The ETI-480 Power Amplifier - Second Edition

[00:00:05]Greetings, the Astro30 here yet again and welcome back to the yet another amplifier channel, formerly known as AAL. Thank you for joining me. No, uh, let's get serious and look at what we're actually looking at today. On the screen here at the moment, we see the classic ETI 480 or ETI 480, however you want to say, 100 W power amplifier. Now, that was published in March of 1977 in the Electronics Today International Magazine of Australia. And thousands of these were sold as kits by Dick Smith and JCAR Electronics back in the day. And I have actually built several of these modules up and they do perform reasonably well.

[00:00:47]However, I was analyzing the circuit the other day using simulation software TINA and I was just interested to see what the actual frequency response with the original components in its original layout was like and it's it's got mixed results. So if we look at a more clearer redrawn schematic of the amplifier which was used for my first iteration that I made a PCB of, we can see that it looks like a pretty standard typical amplifier design. Got a current source up here with a voltage reference longtail pair input stage voltage amplification stage bias and drivers and the output stage which is a complimentary feedback pair or Clay. However, the biggest red flag to this thing is C5.

[00:01:37]This 330 poparad capacitor going across the negative feedback resistor that is suspiciously high. There's also a few design flaws here. We've [sighs] got this 3.3 nanopharad capacitor between the base and the negative supply rail at the base of the bash transistor here.

[00:01:54]This is an attempt to fix an oscillation problem that is occurring most likely on the output stage of the amplifier. And this is a band-aid fix rather than actually, you know, addressing the real problem that is the source of the oscillation. And the other design flaw of this is this output stage configuration here. What they're doing here is they're providing negative feedback back to the drivers to boost the voltage gain of the output stage.

[00:02:22]And how we calculate what the gain of the output stage is is we'll ignore the lower half of the circuit here as it's symmetrical to this. And we only need to look at two of the resistor values in the voltage divider here. So to calculate it, we'll call this resistor here RF or resistor feedback. And we'll call this resistor here resistor gain or RG. So the formula is AV = RF plus RG / RG. And with these current two values here, we come out with a figure of 4.33.

[00:02:58]So that's a gain of 4.33 times or plus 12.7 dibels. And the problem that this amplifier has is with these resistor values is they get extremely hot when driving to full power. These 110 ohm resistors are actually 220 ohm 1 W in parallel. So making a compound 2 W resistor. And this one's a 1 W resistor.

[00:03:22]They're barely large enough to dissipate the current. So this circuit was originally designed by David Tilbrook and regardless of the fact it has stability issues and there has been a lot of forums on the internet where some of these have actually self-destructed and someone actually did build a few of them and test them and found they tended to want to oscillate around about 180 kHz. I've never actually had one self-destruct but maybe I was just one of the lucky ones. So let's have a quick look at TINA and the Bode plot for this circuit in its original configuration.

[00:03:57]So if we look at its frequency response B plot, well, it doesn't look terrible.

[00:04:03]However, in the phase curve section here, it is got this like sort of hump here around about well, I don't know where that starts.

[00:04:13]So, I put a cursor at where it starts to hump off. At around about 300 kHz is where it starts to dip down. And we're only at a crossover point of - 117°.

[00:04:28]So, if I shift this down to well 180, we can see here that it's actually 644.9 kHz is where it crosses over at 180°. It really should be more above the 1meg mark. So rather than messing around with the original circuit trying to get the frequency response curve a little bit more respectable without this suspiciously large 330 pair capacitor across the negative feedback resistor. I decided to redesign it. Well, for the most part redesign the output stage. And there is my redesign circuit. So I've added in a lowass filter capacitor here.

