Hi welcome to Tear down Tuesday Got something interesting for you today. It's a PA amplifier From A Train Not a toy train like a real train here in Sydney People in Sydney may notice the City rail or state rail or whatever it's bloody well called. These Days Um, simple. These are used in our trains.

You've heard them before. If you've been on Sydney trains, you know next St Central All stations the warboard so that's you. Know that muffled sound it comes from one of these amplifiers should be interesting. Lots of high power stuff.

It's got to drive. You know, a whole bunch of uh speakers on the train. So we'll look into the design aspects of it as well and we'll get some really good design. Insight from this.

think we're going to like this one. So you know what we say here on the EV blog. Don't turn it on. Take it apart now.

normally. um, it would come with the cover and would take it apart, but the cover's already off this thing now. Normally, if you're really lucky, you'll find a schematic inside one of your products. but I found something hell of a lot better in here.

We've got the original designer, Taada. It's Doug Doug Ford Hey, Doug you, you designed this. Beast I Designed this beasty back in 1992. so these are now 20 years old and still rattling around in Sydney trains literally.

I Believe So, and to the best of my understanding, there are between 1,200 and 1,500 of these being manufactured. In fact, the most recent batch that was manufactured was an additional 50 units about 2 or 3 years ago. They're still made. They're still made, they're still in use and still pretty damn reliable.

And you can tell us all about the design of this Beasty Cuz you Des Designed it from scratch? Yep, except we ran out of scratch and had to use metal work instead. Three Rack unit Three Rack unit High Yep. 19in standard rack mount Mhm. You can probably tell by the front panel over here that it's got two distinct amplifier sections in there, one for public address and one for crew intercom and emergency.

And what we might do is backtrack a little bit to the specifications of why we had to design such a complicated Beasty. For something that's just a power amplifier, It just yeah. Well, it's a little bit more than just a power amplifier. It's got a few extra Thrills in there and a few criteria that had to be met in the design.

Excellent. So let's go back in time. Way back. Okay, first of all, let's kick off with uh, how trains operate in Sydney Australia Different countries have different standards.

We have heavy rail by the way. we are one of the few people don't know. Sydney is one of the few cities in the world that actually runs heavy rail directly into the city. instead of a what do you call it one of those Metro Services You know, the quicker, smaller, lighter trains, we have heavy rail anyway.

I'm sure you can look that up on Wikipedia it's a lovely drawing. Doug Oh yes, indeed. terribly accurate. Dreadfully accurate.

Okay, our train system runs from 1500 volt DC overhead lines. Our silver trains are generally configured so that we get a motor car with so these are the silver. the traditional Silver trains which went from what vintage are they I Don't know. late' 70s, 70s onwards late 70s onwards Are the Tang new Tangaras all the same? all the New Millennium train.

These amplifiers don't go into the Tangaras or the Malinium right trains or any of those. um, they called them the memons because the the trains were lemons because they when they went in the tunnels tunnels had a significant gradient on the track I believe. and then they draw draw drew too much current and they couldn't get up the they weren't spect right? So anyway, that was a bit of a fire on part of City rail or whatever they call themselves. Anyway, these are the silver trains all right.

We have everyone in Sydney is familiar with those uh what what it call pantex that pick up the 1500 vol uh and in each motorc car set we've got uh, big inverter motor drivers which and also circuitry which drives big 120 volt battery. Banks Now for every motor car set there's also a trailer car. Then we'll get usually another trailer car and then another motor car set with its pant and drive, and Battery set, etc. etc.

And then that could be twice as long or three times as long. Yes, either. Two car sets, Four car sets, Six car sets, eight car sets. But they're always in peirs of a motor car and a trailer car.

Got it? Okay, they're all cross link by rather large connectors which circulate things such as uh commands from the drive let's say, the train going that way. So we've got the driver in there. It's going to transmit motor speed controls from that motor inverter down to this one down here, and etc, etc. Uh, there are public address signals that are circulated amongst the train.

There are intercom signals so that a driver here can talk to a uh, a guard who might be resident in the guard's compartment. Down here, there's quite a lot of signals I Think that there's some, uh, 20 or more wires Within These Rather large connectors. Are they used for existing other stuff? You don't have to piggyback signals on them or anything like that. Yes, there are now, right.

I've opened a can of worms have I Oh yes, indeed, because long ago, they ran out of conductors in the wires. Particularly now that they also have to transmit things like door open and close right? commands. Yep. H There's also now a lot of data signaling uh from destination signs up the front and down the back and etc etc.

There's quite a lot of data superimposition. And indeed, even back in '92 that's where this amplifier came in. Now we'll erase that and kind of zoom in a little bit to what this amplifier has to do. Now we'll zoom into, uh, a plan view of a driver's car set with a driver's compartment up here and next car down here and the next car down here.

and the next car down here with a say, a guard's compartment here. MH Okay, driver wants to talk to guard. Well, we need an amplifier here so that the driver can address the speaker down to the guard's compartment. Is it just speakers? There's no.

Are there headphones? Are there Speakers only? Speakers only? Yep. Ceiling speakers. Got it. We also so we have a microphone here running into a power outet fire driving his speaker.

Mhm. He wants to reply, Guess what? He needs A microphone driving into a power amplifier? Mhm. Back up the same lines to his. How many lines are we talking about here? A pair to do that? Single pair? Single pair? Yep.

Now, the thing is, it's entirely possible that if they get into a bit of an argument, they're both going to be pushing their talk buttons at the same time. Oh no, they're both going to be talking into each other. And basically the amplifiers have to be designed to not only feed dead shorts, but also feed into the outputs of other amplifiers without self-destructing. That's part of the criteria.

