Open Source Pulse to DTMF converter project/problem, a detailed technical discussion

[royalbox]

Hello,
I built an open source pulse to DTMF converter from here: https://bitbucket.org/310weber/rotary_dial/overview (circuit diagram attached) for use on a UK 746 telephone.
The only problem I have with it is that it cuts down both transmitting and receiving volume. I've been told I sound distant.
As soon as I remove one of the power connectors the volume is restored so it is not the phone that's the problem.

Anyone here familiar with electronics who could give me a clue why this might be?

As a side issue, the DTMF volume is a bit too loud.

Thanks for any help.
Barry.

[ThePillenwerfer]

I had the same problem and fixed it by increasing R1 to 1kΩ.

I found that the tones weren't load enough to always be detected.  I increased their volume by reducing R2 to 200Ω.  If they are too loud try changing it to 470Ω.  I'm certainly no electronic expert and it may be better to increase R3.

Details of my version, modified for UK use, is here: http://www.classicrotaryphones.com/forum/index.php?topic=17284.msg185709#msg185709. 

I've now made three and found them excellent.  I use them with an ATA (Analogue Telephone Adapter) for a VOIP (Voice Over Internet Protocol) line so, if yours is on a standard line, it may explain why we've had different results with the tone volume.

[royalbox]

Thanks  very much, that's helpful. I'll give that a go over the weekend.

Reading the post you linked to it looks like you had the same problems as I did. Trying to find how to program the chip and connect it to the telephone. I think I used the rototone instructions as a guide, not sure, it's been while.

I wonder if it would be better powered independantly, and doing away with R2. Less convenient though of course.

Thanks again,
Barry.

[Alex G. Bell]

In the circuit in the first message the zener diode has a near-zero impedance so this circuit looks like 220 ohms impedance to the phone line, which is 600-900 ohms.  So most of the voice signal energy is shunted by the 220 ohms and zener path. 

You need to raise the AC input impedance.  The simplest way would be by connecting an inductor capable of handling the DC without saturation or a constant current source in series with the 220 ohms to raise the AC input impedance at voice frequencies. 

However it would be better if instead of using a shunt zener regulator you replaced the resistor and zener with a simple series pass emitter follower voltage regulator.  The collector of the transistor would reflect a very high impedance towards the line.

If you do this correctly it will have very high input impedance and cause no voice signal shunt loss.  The circuit also needs to be protected against polarity reversals and transient voltages which it will experience if a call is answered while the phone is ringing or other circumstances.  The 2N3904 is not a very conservative choice.  A higher voltage part such as 2N5551 would be more prudent.

The high DTMF levels are another topic.

[TelePlay]

In the circuit in the first message the zener diode has a near-zero impedance so this circuit looks like 220 ohms impedance to the phone line, which is 600-900 ohms.  So most of the voice signal energy is shunted by the 220 ohms and zener path. 

You need to raise the AC input impedance.  The simplest way would be by connecting an inductor capable of handling the DC without saturation or a constant current source in series with the 220 ohms to raise the AC input impedance at voice frequencies. 

However it would be better if instead of using a shunt zener regulator you replaced the resistor and zener with a simple series pass emitter follower voltage regulator.  The collector of the transistor would reflect a very high impedance towards the line.

If you do this correctly it will have very high input impedance and cause no voice signal shunt loss.  The circuit also needs to be protected against polarity reversals and transient voltages which it will experience if a call is answered while the phone is ringing or other circumstances.  The 2N3904 is not a very conservative choice.  A higher voltage part such as 2N5551 would be more prudent.

The high DTMF levels are another topic.

So, replace Q1 (the 2N3904) with a 2N5551

Then, place what sized/type of inductor in front of the zener diode ( D1) or between D1 and GND? Or replace D1 and R1 with what. What would you do, your first choice?

[Alex G. Bell]

So, replace Q1 (the 2N3904) with a 2N5551

Then, place what sized/type of inductor in front of the zener diode ( D1) or between D1 and GND? Or replace D1 and R1 with what. What would you do, your first choice?

I said to place the inductor in series with the 220 ohms.  Connecting it between D1 and GND would be in parallel with D1. 

Not sure of the inductor value, probably 100mH, perhaps smaller, but using a series pass regulator instead of D1 and R1 would be much better.  There are lots of articles on line about designing simple emitter follower series pass regulators.  I don't have the free time to research it.

[ThePillenwerfer]

I'm loathe to comment further as this thread has already gone beyond my comfort zone but I do wish to point out a couple of things.

The modified version I referred to in post 2 has a polarity protection bridge.

