Signal amplifier works

... unless there is no signal at the input.

I started with some information at the following two URLS:

formatting link
formatting link

The first URL goes through the process of building a high-gain amplifier, and it works O.K. with the exception that there is no power. The second URL goes into more detail and shows how to deliver some power to the speaker as in figure 16. I have different parts on hand, and I am working with a 12V supply instead of 9V as in the first URL. Since I am looking to deliver big chunky volts to my speaker, I went ahead and modified Figure 16 and the example in the first URL to obtain the following circuit:

12V +---+----------+----------+ R1 = 510K | | | | R2 = 5.9K | > R2 > R3 | R3 = 220 R1 > < < | C1 = 1uF < | | | Q1, Q2 = 2N2222 | +-- | | Q3 = 2N3055 input | | \ b |c | o---||-+-(Q1) -----(Q2) -----(Q3) C1 | |e / | 1uf | --- SPKR | | | | o----------+---------------------+

SPKR is a ribbon tweeter with a 2 Ohm, 10 Watt resistor in series.

R2 ought to be 6K, but 5.9 is what I have on hand.

Driving the circuit with a 50KHz signal produces 10V across the speaker, which decreases to about 8V as the power transistor heats up. I had a problem with thermal runaway until I got a sufficiently beefy heatsink on Q3 and the 7812 supplying the 12V rail. Q3 runs quite hot, but I think I can fix that by substituting a TO-3 package for the TO-220 currently in the circuit, or by reducing the current. Not sure which is the proper solution, but the speaker handles the existing power.

220 Ohms for R3 was derived by guessing, and noting that 1K attenuates the signal too much, but I don't really know how I should be calculating the proper value. The speaker is usually paired with a 5.1 Ohm, 5W resistor, but I found it got too warm in this circuit, so I used the 10W part instead. I have no idea what the speaker impedance really is in operation.

The horrible fact is that this circuit works as long as there is a signal present at the input. I am using a simple signal generator from my DSO, but it doesn't allow the amplitude to be changed, so I put a 100K trim pot on the positive side of C1. Before I modified the circuit by adding Q2, the trim pot worked as expected. Now, attenuating the signal even a little causes the output across the speaker to fall off very quickly, with serious distortion occurring as it falls (it seems the duty cycle goes from 50 to 90%, but that is occurring somewhere in-circuit.)

If there is no signal present at the input, the circuit goes nuts very quickly and ends up producing a much-distorted 9KHz signal at full power at the speaker. This seems to be occurring at Q3, but I don't really understand what is going on. My thinking is that with Q1c-Q2b at ~6V, the DC load through the speaker is too much for Q3, but when a signal is present, it 'works'.

What I'd like to do:

In the original form of the circuit, the speaker is driven through a capacitor connected at Q2c (3300uF, 50V), but of course there is no power available, and the voltage drops from 11V to 1 or 2 volts at the positive terminal of the speaker. With the above circuit, it is not obvious to me how I should rearrange things to decouple the speaker from the DC signal.

I would like to figure out why the circuit only works at 'full on', and blows up otherwise. The last problem is the power-on state, which launches a huge spike through the speaker driver, and indeed, changing the input signal frequency on the DSO causes a similar discontinuity that is being amplified quite a bit as well. Suggestions as to how to reduce this spike would be appreciated. Off the top of my head, I suppose I could install an SCR in line with the speaker driver and have a switch to turn on the power and another to enable the speaker output via the SCR. I would test this, but I don't have an SCR available ATM.

The entire objective with this little project is to develop a simple amplifier that I can use to drive this speaker I have, where fidelity is not as important as stability. Plus, it would be nice to be able to use the parts I have on-hand.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve
Loading thread data ...

formatting link

Where did you get a speaker that works at 50 KHz?

Reply to
Michael A. Terrell

formatting link

It's a ribbon tweeter, ostensibly rated to 65KHz. It is difficult to say what it's actual output is at those frequencies, but I can put two of them face-to-face, drive one of them, and get signal from the other with a scope. With the above circuit operating as described, the second driver shows 54mV at the terminals -- using ribbons which are sub-optimal. So I guess it works.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve

Actually, here's someting wierd. At 4MHz, I measure 200mV, almost nothing at 6MHz, and 60mV or so at 8MHz. 4MHz may be a resonant frequency of the ribbon, given the speed of sound in copper, but that's just a guess.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve

What's the specific ribbon tweeter? What will be used to drive the amplifier when in practical use? (It's output impedance, Vpp.) What's the application? (Pestering bats?) How much power output do you expect? What's the range of frequencies to be emitted? A pure tone of 50kHz?

