Help with basic amp design

[CraigP]

Originally Posted by RJM1

CraigP, in the circuit you posted above Q3 is in upside down. The emitter should be connected to R5.

Ah! That explains things. I'll retest tonight. Thanks for calling that out!

[CraigP]

Originally Posted by Mooly

From the left...

You experimented with the values By keeping the ratio of Rf and Rin the same then the voltage gain stays the same. Rin determines the actual input impedance.

I think I'm getting the idea of input/output impedance. It seems to be a real game trying to get as much impedance on the input as possible to make the output from the prior section efficient. But then the last one, that drives a tiny 8ohm load, has to stand that idea upside down. So it is a bunch of tweaking to get the efficiency where you want it?

Originally Posted by Mooly

Q1 and Q2 are connected as "emitter followers" to provide current gain to enable a low value load to be driven. By including them in the feedback loop (by moving Rf from the opamp output to the new output point) the circuit as a whole still follows the above rules.

I'll have to noodle this a bit to understand it. I'm trying to grasp the current gain concept. For some reason the voltage gain seems easier to understand for me right now.

Originally Posted by Mooly

Because the transisotors need around 0.7 volts across the base and emitter junction to turn them on there is a kind of dead zone around the centre part of the waveform. That's the glitch in the output. The opamp can't swing its output fast enough across this dead zone and so there is some distortion. The resistor R5 allows the opamp itself to supply the load during this region with the transistors taking over again as the output level increases.

I was actually very close on this but I didn't post it because I wasn't sure. I did notice the glitch was when the transistors "lost power"... I didn't put together that the opamp itself would then power the output directly. That's amazing. When the transistors are providing current gain, wouldn't *some* flow backwards to the base via the resistor?

Originally Posted by Mooly

C1 and C2 are needed to block DC and allow AC to pass. They do form a filter. The input impedance of the main circuit is always the value of Rin. If C1 were to small in value then low frequencies would be attenuated. The same applies to C2 and the 8 ohm load.

If you have Java try this putting in different values. You will see why C2 needs to be so large with an 8 ohm load and why C1 can be much smaller with a 10K "load" or input impedance.

High Pass Filter - Java Experiment

I'll run these tests tonight. I'd like to also build this out on my breadboard, any reason that wouldnt work? Someone said this has 1.5% distortion. Is that "really bad"? I know that is a very tough question. I'm just trying to get some idea of what it *should* sound like when I'm done.

Mooly thanks SOOO much for working with me. You've taught me a lot!

[Mooly]

Originally Posted by CraigP

I think I'm getting the idea of input/output impedance. It seems to be a real game trying to get as much impedance on the input as possible to make the output from the prior section efficient. But then the last one, that drives a tiny 8ohm load, has to stand that idea upside down. So it is a bunch of tweaking to get the efficiency where you want it?

Input impedance isn't so much of an issue as long as you understand the basics which is that the lower it is, the more of a load it places on the source component feeding the amplifier. So if you take a CD player or tuner or any source component like that and look at the specifications it will probably say something like "maximum output 2 vrms 600 ohm". An ancient cassette deck might be 700mv at 10K ohms.
Those figures mean it can supply that voltage across that load impedance. Make the impedance lower and the output voltage will fall away and possible distort too. So the input impedance has to be at least equal to and preferably higher than the minimum value the component can drive.

Another way of thinking of it... a large 12 volt battery can supply plenty of current into a load such as a bulb or motor. If you add a 10K resistor in series with the positive battery terminal there is still 12 volts on the end of that resistor but it won't now light a bulb or run a motor. The "output" impedance of the battery is now 10K and limits the current. Output impedance of a component is similar. So a 10K load will cut the 12 volts down to half and so on.

So we make the input impedance of most amplifiers high relative to the source. In practice that means 10K and above.

Originally Posted by CraigP

I'll have to noodle this a bit to understand it. I'm trying to grasp the current gain concept. For some reason the voltage gain seems easier to understand for me right now.

Well the opamp has current gain too, just not very much. About 10 to 20 milliamps is the most that common opamps can supply. So if the load on the opamp draws more than that, then the opamp just limits and the output voltage is reduced and distorted.

Originally Posted by CraigP

I was actually very close on this but I didn't post it because I wasn't sure. I did notice the glitch was when the transistors "lost power"... I didn't put together that the opamp itself would then power the output directly. That's amazing. When the transistors are providing current gain, wouldn't *some* flow backwards to the base via the resistor?

