Telephone line (POTS, PSTN) to MCU ADC interfacing - SLIC

Thread Starter

ag-123

Joined Apr 28, 2017
276
Telephone line interfacing circuits these days seemed to be a 'lost art'.
A search on Digi-Key, Octopart, with the keyword SLIC and sorting by ICs with existing stocks gives the highest counts for chips that are obsolete !
And the newer chips, which integrates the 'SLIC' (subscriber line interface circuit) with ADC, DAC has the digital interfacing protocols omitted in the data sheet. Merely gives the electrical specs with nothing about the SPI and such interface protocols. I'd guess they'd only cater to bulk orders.

Hence, I ended up searching for discrete circuits, some good ones are like
https://www.epanorama.net/links/telephone.html
https://www.epanorama.net/circuits/teleinterface.html
http://www.next.gr/telephone/Telephone-Circuits/

While there are many circuits posted, for some of them, one may wonder if it'd work. And for 'others' a lot of them uses a transformer. I'd guess it is for isolation.

Are there any good circuits or rather ICs that can do without a transformer, as like most 'transceiver' or 'line driver' chips?
can't seem to find many of them except like HC5503, even the datasheet looks 'old'
https://www.digikey.com/en/products/detail/rochester-electronics-llc/HC5503CCB/12134034
 
Last edited:

Thread Starter

ag-123

Joined Apr 28, 2017
276
One of them that I stumbled into in my web searches is:
https://sound-au.com/appnotes/an010.htm


This is a so called '2 wire' to '4 wire' circuit. It seemed quite feasible to use one 'side' of it say with a common op amp e.g. LM358 to interface to microcontrollers ADC, DAC?
The main thing is one side needs to play a 'modem'.
I'm still trying to think how I'd need to cater to the in-coming ringing signal and to 'pickup' the call.
 

Ramussons

Joined May 3, 2013
1,404
I'm still trying to think how I'd need to cater to the in-coming ringing signal and to 'pickup' the call.
Do you have an old Line Modem? You can make it Auto Answer on an incoming call.

Else, you will have to detect the incoming Ring - about 75 Volts@ 20 Hertz. This ring will be riding on a -50 Volts DC.
Detect the ring and give a DC loop across the line. Extend the loop using an Inductor so that you pick up the "Audio" across the inductor.
 

Thread Starter

ag-123

Joined Apr 28, 2017
276
Thanks for the hint !
I'd think 'old' modems works, just that this time i'm mainly interested in voice.
I think i may use a dc blocking capacitor and inductor after all.
I stumbled into this circuit
https://forum.allaboutcircuits.com/...acitor-in-telephone-interface-circuit.111089/
\\

It seemed this is rather 'common' for a basic phone interface.
I may use a transistor, possibly bipolar or mosfet to switch the loop across a resistor, so as to give an 'off-hook' state.
The 'voltages' are somewhat 'difficult' to handle. I'm originally thinking about using resistor dividers and then op amps to bring the voltages into more sane ranges as like in the 2nd post - that "2 wire - 4 wire" circuit setup.
Maybe I'd connect that resistor/transformer/capacitor output to a "op amp 2 wire - 4 wire' setup, this could reduce the transformer to one set.
 
Last edited:

Thread Starter

ag-123

Joined Apr 28, 2017
276
There is something very curious though
https://sound-au.com/appnotes/an010.htm


- lets present 1v at the "right" side of the 2 wire line, and that Out B and In B is initially 0 v.
- So with this Out B becomes -1 V. That looks good.

- Assuming that "right" side VR1b is 'misadjusted', for that matter just remove VR1b.
- But lets just say now make In B 1 V, as VR1b is removed, this 1 V appears at + terminal of U1Ab, that will cause Out B to swing to the +ve rail of the OpAmp, i.e. the input from In B cause a change in Out B.

