance, so it was designed in. Take up of MF4 (tone dialling) phones was slow at first. As far as I know BT is the only UK Telco whose licence requires it to provide a pulse dialling facility.

Other operators may have a tighter spec for detecting dial pulses. as they expect everyone to be using MF4. Dial phones are a rarity now and mainly used by collectors like ourselves.

Simulation of low digits like 1 and 2 by switch cradle tapping is easier than simulation of say digit 0. This is why 112 was a bad choice of emergency code for networks which still supported pulse dialing. 112 is easily generated by say open wires touching in the wind, by a cable fault, or by switch hook fumbling on lifting the receiver and then dialling a number beginning in 2. For this reason numbers starting in 1 were never used in the early days of automatic exchanges. In fact 1 wasn't used as an initial digit until the 100 operator code was introduced with STD. The proverbial "JCB through a cable" can generate hundreds a calls to 112. There are arrangements which I won't discuss here to reject such calls whilst letting genuine 112 calls through.

When pulse dialling telephones with last number recall were introduced there were a lot of complaints from customers saying that they had received the announcement "You have dialled incorrectly, please try again" or got a wrong number, but when they used last number recall the call got through. It turned out that switch hook fumbling was introducing a spurious 1 or 2 in front of the dialled number. Since this digit hadn't been "dialled" it wasn't stored in the phone, so last number recall would work, even though the initial call failed.

[BakeliteBear]

Thanks for the information.

Does anyone have any experience of setting precise dial speeds other than by eye?

[Station X]

Compare with a modern keypad telephone having a loop disconnect setting?

Scope with a long timebase period?

[BakeliteBear]

That’s an cleaver idea - but unfortunately I am short of an oscilloscope.

I also do a lot of dials and was hoping for a more ‘off the shelf’ solution.

[Station X]

The dial speed tester I remember had two semicircular scales. One scale counted dial pulses by means of a simple armature and pawl arrangement. The other scale was effectively a clockwork clock measuring periods of up to slightly more than a second. The first of 10 dial pulses started the clock and the last one stopped it. Speed was then read from the clock scale. I doubt you'll be able to find one of these testers, as they were only situated in repair centres (wipeup) and on test desks.

If you have no test gear, all I can suggest is saying out loud "One thousand, one hundred and one" whilst watching the dial return after dialling digit 0. It can be surprisingly accurate.

You can check mark/space ratio by unplugging the the phone and connecting an analogue voltmeter and battery (say 9V) in series across the line. Wind the dial round to zero and note the reading. Release the dial and note the average reading. Divide the second reading by the first to give the % break.

What about using a scrap car rev counter? 10 pulses per second is 600 a minute or 300 rpm for a four cylinder engine.

[Leon Crampin]

An easier way to check mark/space ratio is to use an AVO on resistance range. Zero the AVO on Ohms x 1 with the dial switch in circuit. Run the dial and take a reading (a degree of estimation is required) using the 0 - 100 linear scale. The percentage "dwell" is indicated directly.

For anyone with an old petrol engined car, the AVO can be similarly used as a dwell meter. This time, zero the Ohms x 1 range to "90" on the 0 - 100 scale. Crank the engine and read the dwell directly.

Useful instruments.

Leon.

[Station X]

I'd thought about that, but dismissed it as the ohms range had a non linear scale. I completely overlooked the fact that a 0-10 or 0-100V scale could be used instead.

How about this for a homebrew speed tester? Stabilised supply of say 12V has a resistor connected in series to limit the current to that which would be found on a very short telephone line. Dial contacts 4 and 5 are connected across the combination. Also across the combination is a capacitor in series with a diode. Across the capacitor is a DMM on say a 15V DC range. Dial is wound round to digit '0' and released. Each time contacts 4 and 5 open the capacitor is charged a bit via the resistor. When the contacts close the capacitor is prevented from discharging by the diode. Each dial pulse charges the capacitor a bit more, so that, plotted against time, the voltage across it is a staircase. The actual final voltage will depend on the resistor value, the capacitor value and the total time the contacts are open. For a correctly adjusted dial the time will be 666.6mS.

The idea might just work, but you'd need to calibrate the device using a good dial.

