Any stereo encoders that can be made from components from now?

[radionortheast]

I’ve managed to make a pll transmitter from a kit, it is monoral, Are there any simple stereo encoder circuits that use components availble from now that will work, what is the deal with upsamping. I have no stand alone encoder, the one I have in pcs transmitter, the whole thing is broken, i’ve always trusted them they always use filtering from the output of the encoder, I don’t see it in circuits I look up here their also very old!. I was going to try and construct the Harry Lythall one http://sm0vpo.altervista.org/audio/stereo_enc.htm
seems like it has errors in it thought, the one from electonicschematics is missing the 4013 ic to generate the 19khz tone. I don’t know a lot about how stereo encoders work, I have alot of 4013 ics lying about here, probalbly from making some kind of clock, wondering if anyone can help. https://www.electroschematics.com/stere ... eparation/

[rigmo]

this is closest good...

[yellowbeard]

Oversampling on stereo encoders gives better signal to noise and separation amongst other things. A lot of the digital switching is done at 152, 304 or even 608KHz which is piss easy to remove with rudimentary filtering as it's so much higher than the frequencies you want. YO4HFU has some good info, diagrams, PCBs etc. here: (He is not shy about using a £30 component here and there - be warned!)
https://www.qsl.net/yo4hfu/Stereo_coder_sampling.html
Albert put a basic stereo encoder here:
viewtopic.php?t=2757
All the parts are still readily available. If you are looking for something ready made, the SPI encoder is similar in principle to Albert's except it's ready made, tiny and €18 yoyos out of Holland:
https://display-elektronica.nl/Webwinke ... udget.html
Right, spliff and bedtime!

[Albert H]

Yes! Stereo coders are actually not too complex. You need to understand what makes up the stereo multiplex signal, and then you can understand - easily - how a coder works.

The Zenith-GE stereo FM system (developed in the early 60s) is the one that everyone uses. It was designed to maintain compatibility with mono receivers, so has a couple of compromises. There are three basic parts of a stereo multiplex signal:

• The Mono signal which is Left + Right
• The Difference Signal which is Left - Right
• The "Pilot" Tone - a 19kHz sinewave tone that tells the decoder in the receiver that it's a stereo broadcast, and provides part of the signal for recovery of the separate Left and Right components.

So - the stereo encoder takes the audio and generates these three parts of the "Stereo Multiplex".

Firstly, the audio is filtered to remove any content above 15kHz. This is to prevent interference with the 19 kHz Pilot at to prevent "aliasing" distortion, which gives rise to all sorts of non-musically related harmonics, and usually sounds horrible.

• Pre-emphasis needs to be added to both channels - basically a "top boost" that has a rising characteristic from ~3 kHz to 15 kHz at 6dB/octave (the UK and Europe uses the 50µs curve):
  
  FM Pre-emphasis Curve.jpg
• The Sum (Left + Right) and Difference (Left - Right) signals need to be created. Sum is usually called "M" (for Mono), and is the part of the signal that a mono receiver will detect. It's simply produced by mixing equal parts of Left and Right.
• The Difference (Left - Right) is usually called "S" (for Stereo) and is used later to recover the separate Left and Right signals. It's produced by inverting the Right and then mixing it with a equal amount of Left.
• The Pilot tone is generated by an oscillator. This needs to be very frequency-stable, and very pure (a good sinewave). In most serious stereo encoders, it's derived from a crystal reference oscillator, to make certain that it's accurate.
• The next bit is the (slightly) tricky part: The S signal needs to be converted to a "Double-Sideband Suppressed Carrier" signal centred on 38kHz. DSSC is an AM modulation mode, created by mixing the S signal with a 38kHz sub-carrier. The 38kHz sub-carrier has to be in phase with the 19kHz Pilot tone (to prevent decoding errors and distortions at the receiver). It's usually derived from the same crystal oscillator that's divided down for the 19kHz Pilot, to ensure phase coherence.
• Finally, the three signals - the M, the DSSC version of the S and the pilot are added together to give the composite multiplex signal that's fed to the transmitter. The levels are set up so that the 19 kHz tone (the Pilot) gives 10% deviation (7.5kHz), and both the M and S signals are allowed to give 90% deviation.