Uh, another part of the criteria is okay. These amplifiers are run from the 120 volt battery. Banks Previously mentioned, the voltage on those battery Banks can vary from um. well, well.

on a bad day when they're near Stony flat, they might be down at 70 volts Mhm. If they're fully charged to the point of overcharge, they might be up at 150 volts. It's a pretty fair what sort of Battery Technology Are we talking about Lead acid? Lead acid? Big Lead acid? But just to add insult to injury, there's rather a lot of different electromechanical bits of Machinery run from these 120 volt batteries. They're not only used for the PA system, they're used for plethora of other things.

They connected to inverters so that they can run the air conditioning systems so there'd be lots of dips and brown outs and all sorts of crap on dips. There's brown outs. Uh, they actually describe the spikes that they see on these lines which extend from you minor overvoltages of maybe 200 Volts for hundreds of milliseconds down to 4 kilovolts for some Microsc and state rail actually have a list of spikes that they have obser so that and you have to design the amplifiers to accommodate. Do they have Spike simulators to test that? So when you design gear like this to you, test it to their standard they didn't.

We had to create. Oh, you had to create. the generators to do that. y There you go.

Which actually is disturbingly easy. How do you do it? Pass a current through an inductor and open circuit? The inductor capacitor couple the resultant Spike energy onto your main Rail and and ble. It's disturbingly easy to simulate. So this is a half duplex push to talk system If I've got the terminology correct.

Yes, Right now on the this is for the intercom system. We haven't addressed the PA system yet. The intercom system can use relatively low power amplifiers and in this case, we limiting the power here to 30 35 Wats Mhm. The lines connecting all of this run out nominal 100 volt levels y so it's fairly high voltage level.

Very few transmission losses. Uh, but one of the systems that was proposed is the use of Uh passenger help points. That's right, such that in each Uh in each passenger location, you would have a help point. We you can push the button and talk to crew so that effectively overrides the crew intercom system.

Yeah, now, how does it do that? Does it give priority over the crew talking to each other at the moment? It does. Ah, it does. There you go now. How does it do this When a passenger pushes their emergency button as well as connecting their little local amplifier onto that indcom line and squating audio which will be heard by both of these speakers? It also injects a 25 khz tone.

5 khz ton is detected by these power amplifiers. These PES Here, when they detect a 25k tone, they lock out other activity. Got it so that only the uh, the passenger in in dire straight is heard excellent now. Similarly, how do they reply to just him? and not to that one And that one, And that one, And that one, and that one cuz you don't panic the rest of the train? Exactly? m Okay, these when the crew want to reply just to that help: Point M They press their passenger emergency intercom button and they're putting audio back down that line plus a 50 khz tone.

Ah, the 50 khz tone is detected by only the one who's had its button recently pushed right. thus two-way Communications Recently, push has it got a Timeout on that? Yes, right? Got it? Absolutely. Yeah. The logic of all of this was fairly carefully thk through, right? Uh, And not only that, they can't hold their button down for any more than 30 seconds, it times's out.

Similarly, they can't hold their button down for more than 30 seconds. It times out. So if you get someone who likes to have a good gab on your just, you're out of here kill switch. They flicked off the queue.

Okay now public address. Okay, all through here. We've got loud speakers spread right throughout the carriages. Now these are double deck carriages.

So you've got uh, you, maybe 10 or 20 or 30 speakers down the bottom. These the ones you can't hear speakers? Yeah, they're the ones that you can't hear and there's there's a reason for that which I'll actually come to we will go into. Oh yeah, Oh yeah, everyone wants to know why they're crap. so we'll find out.

Folks, stick around the amplifier section used for public address. particularly because it might be addressing the speakers in eight carriages sometimes up to 16 carriages. much higher power: 150 WS right? Separate set of lines reticulated right through. Again, 100 volt stuff.

But uh, each of the each of these little 4-in speakers is I Think tapped for about a half a watt or one watt load onto that line. So uh, if let's say they tapped at half a watt, it means that you can put 300 speakers onto that 150 W amplifier, right? and the amplifier will treat that with perfect the plum. y Uh, once you get to maybe 400, uh, sorry. once you get to maybe 200 WS worth of load on the amplifier, it's VI limiting.

I'll come to that, we'll start kicking in. Uh, but you can talk to an awful lot of speakers Awesome. And half a watt into a little 4in speaker still represents a fair amount of sound pressure level in The Carriage Particularly when you've got a number of such speakers multiple spread throughout. It' be enough.

Yeah, okay, what kinds of problems do we have with PA systems? even on nominally working trains? MH There's two that I'm distinctly aware of. The first is that between some of the Uh, middle-aged and extremely old train, there were some compatibility problems in the links that couple the carriages together. One of those areas of incompatibility was perfectly capable of connecting 120 volts DC straight onto the PA loud speaker lines. Ouch Ouch! Indeed, you don't want to burn your coils out.

Well, what it burned out was the all of the Uh, the coupling Transformers at the individual speakers until they started fitting those speakers with coupling capacitors so that what you wind up having is a little coupling cap. Okay, this is your 100v rail running into a Step Down Transformer Transformer which drives your speaker into the speaker and you would pick the correct tapping on the Transformer to give you nominally half a watt into that speaker from 100 volts there. Okay, so that was number one problem. Incidentally, in the earliest amplifiers, we were not made aware of this particular problem as a result of which.

uh, you'll see in a minute there's a couple of toroidal Transformers under here. Mhm, the larger of the Transformers would get 120 volts straight up it. It was sufficiently large that it would actually take quite some time to die. maybe tens of seconds.

right? During that period of tens of seconds before it ruptured, there would be a lot of smoke emitted. Uh, it was after that first batch of amplifiers went into production that we were made aware of this. So he started on those first ones: fitting extra coupling capacitors on the output of the Transformers so that DC wouldn't affect the Transformers Later versions such as this one had coupling capacitors Incorporated on the PCB got it And note incidentally that they're electrolytics. They are.

they are electrolytics. but they're connected back to back, right? Yes, so it doesn't matter which way around you put your DC on it during a Fa condition, it'll still hack it. Okay, we're having a look at the back panel of this amplifier. Uh, they're all Uh XLR style connectors power.

