Back in 1968, Arthur Bailey used an "amplified diode" to set the quiescent current in the output stage of his famous Bailey amp. That is nothing more than a reasonably high gain NPN transistor with a resistor from collector to base and another from base to emitter. Varying the ratio changes the voltage drop at which the transistor starts to rapidly conduct. While experimenting, you can use a potentiometer in place of the two resistors and trim the voltage across it. That allows you to replace the two diodes, so you don't have to go messing about with one small signal and one Schottky, etc. which is a rather crude way of varying the bias voltage and hence the quiescent current in the output stage. The transistor will thermally track like the number of diodes that it is emulating, be it x1.5 or x2 or whatever. I've found it a good way to examine the effects of quiescent current on crossover distortion as you can continuously vary the quiescent current and watch the crossover distortion on the FFT display of your scope. I would recommend using a current-limited power supply just in case you try to turn up the quiescent current too much.
Hey John, one more nice and out-of-the-box reasoning... always a pleasure. Since you chose not to have a stand-alone biasing network and thus relying on the Global Feedback to stabilize the bias as it is, one good way (and indirect) of looking at how good the output crossover distortion is, is by measuring the signal at the output of the op-amp. If you see a noticeable inverse crossover distortion there, than it is an indication the op-amp is doing the most in correcting the x-over distortion, whereas if you find little inverse x-over distortion at the op-amp's output, then you can conclude the output biasing is good enough on its own.
The LM4562 is a very fast opamp with a typical GBWP of 55 MHz and a slew rate of 20 V/µs. So it is absolutely no problem for this device to eliminate crossover distortions of underbiased output stages. It should even be possible to operate the Sziklai pair without a bias spreader in pure class B mode.
@@gkdresden I never meant to go down that rabbit hole. I didn't even know the 4562 is that fast... I just wanted to point out that, if it's the op-amp who is working the x-over out (iso the x-over being linearized locally at the output stage itself), then you will see its output crossing the zero line at high speed, indicating the output stage isn't properly biased. The more linearized the output stage is, the less the op-amp will need to work itself out to counteract the lacking of the output stage. Now, on your assertion that the 4562 is so fast that we shouldn't care as it would be more than capable of working out the x-over... Well, I don't think it's that simple... The op-amp may be fast, but the output transistors would need to be fast too, to cross that non-linear region so quickly not to leave any swiching artifacts getting through... and one doesn't want an E.F. output stage being any faster than just enough the freq. response specs require, if one doesn't want to deal with nasty E.F. oscillations (which are already prone to it all in its own). That is - of course - my opinion.
@@fabiotrevisan8922 The limitation are the 2N3055 / 2N2955. But their GBP is still 2.5 MHz. But this is still fast enough to cope with cross-over distortions, especially for small signal amplitudes and frequencies of 1 kHz. I guess it is even enough to cope with full RMS power at 20 kHz. John can perhaps test this for us. For power MOSFETs this is certainly a problem, because they have usually larger input capacitances and significant gate charges.
Hey@@gkdresden , you surely have some strong points. I for myself, don't believe "slow" devices such as the 2N3055 would be able to take nicely the step transition, especially in the transition from active to cutoff. It would surely surface as some sort of odd harmonics distortion but, in favour of your argument, I grant we're not talking about high-end HiFi here anyway. I think it will still perform better if the output stage is biased. If there's anyone who can settle this argument, it would be @Johnaudiotech, if he's willing.
@@fabiotrevisan8922 there were harmonics visible in the spectrum, and for sure, it is better to operate the amplifier in the class AB mode with a certain amount of bias current. btw: I have found, that it is convenient to keep the unused transistor at a certain amount of bias current after crossover. I have used the LT1166 for this purpose. www.analog.com/media/en/technical-documentation/data-sheets/1166fa.pdf The circuit suggestion in the datasheet shows more effort than really required.
Thanks for the video John! I would also be interested to see it done with Mosfets (or some version of it with Mosfets) Glad to hear that Snickers is doing better. :)
would just like to say your vids and explanations are excellent even with so.e experience I still find somethingto take away so thank you, and anyone giving a thumb down are obviously on the wrong channel.