[00:05:09]I've messed around with the values of the feedback capacitor and the VAS miller compensation capacitor. I got rid of the bootstrapping which is this capacitor and resistor here coming off of the collector of the VAS going back to the base of the error correction side of the amplifier's longtail pair as well as got rid of that negative feedback gain boosting and modified the currents going through the longtail pair and the VAS stage by changing the configuration of the constant current source up here and a voila we've got a much more respectable looking phase curve here and frequency response curve and our 180° point which is about there close enough is over 1 MHz where it should be excellent so just running a DC test on this we can see our output DC offset here is 2.6 m which is quite respectable as to what the output offset of the original amplifier is well it's more like 23.89 8 9 m. And finally, here's a look at a cleaner schematic of the amplifier. As we can see, it looks pretty much like Tina. I've also added in these two diodes here on the supply rails between the high current side and the low current side with Bob filtering here. And what that is an attempt to do is one isolate the low current input stage from the high current output stage and also prevent the input stage from starving for current during a large transient here on the output which will cause the power supply rail to dip.

[00:06:52]So what we need to do now is we need to build this up on breadboard and actually see if it functions and it does function in tener. It does output. So we can see that the virtual oscilloscope is producing an output. And if I just put a cursor on our output point here, we've got 30.32 volt peak coming out of it. And if we look at that as an RMS AC voltage here, it's 21.48 volt into 8 ohms before clipping, which is about 57 W, which is not bad. But that's using ideal transistors and not taking into account supply rail dipping when it gets to that power. So without further ado, I'm going to build this up now on breadboard. I'm not going to worry too much about including the parallel inductor and resistor here on the output so much, but the rest of the circuit. Yes. And just test that it actually functions and performs the same way that Tina says it does. Let's get going.

[00:07:50]And a fair few hours later, actually a couple of days later, I've got the thing breadboarded out now. I had a few issues that I had to debug and sort out before I presented the circuit. Just makes the the video a lot easier to record. Ignore this trim up here. It doesn't do anything.

[00:08:10]The C the circuit is on. And basically, I've now created a yaba.

[00:08:16]That's not yaba daba doo. That's yet another blameless amplifier. But well, most amplifiers look like a blameless amplifier anyway.

[00:08:26]But as I said, it's currently turned on.

[00:08:28]Bias is adjustable and sitting at where I roughly set it, which was about 32 to 35 MA.

[00:08:36]Just tolerances and different temperatures today. And the multimeter is looking at the output node point and we're getting about 2.5 2.6 molts of DC offset, which after the amplifier has stabilized for 5 minutes is what Tina says it should be. So, that's pretty accurate. It starts out at around about 3.2 when it first turns on cold. The only main issue I had with this was the bias was unstable. You'd adjust it and it would suddenly shoot up in current and you'd have to turn the pot several turns to get it to come back down again.

[00:09:14]Well, after fighting with it for a couple of hours, I just touched one of the driver transistors, which is behind this other transistor here with the seal washer.

[00:09:24]and noticed it was warmish. And as soon as I touched the body of the transistor, I watched the current drop down.

[00:09:32]And that was a Eureka aha moment. So that's the BBE multiplier transistor, the one in front here with the seal washer behind it. That used to sit there between those two output transistors. So it was only monitoring what the output stage was doing, not what the drivers were doing. So, with that now in contact with one of the drivers at least, it's now bias stable, which is excellent. And the original ETI 480 has the drivers mounted to the heat sink bracket. So, all right. So far, we've got a working prototype. But what I have to do is I have to just see if we get an output signal across my dummy load there.

[00:10:15]Currently got it at 4 ohms. I'll stick it back to eight. increase the current limit to say 1 and 1/2 amps for this test and see if we get to full output power before clipping and clips symmetrically as it does in Tina. So let me get that configured now and ready to go and we'll have a look at the oscilloscope. Okay, so with the amplifier loaded into 8 ohms, current limit set to 1.5 amps, I've got a signal coming out. Looks relatively clean.

[00:10:48]And yes, it does get to clipping.

[00:10:52]And it symmetrically clips, which is nice.

[00:10:56]And just want to know where it's bordering on clipping, which is about there. Getting about 19 volt RMS, which actually surprisingly is 45 W RMS, which is not bad. And it's actually a lot more than what most of my amplifiers when I'm doing the preliminary breadboarding and testing power outputs usually is which is about 36 to 38 W. So that's nice. So currently the amplifier seems pretty stable with my arbitrary capacitor values. So now let's see if it's stable into 4 ohms by switching the load around. I'm going to have to increase my current limit to probably more like 2 amps.