The whole composite looks like this:

FM Stereo Baseband Spectrum.jpg

There are a number of ways to generate the three component parts of the stereo multiplex signal - the earliest used op-amps to sum and difference the audio, a balanced modulator with a 38kHz sinewave carrier to derive the S DSSC signal, and a 19kHz oscillator for the pilot with a frequency doubler for the 38kHz sub-carrier, and then a further op-amp mixer to add the three parts together in the right proportions.

At the receiver, the stereo decoder is (usually) much simpler. It has a tone decoder that detects the presence of the 19 kHz pilot tone which switches on the "stereo" indicator on the radio, and enables the decoder. The 19 kHz is received, doubled in frequency and then mixed with the DSSC "S" component to put back the "suppressed" sub-carrier. This allows a simple AM detector to recover that difference audio. Then the decoder does a bit of electronic algebra to recover the separate left and right audio:

At the encoder: L + R = M and L - R = S

At the decoder: M + S = 2L and M - S = 2R

.... so we've separated the L and R audio signals we started with! The last thing to do at the receiver is to deemphasise the audio, to remove the "top boost" and recover the audio as it went into the encoder at the transmitter end.

The reason for the pre- and de-emphasis is because the predominant background "hiss" on an FM signal is more in the treble end of the frequency range, so "brightening" the audio at the send end, and reducing the treble to normal at the receiver will make the "hiss" less obvious.

The "quick 'n' dirty" way of generating a stereo multiplex is the one most often used by pirates. It was noticed that if you alternately send a brief sample of the left and then of the right, rapidly switching at 38kHz, along with an in-phase 19kHz sinewave, the receiver is fooled into "seeing" a stereo multiplex signal (because all the components are there). There are a few drawbacks with this approach, including the risk of lots of harmonic products if you don't filter correctly, and the fact that the "S" component is pi/4 times too loud (which can also be remedied by filtering), but it's usually the most cost-effective way of doing it. The simple coder I gave in another thread uses this switching principle, and gives perfectly reasonable results.

The "switching" type of coder can be further enhanced by the use of "oversampling" which reduces audio distortion and harmonic products, and is the approach used by most modern commercial encoders!

The best cheap switching coder was (probably) the NRG PRO IV, which I designed, but unfortunately isn't available any more. The 15 kHz lowpass filter modules we used are no longer manufactured either (unfortunately), but I can provide really good circuits for filters that can be built using op-amps, resistors and capacitors. I use the "gyrator" approach, to emulate inductors.

Circuits to follow if anyone wants to give them a go!

[LeeCavanagh]

Why do you not want to use the software mpx encoders you cqn run from a computer ?they sound great too
My fave is breakaway one, there are others tho, invluding sone free ones, but they all at least offer free trials.

[Albert H]

I have tried ALL the available software coders, and NONE of them works as well as even a simple, cheap switching coder. They all have problems - whether it's harmonic crap making the bandwidth of a transmission too wide, or it's bizarre audio artifacts caused by erroneous assumptions about audio processing, or it's simply nasty digital distortion.... No matter how expensive the software, and no matter what hardware it's run on, none of them works well enough for serious transmission.

There is one software solution that comes close, but its developer has given up on it (he finds that he needs to concentrate on paying work!). It's written to run on quite minimal hardware, but still needs a lot of work....

Building a stereo coder - the hardware way - is quite easy. The physical layout isn't as critical as it is with RF stuff (since none of the frequencies are above ultrasonic), and if good basic construction practice is followed, it's quite easy to build a coder that works well.

I've recently been working on a design that uses a few high quality op-amps for the 15kHz lowpass filtering, another couple for the summing and subtraction (to get the difference) of the audio, an MC1496 balanced mixer IC for the DSSC generation, and a bit of CMOS logic (and a 4864kHz crystal) for the pilot and sub-carrier generation. I've got a Veroboard prototype that works remarkably well, and I'm designing a nice, neat single-sided PCB to make it easily reproduced. I plan to offer the boards at cost price (along with the 4864kHz crystal) so that anyone can build one easily. The only caveat is that it might need an oscilloscope for accurate final alignment.....

