Detecting pulses, or digits.

[5415551212]

This film shows how the operator switched both common battery manual and local battery magneto telephone lines. While the Catalina and Michigan used combined manual/magneto boards, the Virginia City board was strictly magneto. The dials shown on these boards were for calling other exchanges.

• https://youtu.be/7qLU_urEYVE?si=_tPFp6aa9WSpt9yz

"A visit to the Bryant Pond (Maine) Telephone Company in 1981. Bryant Pond was the last place in the USA to convert from a hand crank magneto telephone system, which was in use until days after this video was taken."

• https://youtu.be/DJdg4A5ZmJI?si=-7EyXXzpyBz7stXJ

Starting at the 6:17 mark-

• https://youtu.be/0XyFEv0B_PI?si=-zptEfmFMeVj42pr&t=377

Neat to see how small a phone company could be, thanks for sharing those videos.

[5415551212]

What i want is to emulate the interface that magneto phones use, but have a connection to the world. I want to have the phone operate as it originally did.

The interface that magneto phones used was a live operator.
I think it would not be that hard to build a DIY board based on a opensource voice command project like jasper, running on a raspberry pi, and add a port that would detect magneto ring and connect you to the virtual operator.
And that operator could initiate a VOIP call and connect the magneto phone.
https://jasperproject.github.io/

[countryman]

I played a little with the voice control of my smart phone and noticed it would not work with different headsets, or would need a new teach-in for each of it. I'm afraid even a good performing system would have a hard time to recognize words spoken into an old transmitter with limited frequency response and high distortion?
It would be nice if it worked though...

[mR_Slug]

Sorry for the long post. I wanted to try to address everyone.

G-Man, Yes coded ringing, as it is simple to implement. My initial aim is to detect a digit dialed by the fxo/standard telephone, say a 1, and then ring the magneto phone. Coded ringing is at that point is a copy-paste procedure in code. Selective ringing is much more complicated and to be honest I've not really researched it much. I'm sure it could be extended to provide these schemes, however a too complex place to start, like with a key system.

To clarify, by party line I mean standard magneto phones connected in parallel. They work without an exchange (railway and other industries) but a telephone service, provided a party line connected at the exchange. As I understand it, in the most basic somewhat-automated form, coded ringing, all parties hear an outgoing call as a constant (magneto-driven) ring. Incoming ringing, is coded, that is different "ring tones". All parties hear all incoming calls. Not as perfect as an expensive dedicated line to the exchange, or the more complex selective ringing.

<offtopic>
Thank you for the videos, thoroughly enjoyable, though brings up another question. The bell jack plugs on the board, they look like UK GPO "bantam". Are they compatible?
</offtopic>

5415551212
Not heard of jasper. What I envisioned was Dial and CB phones connected to FXS ports. CB phones would need to have a dial-on-off-hook function. Then this adapter on an FXO port. (Either with just one, or as a party line.) Then everything is connected with asterix + voice recognition.

countryman
Yes this is a bit of an issue. As I understand it most solutions are trained on 16bit PCM, and voip is 8bit. And the old transmitters, but, it only needs to get numbers. We are not taking a sample and determining an arbitrary word, we can use an analysis that is a choice of just 10 numbers. It narrows it down.

ka1axy
I've been trying to look at optoisolators in phones. found this:
https://www.epanorama.net/documents/telecom/ringer.gif
https://www.epanorama.net/documents/telecom/telephone_ringer.html

I've been trying to determine how to detect a pulse without also detecting rings. I suppose a pulse detector that also detects rings can be paired with a ring detector, and then rings detected can be removed from the pulse detection signal. Must be a simpler way, can we use a choke?. On the first diagram on this page
https://www.epanorama.net/documents/telecom/teleinterface.html#interference
or this http://www.egyed.com/phonework.html

The coils of the relay, also act to filter AC from the power source. As i understand it, a 5 henery choke can be used. Does this stop a ring signal?

In the old time world where a relay detected a pulse, didn't it oscillate during a ring? I guess you could work it out in software, but does anyone have any circuits? Or how to strip the ring signal? Filter circuit like in one of these?

https://www.arrow.com/en/research-and-events/articles/using-capacitors-to-filter-electrical-noise

Although a voltage based removal is more applicable. Now I am having problems seeing the zener back to back across the optocouple led, and not suppressing the ring between tip and ring. Maybe i  need to re-read your post.

for the optoisolator, quick calcs:
150V AC, say 150V DC 20ma for led is 7.5Kohm 3W But we don't want it to trigger.
on  hook     ~48v DC 20ma for led is 2.4kohm 1W But its magneto, so its 0V
off hook  max~12v DC 20ma for led is 600 ohm 0.25W

Think that's everything. Must say, i'm much more familiar with DC than AC. Thank you.

[5415551212]

Mr slug, you should check out forum member Gregg's project for some component ideas:
https://www.seriss.com/people/erco/1a2-multiline-board/

He has done a lot of research on components, including opto isolators.
I have been slowly building up one of his boards.