- The only way it seemed, to fix that, is to make VR1b and R2b 0 ohm - i.e. shorted to GND.
- Then we'd get back -1 V at Out B !
---
next experiment:
- lets disconnect the 2 wire line, make In B 0v and Out B 0v.
- next: keep VR1b and R2b 0 ohm , i.e. shorted
- And make In B 1v, this 1V appears at the output of U1Bb.
- As VR1b and R2b is shorted to GND 0 ohm, now Out B swings to -ve rail of the Op amp. (initially it is 0v)
- The only way as it seemed to fix that is to *remove VR1b - i.e. infinity ohms.
- so now In B i.e. 1 V appears at the + terminal of U1Ab, and Out B goes back to 0v

this circuit is 'hairy' as in order that the 'Out B' won't be affect by 'In B', something has to 'magically' vary VR1b to a 'correct' value
lol
hope to hear your comments ;)
 
Last edited:

Hymie

Joined Mar 30, 2018
1,277
The separation requirements between telephone (POTS, PSTN) and SELV circuits is a complex topic.

The telephone ringing voltage can be as high as 120Vac, and therefore it is important that this voltage cannot reach SELV circuits. Isolation (between the circuits) is one method of achieving this, but impedance protection such that under any single fault conditions SELV voltages remain within permissible limits is also permissible.

You should check out the relevant separation requirements, as set out within the applicable product safety standard for your product, and ensure you comply with them.
 

Thread Starter

ag-123

Joined Apr 28, 2017
276
Thanks, I'm searching around on the IC distributors e.g. digiKey webs and looking for SLIC (subscriber line interfacing circuit) ICs.
I'm kind of hoping to find established documented circuits with the ICs which would likely conform to the interfacing requirements. What is a little surprising thus far is that I see quite a few (many) of the IC part numbers which has the largest inventories are *obsolete* ! Or that information provided in the datasheets isn't adequate to understand the circuit fully - mostly the old and obselete parts. And for the new existing parts, i've seen quite a few that simply provide inadequate info, e.g. missing protocol information when it use a digital interface, I'd guess they'd only provide that to large customers under NDA etc.

Rather commonly, I see many SLIC ICs that use capacitive coupling rather than inductors/transformers. I prefer a capacitive coupling as they tend to be smaller components to work on the PCB. I'd likely explore capacitive coupling further.
But inductive e.g. transformer coupling could be used as well for their merits. This seemed more common in 'older' designs which do not always use 'active' electronics like OpAmps etc. I prefer to use OpAmps in the interface circuit as they apparently are more versatile.

Either way I'm making only a single piece, hence I'd guess I'd make do with a bulky transformer if a transformer coupling makes a 'better solution' lol
edit:
It seemed at least superficially that various designs use transformers, as it is rather easy to have several primary / secondary coils on a same transformer and of course that they are 'isolated'. But that more 'recent' circuits it seemed showed a preference for capacitive coupling. I'd guess it is partly as these days there are 'single chip modem' and the whole board probably fits on a finger nail surface.
 
Last edited:

Thread Starter

ag-123

Joined Apr 28, 2017
276
It turns out for a telephone handset, it keeps coming back to this circuit
https://sound-au.com/appnotes/an010.htm


This 2 wire to 4 wire 'hybrid circuit' is intended to work such that the input (mic) would not appear at the output (speaker) of the handset. However, it requires adjusting VR1b, to 'impedance match' the line. if it is 'unbalanced' some of that input (mic) signal would appear at output (speaker) as 'sidetone'. this gives an impression one is speaking on the line as a feedback.

Without all that 'sidetone' considerations, it probably can be simplified to this?
buffer.png

This is merely using 2 Op amps as unity gain buffers, the input would be simply superimposed on the output !
'digitally' that can still be 'fixed' e.g. output = output - input
lol
 
Last edited:

DickCappels

Joined Aug 21, 2008
10,152
Back in the 1960's I experimented with transformerless telephone line interfaces. At that time out telephone office used stepper relays in conjunction with telephone dials to make connections. At that time the telephone company was fussy and even sent somebody out to snip my connection to the telephone line when they detected less than 500 uA leakage across the line. Later they switched to an electrically switched system that would continually scan the incoming line pairs for problems like that.