The output connector actually needs six connections. I'll come back to that shortly. We're missing a three pin connector there for muting, which actually never got used. It was simply a provision that was put in for later.

Uh, this particular unit that I've got my hands on here at the moment is actually a production sample, right? Which is because let's face it, all of the working ones are rattling around in trains or sitting in Depot waiting for something to break down. Okay, we've got an input here which allows Uh program input such as maybe radio or pre-recorded messages into the intercom. We have two inputs here which allow similar Uh program input, pre-recorded whatever music annunciations to be fed into the PA system. We also have a guard's microphone connection and a driver's microphone connection.

Mhm. Okay, Uh, I Should at this stage, maybe go back to the Whiteboard and discuss the overall topology of the Anif. Possibly I'm interested in this uh heat sink. It's a rather unusual riveted arrangement with a what a stabilizing bar on the back.

did they vibrate loose or what? Uh, Well, at the stage that these amplifiers were designed, the company who I was working for Jans Jans Electronics uh were very good at sheet metal and had excellent sheet metal uh facilities uh, and it was viewed at the time to be expedient to design the heat sinks uh from sheet rather than Extrusion or any other form. So we designed them out of Uh stacked U sections mhm with this bar across the back so that it would made it more difficult to bend an individual fin out of shape. Now also, bear in mind these had to undergo almost military specification testing up to an including uh I think it was 24 hours on a Shaker table Mhm and three week swept through what range? What's the range of vibration on a train Cuz I Know there are standard specifications for vibration for Road, air, rail travel, and in international vibration equipment standards. You're pressing my memory here, but I think it was of the order 5 or 10 Herz through to about 500.

HZ Yep, it sounds right and if you give me long enough I'll remember what the actual amplitude levels were because they varied with frequency too. Yep, and so that that would be a swept sign test. Yes, from through through that range. just continuous sweep in once a 10 seconds or something? Yes, Yep.

and uh. There was a an additional pretest where they actually went through searching for resonances mhm that were visible to the naked eye. Yep, Was it only tested in one axis? The mounted the axes that was mounted on. Was it tested in all three axes? Seriously? I can't remember.

it was 20 years ago. It was 20 years ago. Literally. Okay, we're in.

what we've got down. There is the two Tal output Transformers the power amplifiers. Since they run from a ninal 120 volt DC Supply rail, we designed them simply to be a pushpull. Class A amplifier run directly from 120 volt which gives us uh RMS output voltage before clip of about 35 volts.

So these Transformers are 35v to 100v. Step up Transformers in 30 odd watt size and 150 watt size and that's the one that the smoke comes out of in large quantities when you put 120 Vol DC into the 100v AC side. Oops What? and lots of very very solid 1.2 mm uh, punch steel. All good, solid stuff and we can have a look at this.

Okay, pretty simple wiring. looms, bucket loads of T3 output transistors also used for voltage regulation. What type are they Okay, these are Mj15024 and 15025 Npn PNP pairs rated Uh 250 Vols 16 amps. that's a lot of can cancer Compound on there.

Uh, well actually this is just just post the cancer compound ER So these are actually zinc oxide you go rather than burum oxide. Yep, Uh, I was rather Disturbed at the stage that the zinc oxide uh, thermal compounds came out. to have a look at the MSDS sheets and discover that the Ld50 in other words, the dose required to kill 50% of typical human beings was in excess of 50 kg of that stuff. Now we had 2 kilogram tubs of this and I just had this nasty mental image of sitting down to a meal of 25 tubs of Sil silicone thermal grease.

uh I I I think I'd die of repugnance before I actually died of thermal grease poisoning. So yeah, this is, uh, the non toxic stuff. y uh I don't think I've ever actually come across any true burum stuff. however I have as a teenager being up to my elbows in polychlorinated Bip fennels out of power factor correction capacitors.

Maybe that goes way to explaining some aspects of my behavior now. Okay, this is real good oldfashioned. Earth Technology Uh, we've got opamps down here which are ne5532 now in 92 to theop Amps had been out probably maybe up to a decade I Can't remember when the 5532 first came out, but they look. Oddly enough, they're still pretty much considered to be the duck guts.

With regard to Uh audio opamps, there are only a few opamps that are considered Superior to them. Mhm. So for an old and venerable chip, they're doing all right. Um, we've got some, uh, a couple of Seos switches over here for routing signals to different places.

MH What's missing here and here are God help us a couple of gate array logic chips. Oh, you didn't You didn't design in a gal. We designed in a couple of gals for the logic functions. Why on Earth did you do that? This is 92, remember? Yeah, Okay, we didn't want to use micros because particularly back then, uh, micros.

Unless you knew that they were going to be regularly rebooted, there was always the risk that they get, uh, just sent off into a loop or miss behave in some fashion. Yep, unrecoverable exactly. Now the alternative would have been to use quite a smearing of 4,000 series Seos chips, which with the benefit of hindsight, would have been a vastly Superior Solution because the logic combinations we got down here really aren't that difficult. No, I wouldn't think so cuz you can't do much in a gal.

Ultimately, save might save five or 10 chips or something. we probably we were probably saving 5 or 10. 4,000 Ser Seos chips. Those gals Pro to be the bane of our lives in later variants because the primarily the input pins started off having very, very weak pull-ups Mhm about I think it was one between 1 and 5 microamp then they went to to you know, the B and C versions which had maybe 10 to 20 microamps, 20 to 50 microamps.

The most recent ones that we got had more than 200 microamp pullup ouch. and this completely stuffed some of the uh, some of the analog cheating that we doing around the Gap Ah, cuz you can fly to the moon on a couple under microamps. Yeah, yeah, yeah. so this is simply the change in Uh manufacturing technology over the over the last 20 years has made huge differences to the strength of the input pin internal pull-ups Got it? Yeah, they've also become faster.

but let's face it, this is dumb switching of mic signals. We can't tell the difference between 50 nond or 20 nond performance. Uh, kill it like it is. Doug Okay, what else have we got on here? Okay, this is probably where we might want to have a look at First of all, a system architecture drawing.