Both the OP and I are in the UK and things are a bit different here.  Due to our use of a separate ringing wire I can't see how the ringing voltage would get to the converter, though I accept that I could be wrong in this belief. 

Finally UK line impedence is 300 Ω + 1000 Ω || 220 nF^*,  though don't ask me what that means.


^*Source: British Telecom SIN 351 paragraph 3.1.2

[TelePlay]

I said to place the inductor in series with the 220 ohms.  Connecting it between D1 and GND would be in parallel with D1. 

Fully aware of what you said, was asking "where" to place it S1, S2 or S3. I've always considered "series" in  line, parallel next to. Is that wrong?

Unlike many people, I have to see schematics, I don't think schematics. And, I can't read minds, even though my wife thinks I should be able to.

Had to deal for way to many years with arrogant medical sales people who felt we the servants of them were expected to know what they wanted for their doctors the moment they thought it, and then if they changed their minds, to immediately know of the change, turn the truck around back to the warehouse to have the tech staff reload the implants and instruments, put it back on the truck and then still get it to the hospital to be sterilized and delivered to the OR in time for the scheduled operation. "Said" doesn't carry much with me anymore, hits me as an insult the same way as some treat the ignorant when they are accused of asking a "stupid" question. The only reason I put up with those arrogant Type A's and paid my own speeding tickets for rushing changed orders to them was because they sold to the doctors who paid my company who paid my salary.

A simple, civil question asking for explanation could be simply and reasonably answered by a diagram showing what was meant, such as that attached, which would make it a lot more easy to understand by those of us who are not electrical engineers. After all, a picture is worth a 1,000 words and the corollary, 1,000 words does not make a picture.

[Alex G. Bell]

I'm loathe to comment further as this thread has already gone beyond my comfort zone but I do wish to point out a couple of things.

The modified version I referred to in post 2 has a polarity protection bridge.

Both the OP and I are in the UK and things are a bit different here.  Due to our use of a separate ringing wire I can't see how the ringing voltage would get to the converter, though I accept that I could be wrong in this belief. 

Finally UK line impedence is 300 Ω + 1000 Ω || 220 nF^*,  though don't ask me what that means.

^*Source: British Telecom SIN 351 paragraph 3.1.2

Someone said the circuit produces a lot of loss.  A 220 ohm AC impedance shunt would produce high loss on any telephone circuit anywhere unless the laws of physics are different there. 

300+1000 sounds like a DC resistance statement rather than an impedance statement but if it is impedance then the 1300 ohm sum would be even higher than 600-900 ohms and the 220 ohm shunt would produce even greater loss.

I don't see a polarity guard rectifier bridge in the schematic nor mention of it in this topic.  There are too many descendant references in a string of links to attempt to synthesize what may have been done which is not shown in this topic.

The UK may well use a 3rd wire within the premises for ringing but the line to the exchange line nevertheless uses 2-wire lines.  Therefore ringing is applied across the 2-wire line and appears across the speech network in the phone at the moment of answering. 

Ringing voltage is not removed until sufficient time has passed (a few cycles of AC) to recognize DC answering current while ignoring AC ringing current flowing on the line to operate the ringer.   Therefore reversed polarity will appear across the line momentarily at the moment of answering and be applied to this circuit as a few cycles of AC will be applied before ringing is removed.  If the loop is short the current can be high since ringing voltage is higher than the applied DC voltage.

Even increasing R1 to 1K only reduces the loss from a grossly obvious value to a less obvious one (probably about 3dB) instead of reducing it to 0.  It's a makeshift solution rather than a proper solution.

[Alex G. Bell]

Fully aware of what you said, was asking "where" to place it S1, S2 or S3. I've always considered "series" in  line, parallel next to. Is that wrong?

Unlike many people, I have to see schematics, I don't think schematics. And, I can't read minds, even though my wife thinks I should be able to.

After all, a picture is worth a 1,000 words and the corollary, 1,000 words does not make a picture.

By that logic you could make the case for putting the choke in series with C1, C4, or U1 pin 8 or even between Q1-C and terminal 4 and it would be equally in series with R1 to the same degree as it would be by putting it in series with D1.   Whether it is in series with the cathode or anode of D1 would be absolutely equivalent.  So "in series with R1" means just that: either between Q1-C and R1 or between R1 and R4 as they appear on the schematic. 

Of course it must be understood that the topology of the physical embodiment of the schematic is not necessarily the same as the topology of the schematic.

Sorry, I cannot spend the time to start editing the diagram.  I believe "in series with R1" is not ambiguous and there is no need to.