The first stage looks all wrong to me, anyway. Are you saying it worked? That's hard to imagine. All I see is that Q1 is hard-saturated with its collector very near ground and the Darlington emitter follower not doing much of anything. What in heck were you driving that thing with via C1??? Do you have your design notes that you might share?

Jon

Reply to
Jon Kirwan

formatting link

The amplifier clearly needs a bootstrap >:-} ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

The driver is a proprietary design; the label sez "Quadral / 923 112 /

6 ohms", but it is normally attached to a crossover with a 5.1 ohm 5W resistor. Practical use is in a speaker cabinet powered by a conventional stereo system, but as I'm trying to replace the ribbons, I need a way to test the thing without installing it in the cabinet.

Now that you mention it, there is a vampire infestation in Toronto so I may use one as a perimeter defense weapon, but that's after I prove A replacement ribbon. In practice the device is intended to produce audible audio frequencies, but for testing purposes I'm happy to use ultrasonics to keep the workbench noise level down.

You're correct to be surprised it worked, and I still don't know exactly what happened. I wired up a TO-3 version of Q3 and substituted it into the circuit and found out that the old Q3 had let some of it's Magic Smoke out, but was still working as I described. I have no idea how that happened. Ah, it seems I miswired Q2; emitter to GND, collector to the base of Q3, and to +12 through R3.

Again, wired correctly, and without Q3, I have 10.6V square wave at the collector of Q1, and 3.6V at the emitter of Q2, open circuit. Is it safe to apply the output of Q1 to the base of Q3?

Oh, and I'm driving it with the signal generator output of a DSO Quad, which is a handheld 4-channel scope which retails about $220.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve

If it can hit 50 KHz, it's too late for a jock strap. ;-)

Reply to
Michael A. Terrell

It looked crazy to me too.

Uncle Steve, you need to find some better websites. You could try an opamp with a push pull output stage. Or just a power opamp. Hey Jan had a nice audio amp... If you want to do a class A thing, then google the Zen amplfier.

George H.

Reply to
George Herold

I suppose so. It just doesn't seem to have enough discreet components. The first stage is simple enough. The bias resistor, R1 and load resistor R2 specify that the voltage at Q1 collector is about

6V with no input signal, and allows the output to swing about 10V. The second transistor is used to boost the current, and then in theory Q3 does the heavy lifting. I'm not really clear on how Q2 does its thing, but that's what the experimentation is for.

It may help to know that initially, I started with C1, R1, R2, Q3, and a capacitor coupling the collector with the speaker + terminal. That's when I discovered there was no power available to drive the speaker. I substituted a 2N2222, and then made the circuit above. It didn't work properly because I wired it wrong (see previous message) but as I described it did output something like the the input signal once the 2N3055 halfway burned out.

Maybe. There's no shortage of crap out there.

Hey, I have all these transistors lying around and since I hear they're good for amplifying signals, I thought I might as well use them.

Googling... Interesting, but I don't care about linearity all that much. This is just something to run one frequency through the driver at any given time, and therefore does not have to be special or complex.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve

Um. That's not the way I'd say it. But that would be one of the symptoms.

I think the word is "simplistic." Not simple.

Not in this universe.

No, not really.

Okay. So this is when a little theory goes a long way because, it turns out, there is good theory for a circuit like this. Experiments are for when theory fails you because it is incomplete or too complex to completely solve. This circuit isn't too complex and theory here would be complete enough for the job.

