TL071 distortion - bad amplifier design?

[Mechatrommer]

i thought its class AB near 0V (with the existence of 2 biasing diodes) and B when not.
https://www.electronics-tutorials.ws/amplifier/class-ab-amplifier.html

why class C came into picture? its not suppose to be in audio anyway?
http://www.circuitstoday.com/class-c-power-amplifier

this thread seems to follow the fate of the others... as usual..

[Kleinstein]

There is no need to use a large, expensive polypropylene capacitor at the input. Just a more normal polyester capacitor (PEN or PET) is well good enough.
For the 220 µF a more normal electrolytic cap is likely OK too.

If DC coupling is used, it might be a good idea to adjust the offset - at least with an TL071. It may work with other lower offset OPs.
Even with the capacitor at the input one might consider adjusting the offset in a way to get the correct polarity for the 220 µF capacitor.  Over a long time electrolytic caps may not like reversed polarity, even if only a few mV.

The distinction between class B and AB audio amplifiers is normally from the bias / standing current: class B would be without a standing current and class AB would be with some bias current at low output level, but less than the normal load current. Ideally there would be a small bias, so that in the cross over region the slow onset of operation from both halves add up - so no gm-doubling. There is still the problem of staying stable at that point of operation with changing temperature - so the coupling of the VBE multiplier to the transistors can be important.

The output stage uses a darlington configuration. So it would take about 4 diodes to get class AB.

[Mark Hennessy]

why class C came into picture? its not suppose to be in audio anyway?
http://www.circuitstoday.com/class-c-power-amplifier

As your link explains, the definition of a class C power is one where the output devices are conducting for less than 180 degrees.

If you take a class B audio amplifier and turn down the quiescent current, you will get crossover distortion. This is the effect of the output devices conducting for less than 180 degrees. Hence, the amplifier has moved from class B to class C. As you rightly say, this is not good.

Note that the quiescent current does not tell you what class of amplifier you have. This is a common myth - I'm not really sure where it comes from. In short, transistors are not binary devices, so require a particular standing current to reach class B. Slightly less bias moves into class C, and slightly more will move into class AB. Lots more standing current will move the amplifier into class A (in practice you'd choose a value that fits with the power output and minimum load impedance). All that without changing the circuit; simply adjusting a variable resistor.

It's all in the book I referred to, which is a very readable yet thorough analysis.

The distinction between class B and AB audio amplifiers is normally from the bias / standing current: class B would be without a standing current

Not so; class B requires some bias to achieve 180 degree operation. No standing current means class C 

There is no need to use a large, expensive polypropylene capacitor at the input. Just a more normal polyester capacitor (PEN or PET) is well good enough.

Agree completely. Coupling capacitors lead a simple life - obviously it's good to avoid electrolytics when forming a high-pass filter as they are known to cause distortion in that role, but an ordinary polyester is more than good enough. The VAS capacitor is much more critical, but easily overlooked.

[Bassman59]

The stage following the op-amp is inverting, so the feedback to the + input of the op-amp is correct.

no.

Yes.  The drive stage after the op-amp is inverting, so the feedback connection to the non-inverting op-amp input is correct.

You guys are all correct -- I missed how the drive stage was connected. I even simulated the thing.

Ya learn something new every day, if you're willing to do so.

Thanks.

[spec]

There is no need to use a large, expensive polypropylene capacitor at the input. Just a more normal polyester capacitor (PEN or PET) is well good enough.
For the 220 µF a more normal electrolytic cap is likely OK too.

What does 'there is no need' and what does, '(PEN or PEN) is well good enough' mean?
Capacitors have a big impact on the sound of an amplifier- I had a good reason for specifying those components.
A suitable polypropylene capacitor costs 70p UK and is not excessively big.
As I have said to you before, let us see your design for an amplifier, rather than criticizing/contradicting other people's work.

[AngraMelo]

Spec, the only polyproplyne 1uf cap I can find in my country are these types: https://proesi.com.br/capacitor-partida-cbb61-1uf-450vac-fio.html

They seem to be for mains filtering. As they are 450v would impact negatively on the ESR or other parameter?
About the 220uf Tantalum, I think it is better for me to get a bunch of 100uf and I can combine them in series/parallel and find the sound I like.
What is the voltage rating I need for it? A 6.3V or 2V would do?