Let's do the topology and then A We'll go to a circuit diagram. We have circuit diagrams. We have circuit diagrams Beautiful. 30 W amplifier going through a Transformer Okay, in whenever the amplifier is unpowered, what we want is for the uh, let's call that the the circulating Uh intercom line throughout the train.

Mhm We want that to be switch to uh, I'll just draw a a relay here. In the off: State We want it to be switched here to the local speaker. Okay, Uh, Doug's had a 20year brain fart folks. Okay, so we have here a local speaker in the crew compartment.

Uh, we we actually need two relays here. Uh, I'll just draw them in place and figure out what they do later. Doesn't have to be accurate. We can have stick figures.

Yeah, this will be sticky. This will be very, very sticky. In the off date, we want that local uh sorry, that circulated Enom line to go nowhere. Mhm.

We want that speaker to be dead when it's powered on. We want that to switch down and allow that speaker to be connected to the intercom line so that if, uh, if there's somebody occupying it, occupying the the crew compartment so that Reay is energized, any audio coming through on the intercom line gets heard. and then of course if we hit a push to talk switch on a microphone, we engage that one. So we send the amplifier output onto the Indcom line.

Yeah, so far so good. What have we got over here? We've got a guard's microphone. We've got a driver's microphone. He has three switches associated with him.

okay, guard and driver. He also has three switches associated with him because the guard sits in a different place to the driver. Yep, even when the driver's up the front, he's in the driver's seat. When the guard's down the back, he's in the little guard compartment.

Mhm. The driver's compartment And The guard's compartment might be next to each other. They're still got their own microphones and their own places. Got it? Okay, so he could address PA Indicom or passenger emergency Inom the driver too can address PA Inacom or passenger emergency Inom.

Okay, these are just, uh, switch signals from a box. Now, this is the other aspect about poor quality announcements. The microphones for the guard and the driver are dynamic microphone capsules embedded behind perforated steel mesh. right? Okay, so picture.

We've got these little microphone uh capsules with diaphragm there mounted behind stainless steel with a bunch of holes there. Okay, when you're uh, talking into that, there's a fair amount of low pass filtering applied to the speech by those holes. Oh, okay, due to the diameter of the based on the diameter and the pattern and the yeah, now why are we doing that to try and make them well, not Vandal Proof There's no such thing. but Vandal Resistant Mhm.

Uh, we have been told anecdotally and I'm still not certain if this is true or not. and I don't want to offend too many train drivers and guards. We've been told that some crew are they've got time on their hands, they're bored, they're irritable, and if they can break something on the train then they don't have to use that something on the train. So they've actually almost got a V vested interest in trying to make things not work.

M So, and also, there's the possibility that if members of the public get into these compartments, they can reap damage. So we made these as bullet proof as we could, but at the expense of frequency response. So the frequency response of the mic capsules might once have been kind of vaguely flat, but now look more like that at what sort of frequency were we talking about? Roll off. Couple hundred Herz Couple hundred Herz Roll off.

So the human speech goes up to 3 khz nominal. Yeah, yeah, And indeed, the response. The basically response. The basic response of the power amplifiers here are 100 HZ to 15 khz.

MH So the power amps are quite wide band, and indeed, the various other inputs that I showed you here for program material they are unshaped. They're allowed to be flat up to 15 khz, and they're capable of pretty high quality reproduction. So In theory, they could play music on the train no problems at all, could and you'd hear that. but then you'd get next stop.

Central All stations to whoop Whoop Just so. so. This is why the internal preamps down here have shaped response with a frequency response like that to correct for the deficits here. So in Principle as long as the person speaking is within about maybe 200 m of their panel mhm, you should be perfectly capable of getting quite good enunciation.

Quite good Clarity over the PA speakers. Now back in 92 through to about 96 I Understand that State Rail was actually getting their guards and their drivers and putting them through a form of elocution lessons. Seriously, they apparently had little booth set up where they would, uh, record the person speaking, play it back to them for correction and Amendment and I believe at that time Uh, Speech Quality and comprehension were really very good and it has slipped and slipped and slipped since then. It's not much wonder that uh, because there's not a lot of attention paid to this side of things at the moment that you know the the crew don't necessarily know that they have to speak into the the Dr panel instead of speaking off to one side.

M Uh, they're not. so they got the head half hanging out the door while they're trying to make the announcement. Yep, uh, Interestingly, on some of the diesel car sets which are which run into State and out of the country, the announcements are made with a handset or handheld microphone and results apparently are much clearer, right? because of the fact that they're using a close mic technique. I Think that the crews on those runs are considered to be, you know, A Cut Above and you therefore don't have to Vandal Proof It quite as much, right? What's what's the truth of that? I Don't know.

Seriously, don't know. The mic preamps have frequency response shaping. Uh, we've got all of these various switches so that the mic, the output of the mic preamp can be gated through to the endicom amplifier or to the the PA amplifier to be sent out to the global PA line. Uh, so that switch will kind of run that one, that electronic switch there.

that one will run that one and if they are using that one, it not only Gates that through, but it also gets the local uh, 25 khz oscillator and Gates it through to and they that's just summed onto the signal and summed onto the signal. That's it. Okay, you have to do that a few times. Uh, yeah.

but you said the bandwidth of this was only 15 khz? Yeah, uh. for for speech for speech? Yeah right. Uh, the actual bandwidth of the power amplifier itself is intrinsically much higher. We do the limiting back in the prey side of things.

Uh yeah uh. we're actually pushing sorry 50 k Herz out here. And uh, we figured in the system we'd get these ones to torque at 50 khz because we've got enough power here and enough grunt to do those relatively high frequencies well, right? The passenger help points talking back. we're getting them to do that at only 25 khz Mhm because they're lower power.