Always makes my day when I see an upload from you, I'm glad you have put stuff about the JAT501out there, I'm going to start a layout for fun, have you thought anymore about designing a PCB yourself yet ? ...cheers.
Wonder how that output would have looked like if not included in that feedback loop. Pretty impressive crossover zone for such a simple and low-biased amp. Would have liked to see emitter resistor voltages for actual idle current, though of course it must be most of that 20mA.
" of course it must be most of that 20mA" this is not. The LM4562 consumes 10 to 12 mA when running idle for both channels and another 6 mA flow from the positive rail through the LM4562 output, the bias spreader diodes and the boot strap resistors to the negative rail. I think the are less than 2 to 4 mA of bias current through the Sziklai pairs of the output stage.
@@gkdresden Vt/Re = Ib Thermal voltage divided by emitter resistance equals bias current of the emitter follower output - theoreticaly it gives lowest possible distortion in this stage when voltage dropped across emitter resistor equals thermal voltage (~0.026V in room temperature).
I have questions- 1. What will be the difference between using a transistor (s) or opamp before push pull output stage ? 2. Does using an audio quality transistors at output make any big difference than using normal transistors ?
600 Ohms stereo headphones you can drive by use of a LM4556 dual op amp directly. You can even drive them in class A mode by conneting J113 junction FETs between the op amp outputs and the negative rail, allowing for a constant bias current of about 20 mA.
Hello John (sorry for my bad english i'm French) I don't understand, what's the interest of the capacitor, right before the speaker ? please..thank you
Often I don't and just build the circuit. When I do, I use LT Spice such as when I designed the JAT501 amp and the EZ Amp. I may make a video on LT spice but there are already good tutorials.
@@JohnAudioTech indeed, but I think people would enjoy and learn from a tutorial from you, you have a great style and 'come across' very well !...cheers.
I can recommend LTSpice for audio circuit simulation, because it is very powerful, easy to use and there is a lot of literature and tutorials availlable over the internet. But there is not much stuff about audio circuit simulations and maybe John can help us out with some special videos about audio amplifier circuits and how modifications influence / improve their behavior in simulation compared to measurement.
If I want to mix the outputs of 4 microphones and one audio input from any source like a mobile phone to a single or dual output , then how should I do that if I am using opamp ?
Hey John I absolutely love your videos.. but this time I'll give you an offtopic question still related to audio. I have a old reel to reel player and in the manual they specify for low noise transistors for the read head amplifier. Unfortunately I have no idea how to measure such things neither quantifyably nor relatively. I would like to figure out how to do it properly though. Thank you a lot.
Noise depends on bias currents, resistor values and so on. It's a lot to get into. You could setup a similar stage feeding a low noise opamp and measure the difference in noise of transistors. The KSC1845 and its complement, the KSA992 are well known low noise transistors that may work.
@@JohnAudioTech the original transistors are supposed to be KC149C. They are essentially BC547C. High gain low power. I have the schematic and I could construct the first stage of the preamplifier on a breadboard and maybe experiment with it if you're saying it's highly dependant on where the transistor is biased. Thanks for the recommendation
What would be the best way to protect output transistors on a small amplifier ? OK so I am making a small amplifier with vintage nos transistors mostly for fun but I want to protect these vintage transistors from short circuit or overcurrent situations so what would be the best protection circuit for s small amp that is about 3w per channel. Transistors are rated 1 amp as far as I remember. Thanks in advance.
There's hardly any headroom to protect a 3w amp using 1a transistors. Any current limiter circuit can affect the sound with reactive speaker loads. Perhaps a thyristor circuit that shuts down the signal until power is cycled can protect them. If it's an amp only you use, just be careful not to short the outputs.
@@JohnAudioTech Yeah that's what I thought... If I put any current sensing resistor it will drop too much voltage for such a small amp. So yes probably no protection circuit is the best since this is more for fun and i am not selling it. Thanks for the response.