Okay. So let's look at the circuit you built. Here it is:

,--------+--------+--------+-------, | | | | | | | | | | | | \ | | | | / R3 | | | \ \ 220 | --- | / R2 / | - V1 | \ 5.9k | | --- 12 \ / | | - / R1 | | | | \ 510 | | | | / | | | | | | |/c Q2 | gnd | +------| 2N2222 | | | |>e | | | | | C1 | | | | || 1u | |/c Q1 | |/c Q3 IN---||----+------| 2N2222 '------| 2N3055 || |>e |>e | | ---, | | | | \ | | / SPEAKER gnd gnd \ 2 Ohm / | | gnd

Ignore everything from the base of Q2 and beyond (to the right.) Ignore C1 for a moment (assume it isn't connected up.) Just look at R1, R2, Q1, and the +12V supply rail:

+12 +12 | | | | | \ | / R2 | \ 5.9k \ / / R1 | \ 510 | / | | +----Q1 collector ??? | | | | | |/c Q1 '------| 2N2222 |>e | | | gnd

No signal input (obviously.) So you think Q1's collector will be at 6V? Really?

Let's see. Assume Q1 Vbe=0.7V (within 0.1V, probably) for now. This means Ib=(12-0.7)/510=22.2mA. Ic, even in the worst possible case where Q1 is completely saturated and its Vce=0, would be Ic=12/5.9k=2mA. So the base current is about ten times the maximum possible collector current. Now THAT is overdriven, with beta=0.1. You don't see that very often. But the gist is that Q1's Vce will be driven so close to 0 as to make no difference at all.

It won't be 6V. The collector is hard-clamped against ground.

Now this doesn't mean you can't drive the input with something that would move it. But this circuit is going to be sitting with Q1's collector clamped hard to ground for half the input cycle, at least. If you drive it hard enough, at

50kHz (Xc of a 1uF is only about 3 ohms), you will decidedly be able to pull down hard enough during part of the cycle to actually get the collector to move off of ground, I suppose. But this is a mess.

And the quiescent Q1 collector will not sit at 6V. Measure it.

Jon

Reply to
Jon Kirwan

You forgot the bootstrap effect >:-} ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

I didn't miss out on the "Negative Feedback in Audio Amplifiers" thread. ;) But since this is posted in .basics it seems appropriate to put on the kid gloves. There's still plenty of time to add in a bootstrap, too! :)

Jon

Reply to
Jon Kirwan

Here's a schematic of something close to what you want. It is set for around 50kHz (I'm assuming you wanted that) and uses your three transistor types. If your IN drive is low impedance, then you don't want it to go more than about

40-50mV p-p. I set the quiescent Vc=7.5V or so. Gain is 40*Iq*R2, but since Iq is about 400uA that's about Av=160. It will actually be a little lower than that. But it's still likely to be somewhere around 120 to 140. With about +/-3V around the 7.5V center, roughly speaking, that would mean no more than 6vp-p/140 or about 43mVp-p. Which is why I said 40-50mv p-p input at C1.

I've even included the ever-popular bootstrap (R7 and C4):

Since I've no idea what is driving the thing, except that it appears to be some kind of low impedance driver if it expects a 5.1 ohm resistor, then I feel fine stopping here. You may need to supply that DC path, though, as a resistor divider and then drive C1 from that.

Jon

Reply to
Jon Kirwan

I made a mistake reading your values. You are using a 510k, not 510 for R1. This reduces Ib by a factor of 1000, to

22.2uA. With a beta of 200 or so for the 2n2222, this suggests a possible Ic=4.44mA. That times a 5.9k ohm collector resistor, this is a drop of 26.2V across R2. Not possible at a supply of 12V. Instead, you will see a drop of about 11.8V with Ic=2mA.

Not as "hard-clampled" as I wrote earlier. But it is still in saturation and will clip half-cycles. Not so good.

Jon

Reply to
Jon Kirwan

speaker

How much power can that deliver to the 2 Ohm load ?:-}

The OP seemed to want 10V P-P across the load. The OP was somewhat vague on the load, 2 Ohm in series with a "ribbon tweeter". Wonder if that was a piezoelectric or voice-coil type?

I'd suggest a class-B structure with some feedback (and some engineering thought :-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

speaker

No power supply bypassing, either. Anyway, half that at best. And with plenty of distortion.

Ah, come on. It's got the three transistors in the original configuration and it has a bootstrap, too!

Now _you_ want a differential pair, current mirrors, a VAS and a Vbe multiplier, two to perhaps four output BJTs, loop feedback, speaker reactance compensation, a bipolar power supply or perhaps a bridge-tied speaker, and THD of 0.1% with bandwidth to 50kHz? With crafted design thinking to boot?