They're feebler. We're not asking too much of them right now. Incidentally, at the moment. I Seriously, do not know whether that particular tone signaling system is being used for the help points or whether they've gone to use the uh, how do you put it the Digital Data path that's apparently a part of the most recent Linkin signs, the display signs and stuff like that that's an unknown to me, right? Uh yeah.

Let's face it, my knowledge is okay. the these are designed 20 years ago. my knowledge is probably 12 years old so the hardware still there, but whether or not they're using it and unnown and part of what the Gatorade logic chips were doing was deciding who had priority. Okay, lowest priority of course was these uh program inputs MH Uh, next priority was uh, things like the PA mics M next priority was intercom and highest priority was the passenger emergency gear.

Come on, you need to Gat array for that. that's guilding the lily look. It was thought to be a good idea at the the time and it probably was because it shrunk our you know, maybe eight or 10 chip solution and a whole bunch of passives cuz for that kind of thing. I would have used uh, probably diode resistor Logic for a lot of this.

yeah, uh, a trunk. Maybe a yeah, a six or eight or 10 chip solution down to two chips. Yeah, so that was thought to be a good thing at the time. yeah, minimize the amount of hardw and you came to regret it.

Oh yes, Yes, indeed. Uh, There's a lot of other decisions that were made here that have proven over the period of time to be absolutely spoton. Uh, tell us about them. Well, things like uh, we've got probably three times as many output transistors as are actually needed.

Mhm, they're just in parallel. Uh, yep. Yep, This is where we probably want to go to the circuit diagram. But for example, 150 W amplifier has three pairs of these beasties.

Uh, now ordinarily out of three pairs of those, you would probably get a 3 or 400 WT RMS Rock and Roll amplifier. Yep, Okay, which you've designed many of. Oh yeah, J's did a lot more than uh, Train Carriage amplifiers. Yeah, this was an oddball project for Jans, which is why I think around about 96 or thereabouts.

they decided not to support these anymore. How many amps did you design at Jans Oh, all of them during during the period you were there. Uh, all of all of the all of them during the period that I was there. There's two that I did not design.

Uh, which are the first of all, the J300 and the J600, which were pretty much the first Power amplifiers that Jan's uh manufactured. And they manufactured a lot of those. They were almost a copy of The Phase Linear 700. Okay, the next ones, along with the J1000 and the J700, they're an alfier which almost sent Jan's broke.

Why? So they had some uh, fundamental design problems, some of them electronic, but most of them simple mechanical stuff related to Cooling and connection. The designer at the time, uh, didn't seem to be able to get a good grip on how to fix the problems. Incidentally, field failure rates were of the order 10% Oh, that's a lot. Huge Yeah.

and seriously, just about since James broke, you know they just in the middle of doing nothing. they just go and apparently they sounded pretty much like that. I I I I heard one do it and it's just say like that was it. Um, so lots and lots of lots of output devices later.

Um, the designer at the time, like I said, didn't seem to have a good handle on how to rectify it. Mhm H I came up with a whole bunch of incremental improvements, each one of which did its bit to get them back online. So at the end of the day, we had a set of solutions. You know, an amplifier would come in.

We'd apply the set of solutions, go out and have re reasonable confidence it would survive. Got it? Uh, at that time I just gotten my degree 1983 Uh, the designer left or was pushed I don't know which. uh Jan said hey Doug do you want the job I said yeah, pick me, pick me and that's when I started designing power amplifiers and the Jans S920 was my first one. Mhm weighed a ton I mean it to.

it was a 450 W per Channel amplifier and grossly overdesigned when it came to the heat sinking. the Transformers they weighed a ton. Well actually I think they weighed something like 22 kilo. the Ries loved you I bet.

Oh yeah, yeah yeah yeah. once they got their compo for their back, they. but the thing is they were reliable. they just didn't break.

Yep, uh as far as I know there's still 920s. uh Power amplifies rock and roll. go through a sequence or or they did when I was there. The newest amplifiers were bought by the cream of the crop touring outfits for higher and sometimes by Cream of the crop bands who happen to have their own PA systems.

Now these days, most bands just won't own a PA system. They don't want them, they just want to hire them. So they start off belowing to these Prestige touring companies. After maybe three years in service, three touring.

Seasons Uh, the touring companies don't want to bother about. you know the the the seams that are starting to the folded seams that are starting to just rip a little bit, the little bits of corrosion that are starting to creep in onto the connectors, the components that are just starting to Rattle a little. They don't want to have to tighten up the mounting screws on the transistors, so they sell them to the next tier down and buying the most best crop. Okay, so these amplifiers go to the next tier down who are hiring out to all of your little Pub bands? Etc Once they've done a few seasons there and they're really starting to get that kicked around look, then they go to maybe a quiet retirement in a practice studio somewhere.

Uh, where the local punk bands are just beating the living snot out of them, but they're not actually moving them around much, Got it? Okay, so they're being caned, but they're not being physically punished and after a while there, maybe they'll go into somebody's home recording system and just molder there for a decade or so. Um, and that's that's what happens to Old Amplifi So old amplifies? Yeah, do they eventually get bin? I'm sure they must, but they wind up in someone's basement somewhere driving. Seriously. I Don't know what happens to a lot of this.

Pro Gear at End of Life M Uh, some of it I Think just gets too old. Mixing consoles, especially mixing consoles are nasty. They're full of pots and lots of pots that have to be There's full-time pot jockeys. Isn't there that go around and replace the pot? Is that the correct term Pot Jockey or something close enough? Sol The Jockey.

Yeah, Um, no. Pot Jockey is somebody who sells you green stuff at a pub. Yeah, Mixing consoles and lighting control consoles can be an absolute Mongr because quite often it costs more to replace all of the pots that have gotten scratchy than it does to buy a new console. That's it.