I wouldn't mind sending you a package or I've even seen "fujipoly XR-m". Their spec sheet shows to have more thermal conductivity and electrical insulation than mica and thermal paste. Never used it but it seems worth it on upgrading something people would want to keep for a long time.
It has to connect. But the power ground should go straight to the psu, while signal ground should go between preamplifier (that also had a power ground connection to psu) and power amplifier. Speaker return MUST go straight to psu ground.
To: John Audio Tech re: JAT501 amp I searched for this and came across the schematic at OSHW, authored by Blasttherost. The schematic has gross errors! The PNP power transistor is upside down! Other errors, too. Please do us all a favor. Have the author remove your name from the schematic and have him correct the errors. I don't believe it is the same schematic as yours. He has misappropriated your name. Thanks.
![KSI - Dirty [Official Visualiser]](http://i.ytimg.com/vi/VoDh1h1dPUE/mqdefault.jpg)
Glad to see Snickers recovering.
Hello John, you should make an Amazon public wish list for things that would help you make videos. A way to let us know what you need.
May be some volunteers run some boards, now very cheap with China PCB manufacturers.
Nice Work Earthling
Bless Up
Back in 1968, Arthur Bailey used an "amplified diode" to set the quiescent current in the output stage of his famous Bailey amp. That is nothing more than a reasonably high gain NPN transistor with a resistor from collector to base and another from base to emitter. Varying the ratio changes the voltage drop at which the transistor starts to rapidly conduct. While experimenting, you can use a potentiometer in place of the two resistors and trim the voltage across it. That allows you to replace the two diodes, so you don't have to go messing about with one small signal and one Schottky, etc. which is a rather crude way of varying the bias voltage and hence the quiescent current in the output stage. The transistor will thermally track like the number of diodes that it is emulating, be it x1.5 or x2 or whatever. I've found it a good way to examine the effects of quiescent current on crossover distortion as you can continuously vary the quiescent current and watch the crossover distortion on the FFT display of your scope. I would recommend using a current-limited power supply just in case you try to turn up the quiescent current too much.
Hey John, very good explanation as always!
Hey John, one more nice and out-of-the-box reasoning... always a pleasure.
Since you chose not to have a stand-alone biasing network and thus relying on the Global Feedback to stabilize the bias as it is, one good way (and indirect) of looking at how good the output crossover distortion is, is by measuring the signal at the output of the op-amp.
If you see a noticeable inverse crossover distortion there, than it is an indication the op-amp is doing the most in correcting the x-over distortion, whereas if you find little inverse x-over distortion at the op-amp's output, then you can conclude the output biasing is good enough on its own.
The LM4562 is a very fast opamp with a typical GBWP of 55 MHz and a slew rate of 20 V/µs. So it is absolutely no problem for this device to eliminate crossover distortions of underbiased output stages. It should even be possible to operate the Sziklai pair without a bias spreader in pure class B mode.
@@gkdresden I never meant to go down that rabbit hole.
I didn't even know the 4562 is that fast... I just wanted to point out that, if it's the op-amp who is working the x-over out (iso the x-over being linearized locally at the output stage itself), then you will see its output crossing the zero line at high speed, indicating the output stage isn't properly biased.
The more linearized the output stage is, the less the op-amp will need to work itself out to counteract the lacking of the output stage.
Now, on your assertion that the 4562 is so fast that we shouldn't care as it would be more than capable of working out the x-over...
Well, I don't think it's that simple...
The op-amp may be fast, but the output transistors would need to be fast too, to cross that non-linear region so quickly not to leave any swiching artifacts getting through... and one doesn't want an E.F. output stage being any faster than just enough the freq. response specs require, if one doesn't want to deal with nasty E.F. oscillations (which are already prone to it all in its own).
That is - of course - my opinion.
@@fabiotrevisan8922 The limitation are the 2N3055 / 2N2955. But their GBP is still 2.5 MHz. But this is still fast enough to cope with cross-over distortions, especially for small signal amplitudes and frequencies of 1 kHz. I guess it is even enough to cope with full RMS power at 20 kHz. John can perhaps test this for us. For power MOSFETs this is certainly a problem, because they have usually larger input capacitances and significant gate charges.