His board is using a pic micro controller to re-impliment one of the DP-1900 intercoms I mentioned earlier and it can also detect DTMF.

I highly recommend getting a key system powersupply for your experiments.

[ka1axy]

Detecting pulse vs ring...a ring is AC, so you can detect it with a series cap (see the ringer in the 500 sets). Ring is also higher voltage than dial pulses, so you can discriminate that way too.

I think timing is the key. If your ATA sees loop current and a ring signal (so high voltage) the user is trying to initiate a call. If the telephone sees loop current and pulses, time to decode. If the telephone sees loop current and a ring, it's someone else on the loop trying to wake up the ATA.

I think maybe a high voltage detector and a low voltage detector, and if the HV detector sees a signal, disregard what's on the LV detector, otherwise try to decode pulses on the LV detector.

[mR_Slug]

I've had a bit of a re-think, since we are going to generate rings anyway, we can always strip it off the signal getting to the FXO port, then re-add it later. That is we take the ring signal and digitize it to an out of bound signal, the CPU decides if it gets through.

Say we are ringing from a magneto phone, while a conversation is going on. The DC offset is removed via the 2uF caps. At this point we have audio with a very high ring signal. We clip it here to 10V with the zeners.  Now that the max signal is 10V, we can use a 12V relay to short the connection causing no signal.

The bridge rectifier detects rings, and is used as input to the relay. This should make the ring appear as either a pulse or hang-up. Or have i got this all wrong?

Could we detect pulses in the purple/pink low voltage section?

I have copied the termination of a phone line at an exchange (on the right to the FXO port). However I have just used a couple of 5H chokes, instead of a relay actuating on the audio signal. How does the -48V battery connection switch to talk voltage on loop connection?

Errata: Relay should be NO.

[5415551212]

Gregg uses an LDA110 for "off hook" line and ring detect.
Did you look at his board layout?
https://www.seriss.com/1a2-ksu/rev-j6a/data/1a2-REV-J6A-photo-top.jpg

https://www.seriss.com/1a2-ksu/rev-j6a/data/1a2-schematic-REV-J6A-p001.jpg

https://www.digikey.com/en/products/detail/ixys-integrated-circuits-division/LDA110/203131

[mR_Slug]

Thanks for posting it right here, I designed the previous circuit before reading your links. Bit puzzled by it though. Ok so max current thru the input of the LDA110 is 100mA, volts is 30V. I have read that 20mA is standard for talk. A ring at max 150V is 106V RMS. In this diagram, the 1k resistor acts as a shunt, causing a volt drop, this is then presented to the LEDs. But what is the current put out buy a typical magneto? 4REN is what in mA? I have no doubt it works, I just don't understand it.

Datasheet:
https://ixapps.ixys.com/DataSheet/LDA110.pdf

[ka1axy]

Off Wikipedia:
In the United States, the FCC Part 68 specification defined REN 1 as equivalent to a 6930 Ω resistor in series with an 8µF (microfarad) capacitor. The modern ANSI/TIA-968-B specification (August 2009) defines it as an impedance of 7000Ω at 20Hz (type A ringer), or 8000Ω from 15Hz to 68Hz (type B ringer).

So, back of envelope 150V peak across 7000 ohms is about 20mA

[5415551212]

Greggs board is of course designed for a modern phone line or a VOIP line.
You'll have to do some testing and breadboarding and of course keep us posted.

[leejor]

The issue with using an FXO port is that it is the device/PBX that the call routes through to is going to have to initiate an off-hook. That won't be a function of the gateway. In addition gateways are not made to "see" dial pulses incoming. If anything, when they see a loss of current and the option (CPC) is enabled, they will drop the call. so while you may be able to establish an open circuit, you'd have to use DTMF to pass any digits.
It would require some sort of adapter to "read" ring generator,then seize the line, and repeat the dial pulses, but you'd have a much better chance of this working using and FXS port.

[mR_Slug]

Long post, (Basically is my diagram correct or nonsense)

Thanks. So 1 REN is 20ma so 5 is 100ma. What these magnetos can put out is another question. A party line with 10 ringers is 200mA. 50 is 1A, that is assuming the bells were all 1 REN. So would assume they were pretty beefy.

Evaluating the LDA110 setup:

Effect of the 1K resistor on talk? It's bypassed by the double diode, right? they are 0 ohms. Or say like 3, so parallel diodes with the 1000 ohm resistor is ~3 ohms. That is they present 3 ohms to the circuit. This is a guess, can't work it out from the datasheet.