So I think there's a lot of mixing consoles that die premature deaths. Mhm. You know, before things like power amplifies do so. We got three relays here.

Were was vibration an issue in absolutely? Oh yes, I picked it. did. I Yeah, because okay, see how they well and truly tied down? Okay, that was vital importance, but is a socket there in sockets there? yes, Are sockets more trouble than they're worth in a high vibration environment? For a relay, it's proven not to be so interesting. Yeah, Now, one of the things we were very, uh, conscious of I think is the fact that, okay, the reays are a weak link.

They're an electromechanical component. They're one of the components more liable to failure than others Mhm In in fact, the reays have proven to be ridiculously reliable. Wow. I think what brand are they? Uh, I think we were using Rron Rons Yeah, yeah, uh.

but we're uh, actually, we're using two different brands I think Ron and finder. Okay, actually, if you give me two seconds, I'll tell you what uh, they're on runs y they're definitely on run. But um, what we found? Well, we didn't want to solder the reace directly onto the PCB because if they were to be less reliable, then it made replacement an absolute bastard by socketing them. We might be making a rod for our own back and reducing the inherent reliability simply by socketing them, but that was a better risk.

It was a better trade risk tradeoff, and particularly given that uh, we weren't entirely sure about the relative reliability of relay on PCB versus soed relay weed. Go. soet, just go for the socket. Yeah, right And yeah, they're proven to be very reliable.

Would Would these things get hot? Were dried out? caps an issue? Ah, no. Well, yes and no. It's one of the things we made allowance for. We knew what the highest operating ambient temperature was going to be and I think it was Uh, 55 C Mhm.

Ambient We knew the degree of Maximum self-heating within the shazzy, which incidentally is why it's got these huge number of vent holes on the lid and around the side. Um, so I think that the internal temperature went. Was there ventilation there as the train moved? No, No, right, It's in a sealed compartment. Uh, not quite sealed, but there was no active blow through.

so we were relying on just natural convection for both. uh, heat sink Cooling and also chesy cooling. got it. So just from memory, the Uh, we knew that the maximum operating temperature for the Caps was going to be something like 65 or 70.

MH Uh, we used 105 capacitors with the electrolytic capacitors I got it from the horse's mouth one of the head honchos at Nipon Chemicon that the Uh, the arous coefficient or the coefficient to use in the arous equation for capacitor lifespan was double or half the lifespan per 10 degree increase or decrement. That's always been the rule of th. But I've actually had it confirmed by one head honchos from Nipon K That was kind of cool to have. So increase it by 10 and you can kiss half your life goodby.

Yep. So in the case of 105 degree cap, go down from 105 to 95 85 75 and that's 2 48 times the LIF span of a 5,000 hour cap. 40,000 hours. Now, in actual fact, we knew that uh, because of cyclical operation and they're never going to sit there for.

Incidentally, the lifespan had to be 50,000 hours. We knew that they were never going to sit there at 70 or whatever for 50,000 hours. Yep, it was inherent that with Cycl of Seasons, etc etc. and especially the fact that they were not going to be used at 100% duty cycle for speech.

Yep, they were never going to sit there baking at that temperature so they infinite almost the life and shelf life of the product. Yeah, and look over 20 years, how many electrolytics have I replaced Mhm? None mind you, this is. uh, we've used Nipon Chemicon caps and I Guess the two brands of cap that I tend to prefer Mhm rightly or wrongly are Nipon Chemicon or United Chemicon do? Uh and Panasonic yes I've had very good results from both of those. Both are the duck guts.

There's a few capacitors that I would not touch with a barge pole I Just did a video on that. Oh okay, it caps on all over the place. Wherever we've got capacitors, electros, it's going to be nip on Chicon. Uh for yep.

For a lot of the smaller stuff, these ones and also all of these little red fellas here, they're weer yep, uh and again. In in that class of capacitor, uh, weer are considered to be pretty good. Not necessarily duck scuts, but pretty damn good. Uh, when it comes to little ceramic stuff, uh, we didn't care.

No, it's like he whatever. whatever. The Chinese Factory can yeah, turn out as long as they were Cog or NPO dialectric. and as long as the voltage rating was hugely in excess of what was required, that was all that was required.

Mhm? Yeah. and these are high voltage. Uh, these ones are actually Mains rated. Yes, so they're self-healing Self-healing dialectric.

they are. Yep, Uh, we've actually had to replace a couple of those, But the main reason for failure is because in the environment that these are used just occasionally or more than occasionally, we will get a leak of Carriage wash fluid. Now this Carriage wash fluid. We still don't know whether it's in intensely acid or intensely alkaline, but it is intense and it just anything that it touches, it just stuffs completely and it's all.

It's all the worse if you actually have voltage on whatever this stuff drips onto. Ouch! And here's one we prepared earlier earlier. Let's have a look. Let's put that one to one side.

Okay, let's have a look at that that is. Toasty Folks that look at that. We are talking roasty toasty indeed. Now we've had a drip of or a leak of wash fluid down here and we've obviously had a resultant arcs across the top of the circuit board that have etched through the circuit board while self-perpetuating as arcs.

And it's just completely chewed that section of the PCB away. Ouch. over. On the side of the PCB Here you can see the results on around the legs of these pre chips here of just small amounts of surface wash fluid.

Now, all of this stuff is low voltage. It's 12 Vols mhm. Over on the other highly damaged end of the PCB we've got uh, full 120 volt levels. So uh yeah, this wash fluid has been responsible for I think maybe 34 of the repairs that we've needed to do I Notice you've got some uh snotting there.

Uh, why have we snotted those down? Vibration? Uh yeah uh. Basically a lot of these components such as the capacitors we've joined to each other. It's a sort of self-support society. Yep, it means that it makes it that much harder for any one to start.

Uh, developing a wobble or a shake and breaking through its own leads. y Incidentally, you will notice around these resistors that we've got little holes. Or actually not so little holes under all of the resistors. Are they ventilation? They are.