Hey@@gkdresden , you surely have some strong points.
I for myself, don't believe "slow" devices such as the 2N3055 would be able to take nicely the step transition, especially in the transition from active to cutoff.
It would surely surface as some sort of odd harmonics distortion but, in favour of your argument, I grant we're not talking about high-end HiFi here anyway.
I think it will still perform better if the output stage is biased.
If there's anyone who can settle this argument, it would be @Johnaudiotech, if he's willing.
@@fabiotrevisan8922 there were harmonics visible in the spectrum, and for sure, it is better to operate the amplifier in the class AB mode with a certain amount of bias current.
btw: I have found, that it is convenient to keep the unused transistor at a certain amount of bias current after crossover. I have used the LT1166 for this purpose.
www.analog.com/media/en/technical-documentation/data-sheets/1166fa.pdf
The circuit suggestion in the datasheet shows more effort than really required.
When talking about the cat being old with a crusty nose, for a minuet or so, I thought you was talking about me!
Being my 50s, I'm right behind you!
I'm glad to see Mr. Snickers!!!
Odd fact: This is exactly what I'm contemplating for my own build project, with BD139/BD140 for outputs. Headphone amp project. Thanks. 👍
whomever invents an intuitive way to search YT videos with accurate results should become a millionaire.
Thanks for the video John! I would also be interested to see it done with Mosfets (or some version of it with Mosfets) Glad to hear that Snickers is doing better. :)
would just like to say your vids and explanations are excellent even with so.e experience I still find somethingto take away so thank you, and anyone giving a thumb down are obviously on the wrong channel.
Excellent, got to love Snickers
Always makes my day when I see an upload from you, I'm glad you have put stuff about the JAT501out there, I'm going to start a layout for fun, have you thought anymore about designing a PCB yourself yet ? ...cheers.
Nice one John! What about a signal pass capacitor across the diodes? May improve the distortion?
Always watching master
The problem is you have some of your resistors in backwards. They should all read from left to right.
Oh, this triggered my OCD
@@noelj62 with backwards resistors you only get the dyslexic electrons through them.
Lucky it didn't blow up like that. Negative resistance! 😏
@@worstuserever well we are trying to achieve amplification. So perhaps that is John's secret?
Wonder how that output would have looked like if not included in that feedback loop. Pretty impressive crossover zone for such a simple and low-biased amp. Would have liked to see emitter resistor voltages for actual idle current, though of course it must be most of that 20mA.
" of course it must be most of that 20mA"
this is not. The LM4562 consumes 10 to 12 mA when running idle for both channels and another 6 mA flow from the positive rail through the LM4562 output, the bias spreader diodes and the boot strap resistors to the negative rail. I think the are less than 2 to 4 mA of bias current through the Sziklai pairs of the output stage.
There is almost no current going through that, like 2mA, while by my calculation with those emitter resistors it should be biased to around 260mA 😉
I'll have to scope the error signal. 0 volts across the emitter resistors.
@@akkudakkupl how do you calculate the required bias current? How does it depend on the emitter resistor?
@@gkdresden Vt/Re = Ib
Thermal voltage divided by emitter resistance equals bias current of the emitter follower output - theoreticaly it gives lowest possible distortion in this stage when voltage dropped across emitter resistor equals thermal voltage (~0.026V in room temperature).
Hi John, igbt on output is a good idea? You can drive more current, with the same supply, I think...
No crossover distortion... Hmmm. Is that a function of the cfp configuration, I wonder? Cancelling distortion via local negative feedback?
I have questions-
1. What will be the difference between using a transistor (s) or opamp before push pull output stage ?
2. Does using an audio quality transistors at output make any big difference than using normal transistors ?
John can we use tip140and tip145 transistor in this circuit
hi john will this amplifier will be good for driving 600 Ohms headphones? if not will we get a suitable hi-fi circuit?
600 Ohms stereo headphones you can drive by use of a LM4556 dual op amp directly. You can even drive them in class A mode by conneting J113 junction FETs between the op amp outputs and the negative rail, allowing for a constant bias current of about 20 mA.