The main idea for this amp is to be a good sounding amp. I know it sounds vague but Im a musician and I have plenty of high fidelity amps around. From studio monitors to specialized stuff. Im diving in the audio amplifier world as a way to learn electronics and to be able to put together a amp that sounds the way I like, not that is 0.00000001% distortion.

I dont have a need for high power amps, given that I listen to music in a small space, so when I heard you telling me to choose the capacitors to suit the sound I like I got really excited because that is precisely the way Im tackling this subject.

Yes, it will be great to learn how to make extremely low distortion amps and maybe I will pursue that in the future. But right now, what Im looking for is finding the amp configuration/components/class that best suits my ears and speakers.

[spec]

UPDATE #1  2018_12_12

As to the class of an amplifier, there is no confusion at all:

• In class A both output elements conduct all the time. This is, by definition, the lowest distortion class of all, but at the cost of low efficiency and thus high waste heat.
• In class A/B both output elements conduct for small signals, but move over to class B for larger signals.
• In class B one output element conducts for the positive half of the waveform (180 deg) and the other element conducts for the negative half (180 deg) of the waveform. AFAIK there are no low distortion class B audio amplifiers. In fact there would be little point, because class lowA/B (conventional audio amps) are so simple to implement and have essentially the same efficiency as class B, without the high crossover distortion.

Class AB  amplifiers predominate, because they give good efficiency with reasonably low distortion. Typically, a class AB amplifier will have a quiescent current of 20mA to 40mA, because this gives the lowest crossover distortion. Another approach is to have a quiescent current of 100ma to 1A. This would still be a class A/B amplifier but would operate in class A for a greater part of the signal. Guitar valve amps are nearly always class hiA/B.

Incidentally you can change the crossover point on most class A/B amplifiers, by simply adjusting the quiescent current. Of course, you need to increase the heat-sinking to get rid of the extra static heat dissipation.

You can even turn a class A/B amplifier into pure class A by increasing the quiescent current to a high value. Much more heat-sinking, would of course be required. For example, you could turn the reply #44, 72W class A/B amplifier into pure class A by increasing the quiescent current to 4A375.This wold require a change of output transistors and vastly increased heat-sinking, but that does not detract from the principle.

I have often run a standard Douglas Self amplifier in all three modes: A/B hiA/B and A.

By the way, it is dead easy to make a class B amplifier with no quiescent current: just feed the bases of a complimentary emitter follower output stage with a constant current. In fact the voltage amplification stage (VAS) of most audio power amplifiers do produce a constant current. That is one of the major objectives of most VASs.

In practical terms the absolute value of quiescent current is not that critical, so long as it does not drop below around 20mA. If it were essential to accurately fix the quiescent current exactly to a particular value, it would be done, because it is not a difficult feature to achieve.

[spec]

Spec, the only polypropylene 1uf cap I can find in my country are these types: https://proesi.com.br/capacitor-partida-cbb61-1uf-450vac-fio.html.

There are polypropylene capacitors available for small signal use. I will advise on this.

They seem to be for mains filtering. As they are 450v would impact negatively on the ESR or other parameter?
About the 220uf Tantalum, I think it is better for me to get a bunch of 100uf and I can combine them in series/parallel and find the sound I like. What is the voltage rating I need for it? A 6.3V or 2V would do?

I will change the schematic to show a 100uF Tant, if that is what you can get. 100uf will be more than adequate.
2V will be fine, as will any higher voltage. The objective anyway, is to get rid of the capacitor altogether. The tant capacitor is one of the biggest sources of distortion/coloration because it is part of the feedback network. In general capacitors are troublesome components, so it is always a good approach to get the best you can, consistent with cost/size. Of course, the best approach is to eliminate capacitors in the signal path whenever possible.

[AngraMelo]

Awesome! Thank you again!

[spec]

The main idea for this amp is to be a good sounding amp. I know it sounds vague but Im a musician and I have plenty of high fidelity amps around. From studio monitors to specialized stuff.

nice

I'm diving in the audio amplifier world as a way to learn electronics and to be able to put together a amp that sounds the way I like, not that is 0.00000001% distortion.