The theory is here:-

http://www.play-hookey.com/dc_theory/rc_circuits.html

You could try a lash up using a battery instead of a stabilised supply to see if the idea is a goer.

[BakeliteBear]

Thanks for all your ideas - lots to think about!

[BakeliteBear]

Well I think I have stumbled upon a dream solution to my problem! Quite by chance I have been given a vintage BT ‘Dialled Digit Analyser 1A’ which seems to be exactly what I have been looking for.

I have just tested a dial I set by ‘eye’ and the readings are displayed as follows:

BK/MK 72/21 ms
Speed 10.5 PPS

Near the start of this link Russell W B kindly posted the BT standard for dial speeds as follows:

Minimum inter-digital pause: 400ms.
Permissible range of impulse: 7 - 12 Impulses-per-second (nominally 10).
Standard inpulse ratio: 66.66r% break (tolerance 63% - 72%).

Can anyone confirm that this remains the standard for all current exchanges accepting pulse dialling; and can anyone explain how ‘BK/MK 72/21 ms’ relates to the ‘Minimum inter-digital pause’ and the ‘Standard inpulse ratio’?

[Station X]

BK and MK are the break and make PERIODS. For an ideal dial running at 10 PPS these would be 66.66mS and 33.33mS. Your dial has 72mS break and 21mS make. This is a ratio of 72/(72+21)% or 77.4%, which is out of tolerance.

Inter digital pause is not really a characteristic of a mechanical dial. It's the interval between one train of pulses and the next. It depends on how quickly you wind up the dial after each digit in order to dial the next and the length of time between your releasing the dial and the pulse train commencing. This is unlikely to be less than 400mS.

If you had a modern pulse dialling phone with a keypad then obviously the IDP will depend on the electronics as you will finish "dialling" long before the digits are pulsed out.

[BakeliteBear]

Thanks for your help Graham, I will set to my dial and see how close I can get it.

[BakeliteBear]

OK - I have now got the dial speed to 10.1 pps and by adjusting the position of the fixed dial contact I have a make/break ratio of around 67/33, but only at the upper end of the dial. There is significant variation across the dial range:

2 = 74/24
3 = 73/24
4 = 72/26
5 = 70/28
6 = 67/30
7 = 67/32
8 = 65/33
9 = 66/33
0 = 67/33

The dial speed reeds as a constant 10.1 across all dialled numbers. The dial has been fully stripped and cleaned.

Am I doing something wrong?

[Station X]

It's interesting that the tester displays make and break periods rather make to break ratio. Can you step through the pages to get a ratio display?

I'm not sure what's happening here. Assuming the cam has the correct profile, then the make/break ratio won't depend on dial speed. Thus even if the dial slows down as tension in the spring decreases the m/b ratio should stay constant. I'd be inclined to go with the ratio given when dialling '0' as this will be the average m/b ratio of all pulses.

[BakeliteBear]

The tester only displays the make/break periods, but that is possibly because it is also possible to select a display that summarises the max/min values over a dialled number.

Thank you for confirming my thoughts that the dial speed is not related to the ratio of make to break - other than perhaps the remote possibly of giving the cam follower a but more of a shove when it is running faster. It is a type 21 dial, so I will take a look what the tester says about a no.10/12 type and see if that shows the same type of variation.

As a last thought, is it possible that tightening the contact arm which is moved by the cam follower might make it follow the cam profile more accurately?

[russell_w_b]

Are you absolutely sure that your tester is working properly? Can you put an electronic pulse-dialling telephone on it and see if the readings are consistent?

Just a thought...

[BakeliteBear]

No, I am not sure – but I have repeated the test for each number (1-0) on the dial and the figures are almost exactly the same. Using an electronic pulse-dialling telephone is a brilliant idea, but unfortunately I don’t have one! I will have a play with one of my other dial telephones and see if I get a different set of readings.

As an aside, I have spoken to a contact regarding the optimum make break ratio and he tells me that the BT exchange standard is 67/30, but that he was told that Virgin digital exchanges should be set to 70/30. Can anyone confirm that they are different to the BT standard?

[Station X]

It can't be 67/30. That doesn't add up to 100. 66.66R/33.33R where R indicates recurring is more likely.