[00:11:41]just to make sure there's enough head room there because it's probably going to draw about 1.8 amps. We'll soon find out. So, here we go.

[00:11:51]Yes, we do have an output, but I'm not sure why it starts out triangular and then drops back down to sine wave.

[00:12:07]Okay, still symmetrical clipping, which is nice.

[00:12:13]Uh, 16.2 volt RMS, 65 W into 4 ohms.

[00:12:18]That's not bad at all. Not too shabby.

[00:12:22]Well, I'm quite pleased about that. Uh, the thing is though, I'm not really happy with what it's doing when it first starts producing the signal into 4 ohms.

[00:12:34]So if I lower the voltages per division down to 5 volts and then just quickly give it a transient.

[00:12:45]It doesn't seem to show that waveform going triangular like it was before.

[00:12:53]H power supply fans doesn't seem to be doing it now.

[00:13:07]H. Well, that was kind of odd. So, okay.

[00:13:10]It it appears to work. I maybe just seeing that anomaly because of whatever the scope's time base is doing at the point in time. I don't know. But um I don't have that weird hump at the bottom negative peak like well the old workhorse was and that elector uh viewer contribution amplifier does don't know something to do with the output stage maybe.

[00:13:38]So, the next thing I want to do is I want to do a frequency sweep test on this just to see how um well it performs and where it cuts off at. I'm going to probably have an issue with uh using a sound card output of a computer which is only good up to about 20 kHz anyway.

[00:14:03]Okay, so I'm ready to do a sweep test on this now. So, I've got currently 1 kHz going in and about two odd volt RMS coming out.

[00:14:16]And that was the computer bringing up a uh notification.

[00:14:23]So, I'll just uh get rid of that. So, now what I want to do is I want to do a sweep test. So, I'm going to have to turn this clockwise.

[00:14:32]So, I'm going start actually the other way. I'm going to start with a pretty high time base and I'm going to play the sweep.

[00:14:54]We're up to 8 kHz already and it's looking relatively flat.

[00:15:00]I'm not sure what that volume going up and down issue is, but that could be a connection issue on the board.

[00:15:08]And around about the 24 kHz mark, the the audio output of the sound card just buggers off. So, so far, frequency response looks reasonable for what it is.

[00:15:26]Uh yeah, and considering it's on breadboard, which as I've said multiple times in several different other videos, it's not really a great test bed for power amplifiers or anything that's current sensitive for that matter, but uh it works for what we need it to do. So I guess the next thing to do would be to do a stability test on this by deliberately capacitively loading the output and see how the amplifier performs. I got a 100 nanofarad capacitor here connected to at least one clip lead test lead. There's another one off camera here that's connected across the dummy load at the other end. So if I now connect the other test lead to the other end of the 100 nanofarad.

[00:16:09]Yeah. Well, we can see that it is ringing but it does dampen back off and doesn't continue to like oscillate.

[00:16:18]There's no extra high frequency oscillation on there. So that seems pretty normal to me. Uh but there is no output inductor either. So that could be an issue which I might want to find one before I continue this test. Okay. So I've got a random inductor resistor combination there which is just a madeup values and is not actually designed for this amplifier. But it has improved matters. So, that's all that really concerns me is is it stable into capacitive loads or when it's driving difficult loads. Seems to be um and that's at 20 kHz, by the way. So, so far the values that I've got on the board here seem to be okay and coping well. It could probably do with a little bit more improvement.

[00:17:07]And by the way, uh, if you didn't get it by the video's title, this is the ETI 480 second edition, even though it barely resembles the ETI 480 now.

[00:17:18]But stability wise, it seems okay. I can't get it to go into isolation yet. I need a higher value of capacitor. I thought I had a 220 uh nanofarad somewhere, but I seem to have lost it.

[00:17:35]So, I'm going to look around my stash and see if I can actually locate a 220 nana. Well, I couldn't find a 220 nanofarad. So, I've got a 150 nanofarad and the original 100 nanofharad in parallel. That gives us 250 nanofharad.