Hello John (sorry for my bad english i'm French) I don't understand, what's the interest of the capacitor, right before the speaker ? please..thank you
Have you any plans to show how you simulate your work ? that would be nice one day...cheers.
Often I don't and just build the circuit. When I do, I use LT Spice such as when I designed the JAT501 amp and the EZ Amp. I may make a video on LT spice but there are already good tutorials.
@@JohnAudioTech indeed, but I think people would enjoy and learn from a tutorial from you, you have a great style and 'come across' very well !...cheers.
I can recommend LTSpice for audio circuit simulation, because it is very powerful, easy to use and there is a lot of literature and tutorials availlable over the internet. But there is not much stuff about audio circuit simulations and maybe John can help us out with some special videos about audio amplifier circuits and how modifications influence / improve their behavior in simulation compared to measurement.
@@gkdresden Cheers.
If I want to mix the outputs of 4 microphones and one audio input from any source like a mobile phone to a single or dual output , then how should I do that if I am using opamp ?
Are these D amplifiers' sound quality better than ordinary priced Panssonic or Sony CD walkman?
Snickers 😸
Hey John I absolutely love your videos.. but this time I'll give you an offtopic question still related to audio. I have a old reel to reel player and in the manual they specify for low noise transistors for the read head amplifier. Unfortunately I have no idea how to measure such things neither quantifyably nor relatively. I would like to figure out how to do it properly though. Thank you a lot.
Noise depends on bias currents, resistor values and so on. It's a lot to get into. You could setup a similar stage feeding a low noise opamp and measure the difference in noise of transistors. The KSC1845 and its complement, the KSA992 are well known low noise transistors that may work.
@@JohnAudioTech the original transistors are supposed to be KC149C. They are essentially BC547C. High gain low power. I have the schematic and I could construct the first stage of the preamplifier on a breadboard and maybe experiment with it if you're saying it's highly dependant on where the transistor is biased. Thanks for the recommendation
Hi, John!
Hi
What would be the best way to protect output transistors on a small amplifier ?
OK so I am making a small amplifier with vintage nos transistors mostly for fun but I want to protect these vintage transistors from short circuit or overcurrent situations so what would be the best protection circuit for s small amp that is about 3w per channel. Transistors are rated 1 amp as far as I remember.
Thanks in advance.
There's hardly any headroom to protect a 3w amp using 1a transistors. Any current limiter circuit can affect the sound with reactive speaker loads. Perhaps a thyristor circuit that shuts down the signal until power is cycled can protect them. If it's an amp only you use, just be careful not to short the outputs.
@@JohnAudioTech Yeah that's what I thought... If I put any current sensing resistor it will drop too much voltage for such a small amp.
So yes probably no protection circuit is the best since this is more for fun and i am not selling it.
Thanks for the response.
May I build it with LM301A and NE5534 for comparison?
Thank you
Does the "Thermal Grizzly Minus Pad 8" 0.5mm thick thermal pad work better than mica and thermal grease?
I wouldn't mind sending you a package or I've even seen "fujipoly XR-m". Their spec sheet shows to have more thermal conductivity and electrical insulation than mica and thermal paste. Never used it but it seems worth it on upgrading something people would want to keep for a long time.
Does the signal ground connect to the power supply ground or is it separate
It has to connect. But the power ground should go straight to the psu, while signal ground should go between preamplifier (that also had a power ground connection to psu) and power amplifier. Speaker return MUST go straight to psu ground.
To: John Audio Tech re: JAT501 amp
I searched for this and came across the schematic at OSHW, authored by Blasttherost. The schematic has gross errors! The PNP power transistor is upside down! Other errors, too. Please do us all a favor. Have the author remove your name from the schematic and have him correct the errors. I don't believe it is the same schematic as yours. He has misappropriated your name. Thanks.
Hey John, do you think a OPA2132PA would work on that config?
I'd have to see the datasheet. If it has enough current and good output swing, I don't see why not.
Why u playing with these china chips get some hifi stuff this stuff is junk