Lesson learned: Even though a telephone interface seems to be working fine, there may still be problems that the phone company will find for you. There is a section in the United States Code of Federal Regulations that deals with such things. If you are inside the United States you really should read what it says, especially where safety is concerned. For those in other countries, you might still want to read it so as to be aware of some design issues.

Even though your circuit is only in your house or your office, there is still a chance that a telephone employee will put his hands on a wire connected to your circuit and be put at risk. Also, there are considerations such as keeping your interface from bursting into flames in the case of a nearby lightning strike, and other such things.
 

Ya’akov

Joined Jan 27, 2019
9,070
When the market opened up for third party phones and devices in the US (previously it was simply not permitted to electrically attach anything to the PSTN aside from some PBX gear) the registration required by the FCC to make a device acceptable included a REN (Ringer Equivalence Number) specified in ”B”, where 1B is the equivalent load of a standard Western Electric DM500 phone set.

Before you could legally buy you your own phones and answering machines, the phone company could check how many extensions you had by the ring current. If you had more than they had provided you could expect a visit. When we added extensions in the 70s, they had the ringers disconnected. You could always find old phones here and there.

As a young teenager, the phone in my room was a really nice set from the 1940s I bought dirt cheap at a flea market. It would be worth quite a bit today. For a ringer I capacitively coupled an amplifier with a small speaker to the line which added nothing to the total REN load but rang just fine.

It’s very sad, to me, to think back to the glory days of the Bell System and see the current state of the last-miles copper plant. What was previouly a meticulously maintained network with high standards is now often cut loose and hanging on poles, some pedestals have been smashed (accidentally or purposely) and yet no one repairs them.

The RBOCs are intentionally ignoring the copper plant because they don’t want responsibility for it any more. Fiber to the home, and DSL (which is terminated in a SLAM, not the old pedestals) are what they concern themselves with. The voice network’s important is declining steadily. People don’t realize how incredibly reliable it was. I fear we will find out because it will be needed and be gone.
 

Thread Starter

ag-123

Joined Apr 28, 2017
276
As this is just going to be a 'single piece project', I think I'd make do with a transformer if it helps. I've got some cheap iron powder toroidal cores and I can literally make one myself, especially if it ends up with a multi secondary coil setup. The motivations of capacitive coupling is that these days modems looks like this:
https://www.amazon.com/DriverGenius-UM02-USB-2-0-Modem/dp/B00XW5QYWS
https://www.amazon.com/Robotics-56K-USB-Soft-Modem/dp/B009019KR4
Hence, I'd think it should be feasible to do capacitive coupling, as it is less likely they put a big transformer in that USB dongle.

My intended project is to use a microcontroller e.g. Raspberry Pico or STM32 or some such microcontroller and make a 'soft modem'. But that I'm really just interested in the voice part of it (for now). i.e. the ADC, DAC would convert between analog voice to data and vice versa.

What is more interesting is in my search for SLICs (subscriber line interface circuits), mainly ICs, the online chip stockists e.g. https://www.digikey.com, https://www.mouser.com, https://www.element14.com/ (this is Farnell). Many of the part numbers with a large number of stocks some > 10,000 pieces, are "old" part numbers. There are some recent offerings from Skyworks
(they acquired the infrastructure business from Silicon Labs, the Sixxxx are originally Silicon Labs part numbers)
https://www.digikey.com/en/products/detail/skyworks-solutions-inc/SI32280-A-FMR/7201384
https://www.skyworksinc.com/en/products/voice/si3228x-dual-channel-proslic/si32280
The datasheet is pretty verbose, lots of details, including the modular functions SLIC, ADC, DAC etc.
But that the interesting part is that the protocol and other interfacing details are not provided.
I'd guess the relevant app notes would be provided to large customers and perhaps under NDA.