Yeah, right. It took less than 10 minutes for me to hack up the above with only a few calculations and then use my custom program to convert the schematic into ASCII text. I'm sitting pat at that level of effort to help.

Jon

Reply to
Jon Kirwan

speaker

ohm

OK. How much signal does it put into 2 Ohms ?

My guess is that it just peak charges the 3.3uF (C3) which then decays slowly thru R5 (1K). ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     | 
| Analog Innovations, Inc.                         |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

speaker

ohm

Yeah, I see that clearly now. Active one side only. Well, I'm just an ignorant hobbyist. So I'll chalk up last night to that, plus a kick in the head. Thanks. Well, at least it was only a few minutes wasted time last night.

Jon

Reply to
Jon Kirwan

Sorry, I suppose it is more correct to say that it is about 6V when the input signal is crossing zero.

Hey, I'm just a noob so the theory is not something I have a really good handle on yet. The experimentation allows me to fool around with my scope and see what happens with different values on input and with different component values.

R1 is 510K ohms, so I'll just amend the following diagram to suit.

Ic/Hfe = Ib

.01A / 50 = .0002A

I chose 50 somewhat arbitrarily. Initially I was making this circuit with the 2N3055, for which the data sheet says that the DC current gain is 20min and 100max, so 50 seemed like a good starting point.

Vcc - Vbe = Vb

12V - 1.8V = 10.2V

Again the data sheet says that the Base-Emitter on voltage is 1.8V max. Perhaps I should have used 1V, but there you go.

Vb / Ib = Rbias

10.2V / .0002A = 51K ohms

(1/2 Vcc) / Ic = Rc

6V / .01A = 600 ohms

Next, I decided I could live with an order of magnitude less quiescent current, and so chose 510K and 6K ohms respectively for R1 and R2.

I suppose it doesn't help that after moving on, I changed Q1 to a

2N2222. So, if I fix the calculations based on the assumption of .7V Q12 Vbe, the calculations are as follows:

Ic = .001A / 50 = .00002A (quiescent current)

12V - .7V = 11.3V

11.3V / .00002A = 565K ohms

6V / .001A = 6K ohms

I happen to have a 560K ohm resistor, so I could use that if it was necessary.

Not exactly.

I agree, but not quite for the reason you say.

Sorry, I meant to say that with a driver coupled to the collector with a capacitor, the (zero-crossing) voltage should be about 6V. In theory. After poking around a bit it is now clear that the behavior of this circuit is wierd.

Firstly, the signal generator output has an amplitude of about 1V. I put a 100K trim-pot in front of the capacitor to vary the input from the signal generator. I'd been using a square wave previously, in part because the sine-wave caps at 20KHz with current DSO firmware.

With a 560K bias resistor, and an open output, the sine-wave output at Q1c is thin (duty cycle of 34%, whereas input is 50%) and the output saturates at 11V. Just for fun, I changed R1 to 2M ohm and found that the output waveform is more accurate, although requiring more input attenuation to achieve. In this case the output signal attains a 3.x DC bias and the peak-to-peak output is 7.xV. I assume this is a result of interaction of the input signal and C1, so I suppose I have to look into the proper way to float the input signal. I'm guessing that the impedance of the signal generator comes into play here.

Earlier today, I replaced the broken 2N3055 at Q3 and found that the output of Q2 was now about 10.6V, whereas I measured it's open-circuit voltage as peaking at 3.something. Unfortunately, I had set the scope input to AC, so that was a case of GIGO. Open-circuit, the Q2e output shows 4Vpp with a DC offset of 5V. The waveform is distorted, and I suppose this is no surprise since Q2b is getting driven with too much voltage. Adding in Q3 and a load produces just so much random noise. Driving the speaker directly from Q2e doesn't do anything useful.

So clearly this is the wrong way to do it, and figure 16 on

formatting link
is bogus.

Evidently the original miswired circuit only worked because Q3 malfunctioned in a particular way.

Back to the drawing board.

Regards,

Uncle Steve

--
More than a century has passed since science laid down sound 
propositions as to the origins of the universe, but how many have 
 Click to see the full signature
Reply to
Uncle Steve

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.