[00:17:57]Yeah, right. That's [laughter] pretty bad. Um, but it does does look like the output definitely does need an a parallel resistor and inductor.

[00:18:11]Otherwise, that would be like probably 100 times worse than it is now. It's probably not what I was expecting, but yeah.

[00:18:25]Okay. That is however with only 100 pair capacitors across the base and uh collectors of each driver just to keep the output stage stable.

[00:18:36]Now I'm interested to know whether those capacitors are actually even needed and or does it make the oscillation on the scope worse. So first I will get rid of the extra capacitor in circuit. So, it's now only got the 100 nanofarad across as it did before. Current on the power supply is limited back down to 500 mA just in case it does go into isolation. So, I'll remove one of these capacitors.

[00:19:07]Signal looks the same. So, I'll put that one down there safely and I will remove the second one and be quick to turn the power off if something really does go bad.

[00:19:19]And not much change. It's exactly the same. No change. So let me connect the 150 across the output again with the 100 nanofarad.

[00:19:34]It um doesn't seem as bad as it was.

[00:19:37]That's cuz it's come disconnected.

[00:19:40]Let's try that again. Will you connect, please? Thank you.

[00:19:47]Yeah, there it is.

[00:19:50]Ah, okay. So, I want to see if increasing the capacitance to say 220 across these drivers make a difference.

[00:20:02]So, I'm going to have to carefully do this without shorting anything, but the current is limited.

[00:20:08]If I carefully plug that into there, there was not much change on the output.

[00:20:15]So now I have to connect the other one across the other driver and that's that way around. I got to make sure I get my base and collector right otherwise there's no point connecting it across the emitter and collector then it will oscillate and no there is no change in the output. So yeah, at least yeah, I've just removed the 220 on one side.

[00:20:47]Remove the other 220.

[00:20:50]No change. So, all right. I think what I've got so far going on on this board here is perfectly fine.

[00:20:59]Just checking the condition of the drivers. They're pretty warm to the touch, but not overly hot, so that's good.

[00:21:08]So, I guess now I'll play some audio through it. I'll stick the 100 poparads back in. It's not doing any harm.

[00:21:16]And uh just hook it up to a speaker and actually just see how the thing sounds.

[00:21:21]Okay, I got the thing connected to the speaker, but it's turned off. And I'm using the camera's mic for this particular part cuz well, it picks up the sound better than the desk mic does.

[00:21:34]So, I want to see if there's any turn on transients. So, I'm going to hit the power.

[00:21:40]A slight one.

[00:21:43]And then I'm going to play a random track from the YouTube audio library and see how it sounds.

[00:21:53][music] [music] I am [singing] waiting.

[00:22:05]for bus at the end of [music and singing] night on the street land [music] face [singing] claim my life [music] I'm going away I'm [music] going away [music] I'm going [singing] away.

[00:22:34]And the sun going to [singing] shine on a brand new day.

[00:22:40][music] [music] Must be [singing] somewhere.

[00:22:52][music] >> Sounds all right. Apart from those tweeters sound like absolute crap. I'll fix that at a later time. So, I think that's going to conclude this video.

[00:23:05]I've gone back to the desk mic now. Uh I'm pretty pretty I don't know what that was.

[00:23:15]Anyway, I'm pretty happy with uh how it uh sounds. Sounds all right. Might turn it off now.

[00:23:26]Hardly any turnoff transient which is good.

[00:23:30]So that's probably going to do it for this video. I just wanted to experiment with the original circuit and slightly redesign it and maybe correct a few of the poor design choices of the original which I think I have succeeded in doing.

[00:23:46]Uh the ringing on the output there when it's loaded with 250 nanofarad. Yeah.

[00:23:53]Well, it does look pretty bad, but yeah. Anyway, I think this is where I'm going to leave this video because, as I said, I've explored it enough. Um, I may route a PCB for it. I may not. I don't know yet. Uh, I might do something else.

[00:24:12]But anyway, uh, I'm going to end this video here. I'm the Astro 30 and thank you for watching. If you enjoyed this video, be sure to smash that like button, share the video, and subscribe.

[00:24:24]It's totally free. [music] [music]