The resistors are slightly spaced above the PCB. Well, actually, they should be. They're not on this one. This is a pre-production Uh uh, no.

This one's actually a production unit. The the other one was the production sample. Uh, these should buy rights Be spaced up a little. Um, are you a fan of doing the little Loop in the lead to d-stress the when they expand due to temperature? Uh, no.

I'm a fan of doing the some inductor that's some indu creating an inductor. look again. A note for audio fools: Anybody who's silly enough to worry about a couple of tens of Nanah Henries of stray inductance in there low value of met resistors is a bloody Looney Seriously, it makes no difference. It adds a small amount of degeneration.

Does it affect the stability of your amplifier? Well, if it does, you've done a pretty good pretty crook job of Designing a power amplifier? Exactly Okay, What I am a fan of with regard to resistors is doing a bit of shaping on the legs. Okay, simply so that when you drop the resistor in the H, it self spaces it self spaces. not so much for stress reduction or any of that stuff. M but any of your larger components such as these five resistors.

yep, and sometimes the one Waters These fellas down here? Okay, Yep. Uh, I like to see and in fact I have a pair of uh, leg crimpers that put that shape in to space. Them got it And it's it's a nicity. On the downside, because your component is stood up off the circuit Bo it does leave it prone to rock, that's why, which is where you you whack your sastic on to fix it.

And here's the schematic: you've done well. Is it on one sheet? Doug Uh, yes and bear in mind because this is done back in '92 This was drawn up in Uh, Protel Schematic: Uh sorry. Protel schema. Dos Dos dos.

Yes, Okay, let's have a a look first of all at just one power amplifier and this is a power amplifier here. Yep, y Okay, we've got an input longtail pair differential pair with a small amount of resistor uh sorry. small amount of emitted degeneration current through the longtail pair is set by this resistor to ground. Ordinarily we might have say a current Source in there, but a current source is gilding the Lily on something like this.

the because we only want to go to Central Shush. Okay, the the output of the oh sorry, the longtail pair is differencing. Okay, the non-inverting input here of actual signal and the negative feedback here which is 330k into 6 K2 all right Now for AC it's 3 33k into 6 K2 for DC it's 330k into 6 K2 plus 100K Mhm. This means that uh for 60 Vols there we're expecting maybe about 15 volts there.

and indeed the bias for here is from a 15v reference Mhm derived by 47k resistor from the 120 volt Rail and 6K 2 down here. Okay, so uh, another diff pair up here here to give us our different signals. Uh, this is our voltage amplifier stage here, so that's basically a mirror. Okay, we then have MJ E 340 and 350 pre- drivers MJ 340 and 350 drivers mj15024 and 025 output stage.

got it. Output from this point here goes to Uh Coup capacitors and into the output Transformers we'll Trace that through in a second. One of the most important points is the Uh output limiting the short circuit protection. Output current here comes down through those two emitter resistors which in this case are 1 Ohm.

We use the voltage sensed across those resistors to well, if the voltage across there is high enough, it starts turning on these transistors. Got it? Which shorts drive away from the pre- drivers down to there or away from this drive down to there. How much Distortion does something like that add if we're talking about like a highi would would you do that on a rock and roll amp? Absolutely. Yep, Absolutely.

It's essential, of course. Yes, the whole point is you design all of the circuitry around these VI limits so that it simply does not take any action during normal operation. So if you've designed a power amplifi to run say of 4 Ohm speakers, you would typically not have those take action until the impedance drops below, say 2.7 Ohms go or maybe below 2.7 Ohms plus 2.7 R reactive Mhm or 2 sorry, 2.7 J reactive. Uh, so that normal reactive loads that you might and encounter in speakers just have no effect.

But as soon as you start getting serious overloads, which are enough to really stress your output devices. Oh, incidentally, of course you have to size your output devices so that they will cope with normal loads. No, that's yeah. Who' have th it? Yeah.

Okay, so once you you have an issue with fake transistors in the day, no, we didn't Uh, they just didn't bother to fake these ones or you just were lucky enough not to get caught out. I think it was a little bit of both I think that whoever was supplying our devices at the time, um, were getting them from the horse's mouth. Yep, uh, which was uh Motorola who became on semi? That's right. Uh, for a while we were using some devices I think Eb23 and Ed23 from a manufacturer called Hell oh uh.

Chinese Uh, sorry. Japanese I Believe they were the moral equivalence of these fellas here. Similar bandwidth, similar voltage, similar current, similar gain structure G G uh, all same poo. got it.

Um, so that's the you know Power Amp section. What else have We got? Another power amplifier? Oddly enough. Uh, Incidentally, you will see here that uh, this output rail comes down here through a big coupling capacitor 470 MK into our output Transformer mhm the output transform. Well, look if you, if you're that desperate.

yes, you can trace through all of this Reay line. but who cares? Uh, one of the items that you might be interested in having a look at though, is that Op amp there I Was going to ask about that one. Yes, Okay, it's having a sniff of signal going into the Vas the voltage amplifier stage. Now, while ever the power amplifier is acting within closed loop conditions, there's going to be very, very little signal there.

Yep, But as soon as it falls out of Clos Loop conditions due to either clipping or short circuit limiting. or uh, well, actually, those are the two main conditions: clipping and short circuit limiting. Uh, there's going to be a lot of signal appears there, amplify it. Put it through a rather crude halfway of rectifier.

Uh, feed it down into that lead there and you've got a short circuit indicator clip. Lead: Yep. B Go Too easy. Duplicated over here.

Yep, too crude, Too easy, Too simple. Yep. So is this second stage an absolute duplicate of the first. Looks almost identical.

Almost identical. The main difference is, whereas this one had one pair of output transistors, there we go. This one's got uh, well, Actually, we've got uh, this time this one had little pre- drivers, little drivers, big outputs. Yep, here we've got little pre- drivers, big drivers, and big outputs.