[LeeCavanagh]

I have to disagree about the wide panning of software based, a lot of those sold in racks are software based anyway

Are you saying that id be better turning off the mpx and putting its audio into a dedicated mpx encoder/.? What can that do that a decent soundcard with 192khz output? (Oversamply maybe, noise reduction not so much as doftware slliw that to be confugured to reduce noise, you can ajust pilot tine, adust the rds etc, stereo seperation about the same when set decent,

I have the fully paid version of Breakaway One (by engineer for omnia) and love it, i also tried in the past purchased from john burnhill (innovonics), i dont like stereotool tho. Ive also heard some ok results from bw-dspx (but nothing my dbmax at the time couldnt do)

[Albert H]

The BW DSPX is truly horrible. A friend of mine had one - bought at full price from Roger and Scott at the NAB trade show some years ago - and he was never able to get anything approaching acceptable results from it. He eventually stripped it down, and used the 19" case for a homebrew version of the Inovonics David III!

The software approach that I saw that was pretty good used a 192 kHz audio card (as a minimum) and if carefully configured could sound (and measure) really top-drawer. Unfortunately, its developer got so far and abandoned it. I have the Source Code here, and was considering using it as the basis for my own software processor / stereo coder, but it's mainly a question of finding time to work on it....

Breakaway One was always rather disappointing to me. It's a laudable attempt, but still doesn't win against my hardware approach (take a look at the Inovonics 718 David III). The Inovonics use of split-band audio processing, PWM attenuators, and carefully tailored sidechain responses, together with accurately generated digital subcarrier and pilot tones with analogue filtering leave all the competition for dead. It costs about one tenth of the price of much of the competition, and out-performs all of them! It's the first processor / coder combination that I've worked on that sounds good with all kinds of programme material without adjustment. It's also a doddle to install and set up.

[LeeCavanagh]

You’re clearly biased
Anoraks have budgets to consider

[mikroman]

With due respect to Albert, because he brings good arguments and certainly has great knowledge, when it comes to the mentioned machines, I have a completely different opinion. In my area there are two local stations. One drives David and although perceptually it sounds louder than the competition (but the deviation is above 90KHz in this particular case ), the sound is muffled, and the music productions from different decades and musical styles are non-uniform. Maybe the setup is messed up, maybe the caps are ready to be replaced, but it sounds worse than the local neighbor who uses DSPXtra. One of the few people in Croatia who deal with this field (with longer experience in broadcast) adjusted it this year and it sounds good regardless of the program content (unless something with a low bitrate is added to the playlist).

The Breakaway one is an interesting toy, but you must spend a lot of time if you want it to sound good. No distortion-free dry vocals can be obtained with any LOUD generic preset. Also, the loudest ones have a problem with HF limitation, which is very audible. Each sibilance processing ends with a short-term suppression of the HF band. There is also an audible distortion of the musical material (solo piano, guitar), overemphasized brilliance and an audible input of noise. In short, none of the offered presets impressed me.

After approx. 20 working hours spent in the simulator using different sound materials and listening equipment, I made a preset in which I sacrificed approx. 1.5dBr in order to reduce mentioned lacks to the limit of inaudibility by ear. Result: the sound is perceptually very little quieter than its competitors, but it is pleasant to the ear. The best indicator that you did a good job are listeners who have no idea what is in the way of their radio receiver and praise good sound.
I can PM stream link.

[Albert H]

Of course I'm biased! I designed a couple of the best products on the market. I spent forever trying to get a DSPX to sound good, and no matter what I tried, it still sounded nasty. I took it back to Carshalton for Roger and Scott to check, and they said that it was working correctly. If that was "correct operation", the basic design criteria were wrong!

Which version of the "David" have you heard? I don't much like the latest version (IV) - the "all digital" one. The III was exceptionally good, and the II was even better (but some of the components became unobtainable). The III is an effective mix of analogue filtering and digitally-driven stereo coding. The PWM-based AVC and multi-band limiter eliminate the need for accurate component matching channel-to-channel, and if adjusted carefully, it can sound fabulous. I don't like the "high density" settings favoured by some stations.

I had a complaint of "splashy" audio from a David III that was being used by an Italian station. I visited them, and quickly realised that they'd enabled 75µs pre-emphasis (should have been 50µs) and their rig didn't have a flat audio response either! Correcting the settings, and adding a couple of capacitors to the audio path to their modulator completely cured their problem.