I don't have a need for high power amps, given that I listen to music in a small space, so when I heard you telling me to choose the capacitors to suit the sound I like I got really excited because that is precisely the way I'm tackling this subject.

OK, that gives a good picture of how perceptive you are, and the audio quality standard you would be happy with.

The amplifier of reply #44, with good +-40V supply lines will produce 72W into 8 Ohms. Which is ideal for music reproduction in a small room with modern speakers, which tend to be rather inefficient. If 72W sounds a lot, it is not because music normally consists of low level base-line signals with huge peaks every so often. So while the average power at fairly loud listening levels may be 100mW, the peaks could easily reach 72W instantaneous (theory says 1kW). Don't forget that the human ear's response to sound is logarithmic so, for example, increasing the power by 10%(+1dB) is barely perceptible. 

Yes, it will be great to learn how to make extremely low distortion amps and maybe I will pursue that in the future. But right now, what Im looking for is finding the amp configuration/components/class that best suits my ears and speakers.

It is extremely persnickety and tiresome to make a top end audio amp. Much of it is mechanical: the thickness and kind of wire and solder. The order of the connections. etc. And there are capacitors splattered all over the place- just take a look at a picture of the innards of a high end amp.

So I would advise this as a theoretical study for the future.

To summarize then, I suggest that we get the amplifier of reply #44 working well so there is a reference and reliable test bed and after that stage, which will be no mean achievement, look into improvements, like changing the opamp. Of course, if anything simple crops up on the way we can incorporate that.

[AngraMelo]

That is great Spec!
That is exactly what Im looking for, im ordering the capacitors and preparing a board for the new transistors. It will be awesome!
Ill let you know of the performance and measurements I get here.

[spec]

That is great Spec!
That is exactly what Im looking for, im ordering the capacitors and preparing a board for the new transistors. It will be awesome!
Ill let you know of the performance and measurements I get here.

Glad that you like that approach because, I think it will give the best progress.

By the way, did you set up the quiescent current Iq OK?

[spec]

[I am] preparing a board for the new transistors. It will be awesome!

Can I just say this to you. The physical layout of practically all circuits has a geat impact on their performance, and with high end audio amplifiers more so.

If you look at the schematic of reply #44 you will notice that I have changed the supply line and 0V connections to indicate how they should be physically implemented. Also the reply #44 schematic is only an outline and needs a few more components: decoupling capacitors for example.

So may I suggest that you hold the board layout and I will produce a full schematic for you to work from. The schematic will look a bit weird, but don't let that put you off. When the schematic is posted we can then discuss the board layout.

Just to illustrate how important simple things are with audio amps, you can take the most stable and best sounding amp ever and just by changing a couple of 0V connections, turn it into a temperamental distorting monster.

Apart from the OV line, which is crucial, remember that a class A/B amp is inherently distorting and, as a consequence, the output transistors take huge gulps of current from the supply line for 180 degrees of the signal and no current for the next 180 degrees. And, due to various effects, this current is a rather distorted half sine wave.

Wires and PCB traces are not perfect conductors and have resistance, so the distorted half sine wave current gulps produce a corresponding distorted voltage waveform on the supply lines. It only takes a whiff of that voltage to get induced into the amp circuitry to cause high levels of distortion.

[AngraMelo]

I just got home so I didnt have time to change de resistor and pot. I will do that tonight.
I have no education about pcb layout so any help would be greatly appreciated!!

[spec]

I just got home so I didn't have time to change de resistor and pot. I will do that tonight.

No sweat. By the way most small signal BJTs will be suitable for the Iq setting transistor, including BC337 family, BC546 family, BC182 family, BC107 family, etc The transistor only needs a VCE of 10V  minimum.

I have no education about pcb layout so any help would be greatly appreciated!!

This is the most important and difficult area. The best way forward is to take a look at some PC layouts that are already done for high end amplifiers and the schematic I mentioned previously should be a help.

[AngraMelo]

I will do that. It has been an incredible journey learning electronics.
Sometimes I feel that everyday is a kick in the butt given that the content is so extent.
Slowly Im getting the hang of it!

[spec]

I will do that. It has been an incredible journey learning electronics.
Sometimes I feel that everyday is a kick in the butt given that the content is so extent.
Slowly Im getting the hang of it!