Hence, I decided to just go discrete, it turns out suddenly I'm in an area where there is a lot of things to learn and understand.
I always thought 2 wires - telephone line - easy. The 'simple' telephone has a lot of interfacing requirements, ringing, on-hook, off-hook, DTMF, transmission, "2 wire to 4 wire" "hybrid circuits" etc. It isn't even just as simple as connect some resistors and capacitors or even for that matter transformers. Though some of these components e.g. transformers, resistors, capacitors, diacs, mic, speaker etc wired in a proper circuit forms a telephone !
It is an interesting area which is literally a 'lost art'
 
Last edited:

Thread Starter

ag-123

Joined Apr 28, 2017
276
oh it seemed I stumbled into one that apparently has 'existing stocks'
https://www.digikey.com/en/products/detail/skyworks-solutions-inc/SI3018-F-GSR/4069188
this chip seemed to have an origin that is Si3050
https://www.skyworksinc.com/-/media/SkyWorks/SL/documents/public/data-sheets/Si3050-11-18-19.pdf
and more interestingly, I stumbled into another 'relic' the evaluation kit manual
https://manualzz.com/download/52482062
while on the page prints a year of 2014, in the PDF document properties, it is:
Created 20 Nov 1997 !
Back then 'serial' port is 9-25 pins ! oh well lol

oh well, "obsolete" become "less obsolete" these days, and 'hidden' behind NDA etc for the 'newer' replacements.
The 'newer' chips mostly integrates 'everything' SLIC, ADC, DAC, control logic etc. e.g.
https://www.digikey.com/en/products/detail/skyworks-solutions-inc/SI32280-A-FMR/7201384
I'd guess between that and 'single chip modem' is just software/firmware.

oh i'd think this is also the (a) SLIC (subscriber line interface circuit) lol
for a "1997" invention, it is pretty impressive, the Si3050 datasheet reads
https://www.skyworksinc.com/-/media/SkyWorks/SL/documents/public/data-sheets/Si3050-11-18-19.pdf
The Si3050+Si3011/18/19 Voice DAA chipset provides a highly-programmable
and globally-compliant foreign exchange office (FXO) analog interface. The
solution implements Skyworks Solutions' patented isolation capacitor technology,
which eliminates the need for costly isolation transformers, relays, or opto-
isolators, while providing superior surge immunity for robust field performance.
The Voice DAA is available as a chipset, a system-side device (Si3050) paired
with a line-side device (Si3011/18/19). The Si3050 is available in a 20-pin TSSOP
or a 24-pin QFN. The Si3011/18/19 is available in a 16-pin TSSOP, a 16-pin SOIC,
or a 20-pin QFN and requires minimal external components. The Si3050
interfaces directly to standard telephony PCM interfaces.

---
edit:
but wait a minute, i'd continue to explore the "discrete" option first hand, there is something to learn there still.
And it may after all be an alternative option.
Oh btw, if you are trying to make a modem that can be used 'round the world' maybe that chip is it.
'arduino' modem? why not
lol
 
Last edited:

DickCappels

Joined Aug 21, 2008
10,152
The nice thing about the 4 wire to 2 wire hybrid interface was back in the day before everybody carried a hands-free speaker phone in their pocket or purse was that you use both hands to work on electronics while have a conversation with a friend. The bonus was that if using proper transformers (600 ohm line) you could do it without upsetting the phone company. At least I did that for many years.
 

Ya’akov

Joined Jan 27, 2019
9,070
The nice thing about the 4 wire to 2 wire hybrid interface was back in the day before everybody carried a hand-free speaker phone in their pocket or purse was that you use both hands to work on electronics while have a conversation with a friend. The bonus was that if using proper transformers (600 ohm line) you could do it without upsetting the phone company. At least I did that for many years.
I used the hybrid from inside phone sets for something very similar. They were set up for carbon microphones but it wasn’t hard to match a dynamic with with a preamp to get pretty good full duplex hands free.
 

DickCappels

Joined Aug 21, 2008
10,152
The hybrid network inside the phones was intentionally made unbalanced so that the microphone signal could be heard in the handset speaker so the user did not have the feeling of talking into a dead mic. This feature was sometimes a problem with speakerphones because it could result in a sort of echo that the speaking person would hear in the background. Fun and games - the most interesting part was figuring out a network that had a similar impedance to the telephone line so that good cancellation could be achieved.
 