Unfortunately, some commercial FM exciters have a less than ideal modulation characteristic, and no matter how good the audio processing and the stereo coding is, they just can't be made to sound good!

[LeeCavanagh]

Don’t panic, I know

[mpx]

I have tried both hardware and software encoders. I found one of the best hardware ones to be the fairly low-budget NRG Kits coder. Good separation and clarity. You need an overshoot compensated clipper after it though, to get any reasonable loudness. And if you're running RDS with it, pay attention to the pilot and subcarrier phase.

Software-wise, Airomate can give a good sound, but the clippers leave a lot to be desired. Drive any more than about 2dB and you start to hear some intermod distortion. Breakaway does a reasonable job, but as has been said, there can be some issues with controlling sibilance at the cost of HF clarity, sometimes punching holes in the spectral balance. I'd recommend putting a parametric EQ before Breakaway.
Stereo Tool has a very good go at it, and the advanced clipper allows for more loudness if that's your thing.

There are some better options available but it comes down to budget - and the expensive options aren't always the best ones, I've seen some shockingly terrible performance from some of the well-known commercial brands!

Just my two pennies worth

[Albert H]

MPX - which NRG coder did you like? Was it the Pro IV? That was my design! I'd be interested in how you implemented your compensated clipper - it could make for a "bolt-on" for the next iteration of that coder.

[radionortheast]

I might have a go at making alberts encoder sometime, it is already on stripboard, I will have to work out the resistor color codes, the 4.86 cystal is expensive, it can be got from engima, they do seem to deliver fast. There are at least less parts than trying to get a transmitter together, if I was to get a transmitter board get the parts its something I wouldn’t beble to get working at all, a kit is something I often can, I do have most of the parts apart from afew of the ics.

I able to encode stereo using windows using stereo tool, it took me ages to do it, it was only when trying some other software I found out you need the virtual audio cable to make it work with winamp. Alot of my music is on another computer with a different os, it would be a pain to move it over remake playlists when I want to hear some music over a transmitter, having something that plugs in is easier.
I don’t think stereo tool sounds bad, it can sound ok as audio limiter, i’ve used the standalone version in the past, I found I end up turning most things off, your left with something which is hardly doing anything. You do also get this sound is processed by stereo tool now more often, i’ve never bothered with rds that much, it also defaults the rds back to stereo tool, it dosen’t get the song title, jmpx the rds will just keep whatever you write, also get the song title. I suppose stereo tool is more commercial type of software, were the owners are trying to make money, I do think it is over engineered.
I’m thinking of going back to hardware limiters, maybe looking them up future, I used to have an nrg limiter and an nrg pll, would relay dab stations, it would sound really clear. I see the 4.86mhz crystal is mentioned again, I suppose that means it must be quite hard to get other kinds of crystals working.

[Albert H]

Northeast: Send me a PM so I can get in touch with you - I have some 4864kHz crystals - I'll be happy to post you one.

That simple coder just needs two squarewave signals:

• 19kHz squarewave that gets put through a filter to make it into a (pretty good) sinewave for the Pilot Tone.
• 76kHz squarewave in phase coherence with the 19kHz - this squarewave toggles one of the bistables in the 4013 giving two squarewaves at 38 kHz in phase opposition to each other to drive the switches in the 4016 (or 4066).

Another popular way to derive these signals was to use a 455kHz ceramic resonator, lightly loaded to oscillate at 456kHz. This is then divided by 6 (I used a 4018) to give 76kHz, and then by two cascaded 4013 stages to give the pair of 38kHz and the 19kHz. You need to add an extra 4013 for the synchronised monostable that I used to guarantee that the coder knew its left from its right, and not come on with random channels (like the A***F hunk-'o'-junk). The 455kHz ceramic resonator won't be quite as accurate or stable as the crystal, but is good enough.

[radionortheast]

wow thanks, unforuately i’ve already ordered the crystal from them, it took a while to get here it arrived today, i’m sure i’ve seen the 455khz resonator in old radios thanks for the explanation, I think i’m understanding abit more that the frequency is been devided, I will start with the 4060 it seems the easyiest to start with