I think everybody feels like that as a newbee- I certainly did. One of the most important things is getting familiar with the basics: essential math, Ohms law, and Kirchhoff's two laws are the foundation. Then you need good teachers, and good books and videos. But most of all, you need a logical mind and to be able to make mistakes and learn from them.

The trouble is, when you are young there are so many distractions: booze, music, girls, motor bikes, cars...

[spec]

I mentioned the importance of the physical layout of audio amplifiers (or any amplifiers for that matter), so attached is the schematic for a simple class A/B audio amplifier I did some years ago. Notice the connections of the power line, 0V line and the decoupling. That is the sort of schematic I will do for the reply#44 audio power amplifier.

Just as important is the layout of the PSU, so I will include a PSU schematic too.

[AngraMelo]

Spec, I soldered the resistor and pot correctly nothing I do brings the current between +rail and 5200 Collector above 0ma

[Zero999]

The proposed modifications by spec are good: adding a V_BE multiplier and a constant current source to make the bias point less dependant on supply voltage variations and only using one common emitter amplifier.

The value of the 220μF capacitor can easily be reduced without affecting the sound quality. The input capacitor is 1μF and the input impedance is 22k, which will dominate the lower cut-off frequency, long before the 1k and 220μF. See the calculations below:

F_C = 1/(2πCR)

C = 1μF and R = 22k
F_C = 1/(2π*1×10^-6*22×10³) = 7.2Hz

C = 220μF and R = 1k
F_C = 1/(2π*220×10^-6*1×10³) = 0.72Hz

The 220μF capacitor could be reduced to 22μF. I think the value was chosen due to a calculation error, involving the decimal place, which is very easily done!

And does it have to be solid tantalum? I don't see why aluminium won't do.

To solve the biasing issue, just connect the 22μF capacitor to -15V.


I doubt the input capacitor is that critical, just don't use a ceramic capacitor: anything else should be fine. There are many myths surrounding capacitors and audio some of which are fact, others are fiction, The main issue is ceramic capacitors which can be microphonic and introduce distortion, but all film types such as: polypropylene, polyester and and Polyethylene terephthalate (PET) are good.

Another possibility is to connect two aluminium or tantalum electrolytic capacitors back-to-back, forming a non-polarised capacitor, with half the value of the two capacitors.



For the 1μF capacitor, use two 2.2μF capacitors.

I thought the board was already done? Spec's modifications are good, but will be quite tricky to perform on an existing board. If the board can be redone then great!

Spec, I soldered the resistor and pot correctly nothing I do brings the current between +rail and 5200 Collector above 0ma

What's the voltage across each of the 100R resistors, when the input is shorted to 0V?

Did you implement all of spec's changes or only some of them?

You need to add the BC546 and other 1k resistor as well as the 1k pot for this to work.

[spec]

Spec, I soldered the resistor and pot correctly nothing I do brings the current between +rail and 5200 Collector above 0ma

Very strange. I will go back and check the circuit values.

[AngraMelo]

Hero999,
Great tips!
About the mod with the pot and 1k resistor:
Im implementing the changes as they came on the thread. Im now following the attached schemaitc

[spec]

CROSSED POST

The circuit and values are OK.

Are you using the new design with the PNP transistor in the emitter of the top PNP VAS transistor? Whichever way there should be 10mA flowing out of the PNP VAS collector.

The voltages on the VBE multiplier transistor (BC337) should be:
Collector 1V2
Emitter  -1V2
Base      -0V6

What are you getting?

[spec]

About the mod with the pot and 1k resistor:
Im implementing the changes as they came on the thread. I'm now following the attached schematic

That may be the problem. You need to embody the latest circuit mods as shown on the reply #44 post.

[AngraMelo]

Spec, I did only the following modifications:
- Changed the 2 4.7nF for 2 47uF capacitors
- Added a BC337 between the colletors of both VAS transistors
- added a 1k resistor between 1015 collector and BC337 base
- added a 1k pot between 1815 collector and BC337 base
- added a 100uf cap accross the BC337 collector and emitter

I believe that is all that is included in the schematic I posted on reply#71, am I mistaken?
Do I need to short the input to measure the current?

Hero999
The board with the new transistors is not yet made. Spec will make a layout for me and Ill etch it here.
Im bodging the components I described above on the existing board.