Thread Starter

ag-123

Joined Apr 28, 2017
276
Thanks all, there is really a lot to learn
https://www.digikey.sg/short/qqzj7r55
I'm only beginning to understand the "jargon", ("hybrid circuits", "daa" etc) so the attempts to try a "discrete" circuit is discovering where the "problem" is.
I think this is still one of the good starting point for "beginners"
https://www.epanorama.net/links/telephone.html
https://www.epanorama.net/circuits/teleinterface.html

I'm kind of learning that "DAA" (direct access arrangement) is what i literally wanted of "SLIC" (subscriber line interface circuit).
e.g. in that Silicon Labs (Skyworks) part number Si 3050 - SI 3019. It is a 2 piece circuit. i.e. 2 different ICs.
SI3019 is probably what is called a "DAA" - i.e. it interfaces the telco (i.e. phone line directly), and it takes care of all that isolation, ringing, on/off hook, and various electrical specs requirements and probably provides the "hybrid circuit" - one of that "hybrid circuit" examples is that shown earlier with those op amps, it is one of the 'best' i'd think, it is 'easier' to understand, because the way the 2 op amps are configured is mainly to reduce "sidetone" i.e. when you are speaking on the mic it cancels out what you hear in the speaker on the handset.
if it is 'perfectly' impedance matched, i'd guess there'd be "no sidetone", i.e. you would not hear 'your own voice" in the speaker.
But that to give feedback, slight imbalance is deliberate so that you'd near some if it. Otherwise, i'd guess people will shout into the phone lol
In a 'digital' circuit, that 'sidetone' can probably be 'cancelled out' by simply subtracting it. this could simply the electronics, but place more overheads in the software processing: i.e. input = input - output. That in a way is what the 2 wire to 4 wire 'hybrid circuit' is doing. On the '2 wire' side, output (i.e. speaking on the mic) and input (what you hear on the speaker) are superimposed. i.e. for a 'digital' circuit, it can take the input (and output) directly from the 2 wire line and digitally process it.

And Si 3050 would likely actually contain the ADC, DAC and likely has other digital electronics, possibly even a microcontroller in it. Si 3050 literally interface a microcontroller directly presenting a SPI digital interface. This isn't the codec, it is just data (ADC/DAC), your microcontroller will need to play the codec.
----
edit:
As i look at more 'complicated' circuits, i'm thinking it is simplier to maybe have a transformer and zener diode limiter on the secondary,

I'd need only one set of that transformers, then use my LM358

to do the job, probably as 'easy' as it gets, lol
and oh wait, i'd need to figure out how to "pick up" the call, using a relay and resistor (600 ohm?) would probably be ok, but I'm thinking I'd try to do with a mosfet in place of the relay, it is still 'isolated' as the gate pass no currents in the loop to my interfacing circuit.
 
Last edited:

Ya’akov

Joined Jan 27, 2019
9,070
The hybrid network inside the phones was intentionally made unbalanced so that the microphone signal could be heard in the handset speaker so the user did not have the feeling of talking into a dead mic. This feature was sometimes a problem with speakerphones because it could result in a sort of echo that the speaking person would hear in the background. Fun and games - the most interesting part was figuring out a network that had a similar impedance to the telephone line so that good cancellation could be achieved.
A handset without sidetone seems dead.
 

Thread Starter

ag-123

Joined Apr 28, 2017
276
I took a look at another SLIC IC - HC5502B
https://www.digchip.com/datasheets/parts/datasheet/235/HC5502B.php
This is another chip with lots of inventories.
A good thing about this chip datasheet is the internal schematic is rather clear.
Reading that along with the application circuit, it seemed that this is more appropriate for the PBX switch connecting to the telephone. Hence, it has the ringing circuits, the line voltage and the various hook switch detector, loop current limiting etc features.
While it is possible to use it as a "hybrid circuit" many of the features would be unused for a 'phone' or 'modem' application.
But this is rather good in a sense it helps one understand what could be a likely circuit at 'the other end' (e.g. the PBX switch) rather than the phone.
 
Last edited:
Top