Got it. And part of the reason for that is we we actually configured the big drivers to contribute a fair bit to directly to the output. and you've got a, um, a couple of Uh four one in 47s on the output. Yeah, Reason for those is if the Vi limiting kicks in while the amplifier is driving a heavily inductive load, well, let's face it, inductors hate to have the current to them interrupted.

You try and interrupt the current to an inductor and the voltage is going to fly to one rail or another, so we catch that to either the positive rail or to ground. The dodes don't have to be particularly huge because the amount of energy involved isn't particularly high. No, that's right. So even though let's face it, we've got some 50 amps worth of output device here.

We've got 1 amp worth of diode there and it's suffices just fine. and dandy. Yeah, right. Another little bit of circuitry over here.

We've got an Lm35 temperature sensor, which is, uh, a fundamental part of the heat sink. we s in its temperature when it gets to 80. In other words, if ventilation really sadly blocked off. For some reason, if the heating gets to 80, we mute the PA system.

Got it? But we let the indicom system keep on going until it hits some 85 And at that stage, we decide. Okay, enough enough. let's kill the whole shooting match. Got it? So we've got a progressive shutdown down? Yeah, that's rather clever.

Uh, well. this is the if you like the the preamp signal steering bit. These are where all of the various switches. You know, the guards, intercom switch, drivers, intercom switch.

these are the switch contacts. They're dirty. great big industrial push buttons. Nothing fine or dainty about these.

So we wanted to have quite High uh, voltage switching levels on these. So we pull those switch contacts High to plus 120 volts withth 47k resistors and we switch those to ground with the actual switches. y uh. We then use the transistors within Uln Tw3 uh to detect what the level there is, whether it's high, low, or indiff and to act.

Usually these are drivers. these are Darlington pair drivers. That's right, we're simply using them as uh, transistors, logic input buffers. Really? Logic logic level translators may be 120 Vol to 5 Vol Logic translators.

Look, call them what you will. Okay, Overkill Again, okay, gross Overkill But it means that we can have Kilovolt transients on those lines and it doesn't matter. Nothing dies. Yeah.

Okay, we've even got a bit of switch debounce in there. Yeah, not a whole lot. but enough terrific. Now comes the ugly bit where we get all of those 5V logic levels and we feed them into the Gat array logic chips.

Little yeah, Okay, dismal fail. Uh. I'm disappointed in you Doug Oh shush, uh. including you're an analog man.

Yeah, uh. mind you, between you, me and the gay post, I was dragged into this side of things kicking and screaming and in fact it was I think Peter Godwin at Jans Who did the logic? uh, uh, what do you call it the logic equations required for these and program them in? Got it? He? Um, okay, including. We've got a an oscillator over here, which we tune up to 50 khz. So oh, Lm567.

Yep, uh. now you can use these Lm567 as either uh, oscillators mhm or down here as a tone decoder. A sniffer, they are 567 is the classic tone decoder IC that's what it's famous for. We also use them as a Precision oscillator.

Incidentally, getting these amplifiers back you know, 18 and 20 years after manufacture, we discover for example, that the 50 khz oscillators have maybe drifted 250 HZ 200 HZ 300 Herz over that period of time. Is that a drama? No. Because the tone Dakota can capture, it has a wider capture bandwidth than that. Yeah, uh, that's a credibly small range and generally not due to chip drift, but of course, due to capacitor and resistor drift.

Yep, Okay, now that's all the boring bit. Now let's come over here to what happens at each of our various inputs. The inputs are balanced configuration so that they reject common mode noise and only respond to differential input, which is exactly what you want. Okay, we've got Bridge rectifiers here so that extreme transients present there just get clamped to the supply rails through the bridge.

Rectifier Balance Amplifier Now in these early versions God Help us. We're using J Fets for audio switching and nobody uses Jfets or anything anymore. do they? Uh, yes, yes. Name an Nish Uh.

One of the best possible niches is of course, the uh input buff amplifiers to condenser microphones. Yeah, uh. condenser microphone preamplifiers where you want staggeringly low leakage currents and staggeringly high input impedances and staggeringly low parasitic capacitances. Got it? And that's what I did for crusted Road microphones.

Mhm. Either that a whole another story. Yeah, either that or I Used Jfets with Pilot lights also known as like Tri Tubes Yeah Tubes. Tub tubes, No Tubes Tubes with the CH Yeah C H O B S tubes.

Oh boy. So okay. We're audio switching there with J Fets and spitting them out into the two mixing amps to go out of the two power amplifiers. Uh, well, that's it.

Oh, and the only, the only particularly interesting thing here is where we're getting the uh, the Indicom line, doing a whole stack of high pass and low pass filtering and buffering and then sending it into this detector for 25 khz. That's that's our tone decoder. Yep, his output then goes back into the Gatorade logic chip. So there you go.

That's your lot. That's it. That's it. That is the Sydney rail.

What is it? What's it called? It's a PA It's a Pa uh Crew intercom amplifier PA Crew intercom amplifier for City rail. Everyone hates City Rail. Yeah, and not only that, but apparently there have been Um court cases bought recently to uh State Rail because of the poity of PA announcements. Uh, there's cases some people didn't Some some lawyer missed his stop and decided to se no, some uh, blind fellow site imp uh, has.

He's been catching the train for decades and he's got the Irit because quite often they just don't bother announcing the station. Skip them. Y And being blind, he's got no real way of looking out the window to see where he's at. Fair Call: yeah and look good on him.

Look I want to see these things used more? Next off Stal Come on Doug You must C it all. Mind the doors please. Yeah, the announcements are so bad. Enunciation: please guys.

Enunciation: Thanks Douy Yeah. Speak like I Say you should not the way I speak. That's it. That was an interesting tear down.

Okay, um well. maybe you'll somebody will bring in something else vaguely Industrial in nature for you to have a look at one day. Waet put the word out. Catch you later bye-bye.