That glitch is called "crossover distortion" and is the number one problem with all class B and class AB amplifiers. The resistor across the base and emitter is a quick fix but it's not the best way. Current doesn't flow back because the opamp is always maintaining the bias in the "right" direction.

When a transistor is used as an emitter follower the emitter voltage follows the base voltage less around 0.7 volts. The transistor has a "high" input impedance on it's base and low output impedance on its emitter. The current gain is determined by the property called "hfe" and for a small transistor is around 100 or more. That means with 1 milliamp flowing into the base and emitter junction, a curent of 100 milliamps will flow from collector to emitter. The emitter current would actually be 101 milliamps as Ie = Ic + Ib

Originally Posted by CraigP

I'll run these tests tonight. I'd like to also build this out on my breadboard, any reason that wouldnt work? Someone said this has 1.5% distortion. Is that "really bad"? I know that is a very tough question. I'm just trying to get some idea of what it *should* sound like when I'm done.

Absolutely... get it built, get it working, then it can refined and elaborated on. Is it bad ? Try it and see.

Originally Posted by CraigP

Mooly thanks SOOO much for working with me. You've taught me a lot!

No problem... and it give me a chance to play with LTspice which is a bit of a steep learning curve.

[Mooly]

Originally Posted by AndrewT

Pity, St Andrew's let's the side down, by spelling "through" as thorough and then letting the spell checker accept it.

I always have to think when I see the old Leak troughline... and I just had to Google that and found I'd spelt it wrong because I always think it must be throughline. It's trough as in pig trough though. Quiet and quite is the other one that crops up all the time.

[CraigP]

So I need to put together an order for parts to build these two circuits. I have the opamp (a 741) and BJT (2n3904/2n3906) but not the emitter follower transistors. I'm looking for suggestions for a few other things to buy that you folks think I might quickly find I want (upgrades) as I'll have to pay shipping and need to get my order up over 10$ anyway. Any tips would be great. I'm thinking:

10 - 2n5401
10 - 2n5550
10 - 10uF caps (ceramic?)
10 - 1000uF caps (ceramic?)

I was thinking of getting some of the TIP31 and TIP32C units also as they are used in RJM1's post.

I have a few LM741s - is there a "upgraded one" I might want to order also?
I also have a large assortment of 5% 1/4 and 1/8 watt resistors. The reason for the capacitors on the order is all of mine are the electrolytic and they are strange values compared to what these circuits are using.


Thanks all!

[Mooly]

Electrolytic caps in 10, 47, 100 uf are always handy. Depending on what voltages you are working perhaps aim to get 63 volt working. Small film type caps in smaller sizes are useful. Say 0.1uf and 0.22 uf. Anything over these sizes are going to be electroylitic for the circuits you are looking at.

Opamps, it's worth getting a couple of TL071 FET ones.

Transistors as you mention plus say BD131 and BD132 and TIP41 and TIP42. These are medium and high power devices... a bit old but easy to work with and easily available.

If and when you build a specific design, then that is the time to think of getting specific parts, otherwise common generic parts will allow most circuits to be built and experimented with.

[CraigP]

Thanks again. My order should arrive today (tomorrow at latest) so I'm excited to build. I was thinking through this in my head today and realized I dont have a volume control.... As I'm testing I may not want this blasting at full volume all the time. Any tips to satisfy this? Or should I control the volume via the input device (ie. blackberry) for now and wait to get into a volume control later?

[nigel pearson]

I learnt most from JLH .

http://www.tcaas.btinternet.co.uk/jlh1969.pdf

STEREO CLASS A 10W AUDIO POWER AMPLIFIER KIT! JLH 1969 | eBay

[Mooly]

You can use the input device for volume control or use fixed resistors at the input to make a divider. There are reasons why a volume control is normally better though. The output of most devices with software controlled output deteriorates at low level while the noise stays the same. So if you amplify that you get a poorer output than setting the device to full output and then attenuating it with volume control which gives no real loss of quality.

[nigel pearson]

Molly . Item 31 did you ever get it to sound nice ? I have used it for motor drives . There is a classic design like that which is used to teach current dumping ( Quad 405 ) . Could you simulate it becasue I don't have one using 100 R feed forward resistor ( not 1 K ) and a 3K pull up resistor to - V supply from the op amp output ( pull down if - ? ) . I've got a hunch it will work and it it does I will tell you a story dating back to 1977 . If you have time that is . Looking at 0.01% distrotion I think if it does what I suspect .