[ThePillenwerfer]

Point taken about the ringing voltage.

As to whether the figure I gave is resistance or impedance the picture below is taken directly from the official documentation.

[TelePlay]

Someone said the circuit produces a lot of loss.  A 220 ohm AC impedance shunt would produce high loss on any telephone circuit anywhere unless the laws of physics are different there. 

300+1000 sounds like a DC resistance statement rather than an impedance statement but if it is impedance then the 1300 ohm sum would be even higher than 600-900 ohms and the 220 ohm shunt would produce even greater loss.

I don't see a polarity guard rectifier bridge in the schematic nor mention of it in this topic.  There are too many descendant references in a string of links to attempt to synthesize what may have been done which is not shown in this topic.

The UK may well use a 3rd wire within the premises for ringing but the line to the exchange line nevertheless uses 2-wire lines.  Therefore ringing is applied across the 2-wire line and appears across the speech network in the phone at the moment of answering. 

Ringing voltage is not removed until sufficient time has passed (a few cycles of AC) to recognize DC answering current while ignoring AC ringing current flowing on the line to operate the ringer.   Therefore reversed polarity will appear across the line momentarily at the moment of answering and be applied to this circuit as a few cycles of AC will be applied before ringing is removed.  If the loop is short the current can be high since ringing voltage is higher than the applied DC voltage.

Even increasing R1 to 1K only reduces the loss from a grossly obvious value to a less obvious one (probably about 3dB) instead of reducing it to 0.  It's a makeshift solution rather than a proper solution.

I'm loathe to comment further as this thread has already gone beyond my comfort zone . . .

I fully agree, 99% of the technical discussion in the above quote is off topic. It may be a very good technical lesson but way to complex to the spirit of the topic.

As such it was moved into the THTD board since it no longer fits into the technical repair/help board. Would have been nice to give royalbox time to try the "fix" first offered to see what would have happened before having a Bell Labs lesson on telephone theory. If necessary, royalbox is free to start a new topic in Technical or Troubleshooting to ask for help and get a physical fix to the problem, in 6 replies or less.

[Alex G. Bell]

Point taken about the ringing voltage.

As to whether the figure I gave is resistance or impedance the picture below is taken directly from the official documentation.

OK then, 220nF is a small value only effective at higher frequencies in the voice band and effectively at the lower frequencies the Z = 1300 ohms.  Therefore if the telephone set speech circuit itself has an impedance which is matched to the line, a fundamental requirement to minimize signal power loss, the telephone set has a similar impedance to 1300 ohms. 

Consequently 220 ohms would present a substantial shunt loss and raising it to 1000 ohms would still cause more than one half the voice frequency signal power received from the line to be dissipated in the 1K resistor since it is less than 1300 ohm impedance of the telephone set speech network.  But it's worse than that.  The terminating impedance presented to the line becomes 1300 of the telephone set in parallel with the 1K resistor so there is mis-matching loss too.

You might do well to measure the actual signal level on the line with and without the converter connected if there is a standard transmission loss measuring tone available by dialing a code in your exchange area and you have a transmission measuring set or any instrument capable of measuring AC mV down to about 20mV.

Alternative to dialing a code for a transmission test tone, and since you do not need to measure an absolute level, just a relative one, you could call someone who has a DTMF telephone set and if DTMF telephones in the UK are like those in the US, pressing 2 buttons in the same row or column will cause a single tone common to both buttons to be generated.  The 4th row is 941Hz.  The 1st column is 1209.  The 3rd column is 1477. 

You can measure the signal level at one of these frequencies with an assistant sending one of these tones with the converter connected and disconnected, measure the tone level across the line in both conditions and calculate the power loss from the voltage difference.

[Alex G. Bell]

I fully agree, 99% of the technical discussion in the above quote is off topic. It may be a very good technical lesson but way to complex to the spirit of the topic.

I agree but it's inherent that seemingly simple questions about complex electronics do not have simple answers.  "You should simplify as much as possible, but not more..."

And BTW, I was reluctant to get involved in this for this very reason.

[royalbox]

Interesting replies so far, thanks everyone. I won't pretend I understand all the theory and I also can't picture things in my head easily which is often frustrating. Appreciate the help though.

Regarding the polarity protection, I've fitted a bridge rectifier to the incoming lines as recommended in the rotatone installation guide, like the second diagram here: http://www.rotatone.co.uk/fitting-a-rotatone-to-a-gpo-706-746-8746/.

I'll look into series pass emitter follower voltage regulators but not confident I'll be able to do anything with it myself.

Barry.