So in a ring situation, we have the load of the ringers 20-100mA and 2333-7000 ohms. This additional 3 ohms is added and can be thought of as R1 of a voltage divider, R2 the ringers. So, (RMS)

REN 5 (100mA):
100v x 2333 ohm = 233300
                  ----------- = 99.87V across r2 so 0.13v across R1.
3ohm + 2333 ohm = 2336 ohm

REN 1 (20maA):
100v x 7000 ohm = 700000
                  ----------- = 99.96V across r2 so 0.04v across R1.
3ohm + 7000 ohm = 7003 ohm

REN 50: (big party line, 1000mA + HV)
200v x 140 ohm = 28000
                ----------- = 191.78V across r2 so 8.22v across R1.
3ohm + 143 ohm = 146 ohm

As such this seems like a very volt tolerant method. However, nearly all the current goes through the diodes. Is there another way? I've tried changing the resistance of the diodes with a series resistor, and recalculating everything, but I can't make it work. So, voltage divider:

        Vs * R2
Vout = ---------
      (R1 + R2)

At talk:
        6 * 2000      12000
Vout = ------------- = ----- = 1.5v
      (6000 + 2000)  8000

I = V/r = 6/8000 = 0.75mA thru the circuit.

now, ring voltage (200V RMS):

        200 * 2000    400000
Vout = ------------- = ------ = 50v (but clamped to 10V by the zener)
      (6000 + 2000)    8000

I = V/r = 200/8000 = 25mA total
I = V/r = 10/2000  = 5mA So does that mean 20mA goes thru the zeners? Confused

Now in talk @ 6V, we have 1.5V at 0.75mA at the LEDs.
At ring @ 200v we have 10V clamped at the LEDs, and 25mA?
At ring @ 100v we have 10V clamped at the LEDs, and 12.5mA?

Either way at talk voltages, we can trigger an LED, and use very little mA. And at ring voltages, even high ones, the mA consumed is about 1 REN. Have I got this circuit correct? is this what you had in mind ka1axy?

It is designed to detect pulses but as seen also detects rings. And, so we need to still detect rings separately to mask them. However since only this device can initiate a ring on the FXO side, we will know this anyway.

I did a quick search for the LDA110 and it looks like its out of stock, The Vishay H11AA1 was recommended in a forum, but the datasheet seems to be missing the min voltage. Can anyone recommend optocouples?

Datasheet;
https://www.digikey.com/htmldatasheets/production/1280241/0/0/1/h11aa1.html

I've been looking at dial tone generation, and DTMF. how does the WE 1500 & 2500 generate tones? Datasheet: https://www.telephonecollectors.info/index.php/browse/wiring-diagrams/western-electric/2500-series

leejor, I think you are misunderstanding. The FXO port has to dial the device. It can use tone or pulse, but the intention is to signal TO the device, FROM the FXO. This then selects a ring tone. The magneto phones can't dial   When you wind a magneto, this device detects it, and sends a ring to the FXO port, which is then forwarded to an operator.

[G-Man]

Initially the original Ringer Equivalency Number was based on the load presented by a Western Electric type C4 ringer. Later on as other standards organizations became involved, a new method was arrived at to compute it.

When attempting to compute the REN of magneto telephone ringers, other factors come into play including whether they have low or high impedance coils and if condensers (and their value) are present in the circuit.

Also keep in mind that an FXO port does not switch calls, that is the function of the central office or PBX that it normally would interface with.

Below is a description by Western Electric regarding the capabilities of magneto party-lines.
Note that the number of telephones that a magneto instrument can ring is dependent on whether it is equipped with a two, three, or five-bar magneto.

MAGNETO TELEPHONES

Definitions of Terms

     The following definitions refer to terms used on the following pages in connection with our magneto telephones.

SERVICE
     The number of telephones that can be connected on the same line varies, ranging from 1 to 40 or more. However, a line having more than 20 or 30 telephones connected is usually very unsatisfactory from a service standpoint, except in case of necessity or for temporary service. The reason for this being that a line carrying so many instruments is bound to be in use almost continuously, the bells ringing at very frequent intervals and the user almost sure to be "rung in the ear" or otherwise interrupted during the conversation.

     The following definitions of what may be considered a lightly loaded, medium or heavily loaded line are submitted with the thought that its limits are conservative enough so that under all but extreme conditions the figures given can be relied upon. On the following pages will be found a complete catalog of telephones and opposite the listing of each type is specified the kind of loaded line upon which the particular telephone will give best service. Telephones should never be used on lines loaded heavier than indicated as the maximum for each type.

     The telephone lines referred to are assumed to be well insulated and free from high resistance joints.

Light Loaded Lines A light loaded line is one of less than 15 miles in length and not equipped with more than 12 telephones.
Medium Loaded Lines A medium loaded line is one between 10 and 30 miles long and equipped with from 10 to 30 telephones.
Heavily Loaded Lines A heavy loaded line is one up to 40 or 50 miles long or equipped with up to 40 telephones. Lines loaded with this number of telephones are rapidly going out of use and being broken up into shorter lines with fewer number of telephones. Lines of this length or loaded with this great number of telephones should be discouraged in all cases except as before stated, in cases of extreme necessity or for temporary service.