Ok here is something rather interesting. In fact i already build the circuit and it worked pretty well too. It was exactly based on ESP sound pages project 113 which was a headphone amplifier. Initially i build the exact circuit which was the headphone amplifier without altering anything else there. with input signal and a small speaker, it worked more than a headphone amplifier even able to drive small watts speaker too.
Link to the ESP project pages can be found here
https://sound-au.com/project113.htm
Getting digging further, i've changed the power transistor which was originally BD139 and BD140 to TIP35 and TIP36, following by increasing only the supply voltage to the output transistors and retaining the supply voltages to the opamp as usual, the amp worked as it supposed to be and seems now it able to drive higher wattage loudspeakers. Opamp were NE5532. On progress to upgrade to OPA2604 as they can take 24V positive and negative supply.
Now i am wondering is it a good idea to continue in this design path since now in its final form (TIP35 and TIP36 with +/- 24V supply) they worked excellent and what i do is properly bias the power transistors and give them a giant sized heatsink. If it does, why i didnt see this design is being commonly implemented? or if i were to use this concept, what else you all members suggest to upgrade and improve it? thanks
the photo shows with the large heatsink with the larger board is where the power stages sit ... the smaller one is the opamp circuits
Link to the ESP project pages can be found here
https://sound-au.com/project113.htm
Getting digging further, i've changed the power transistor which was originally BD139 and BD140 to TIP35 and TIP36, following by increasing only the supply voltage to the output transistors and retaining the supply voltages to the opamp as usual, the amp worked as it supposed to be and seems now it able to drive higher wattage loudspeakers. Opamp were NE5532. On progress to upgrade to OPA2604 as they can take 24V positive and negative supply.
Now i am wondering is it a good idea to continue in this design path since now in its final form (TIP35 and TIP36 with +/- 24V supply) they worked excellent and what i do is properly bias the power transistors and give them a giant sized heatsink. If it does, why i didnt see this design is being commonly implemented? or if i were to use this concept, what else you all members suggest to upgrade and improve it? thanks
the photo shows with the large heatsink with the larger board is where the power stages sit ... the smaller one is the opamp circuits
by the way is not my circuit anyway, i just modified the output transistor part and wonder it might work out that way
Enjoy it anyway. Several times happen to me believe I created a thing and googling patents it had alredy built 60 years ago.
Only want to add, take care with too long wires and cables, you may be creating an oscillator and the amp will make smoke. Or a shortcircuit.
Only want to add, take care with too long wires and cables, you may be creating an oscillator and the amp will make smoke. Or a shortcircuit.
oh ok thanks ... almost forgot the Zobel network at the output as i was in a hurry to hear how it sounded like 🤣
I suggest to replace the diode biasing for a VBE multiplier in order to achieve optimal bias for the ouput transistors.
It is limited by the fact that it cannot put out any more voltage than the op amp, even with increased rails to the output transistor. If you hear an improvement with higher rails, it is only because of failings in the output transistors (quasi-saturation, which radically drops gain at low Vce so it clips a bit early). It does however, work and pretty fool proof if you live within the limitations. For the original application (headphones), by the time it is clipping your head should be about ready to explode, and I doubt you’d be hearing anything other than pain.
“Improvements” can be made by using an output stage with voltage gain. Many of those can get horribly unstable and are tricky to work with. The easiest one to work with is the topology used in early QSC amps, but even they are subject to issues and are not totally foolproof. They turn the power supply inside out to force the output followers to run as a common emitter stage. They require separate power supplies for each channel and boot strapping the op amp supplies. If you don’t know what that is, study before proceeding. If interested it is WORTH the time/energy to study. To improve it has a headphone amplifier, remove the PNP output and replace it with a 100-300 mA CCS and run it single ended class A. Night and day. This even removes crossover artifacts in the op amp itself - and at headphone output levels (considering both voltage swing and current) it is a big deal. Heat sink the outputs, of course. To do that with small speakers you’ll need a darlington output and a 1.5A CCS. It works too, and is nearly foolproof. TIP35 and a BD139 driver will work fine - would probably be my first choice.
There are also plenty of other circuits out there that use op amp input stages, that are variations of more traditional circuits. More complicated, though. And the jury is out whether the op amp improves anything - in terms of sound quality at least.
“Improvements” can be made by using an output stage with voltage gain. Many of those can get horribly unstable and are tricky to work with. The easiest one to work with is the topology used in early QSC amps, but even they are subject to issues and are not totally foolproof. They turn the power supply inside out to force the output followers to run as a common emitter stage. They require separate power supplies for each channel and boot strapping the op amp supplies. If you don’t know what that is, study before proceeding. If interested it is WORTH the time/energy to study. To improve it has a headphone amplifier, remove the PNP output and replace it with a 100-300 mA CCS and run it single ended class A. Night and day. This even removes crossover artifacts in the op amp itself - and at headphone output levels (considering both voltage swing and current) it is a big deal. Heat sink the outputs, of course. To do that with small speakers you’ll need a darlington output and a 1.5A CCS. It works too, and is nearly foolproof. TIP35 and a BD139 driver will work fine - would probably be my first choice.
There are also plenty of other circuits out there that use op amp input stages, that are variations of more traditional circuits. More complicated, though. And the jury is out whether the op amp improves anything - in terms of sound quality at least.
IIRC OPA2604 had overheating related issues when used at the max ±24V and high output current and pulled from production. OPA445 works well up to ±45V, but not sure about stability.
The circuit is good as a learning exercise, but it is limited by
the op-amp's voltage rating, slew rate, frequency compensation, and linearity. To do better, one can't use an op-amp. This is where discrete transistors shine.
Ed
Here is an example.
OPA552 takes +/-30 Volt. It has acceptable performance. It comes in 8 pin DIP package.
Here I simulate with +/-27V which is about what you get from a 2x20VAC transformer.
Result is max 33 Watt output with rather low distortion.
The bandwidth is high.
OPA552 takes +/-30 Volt. It has acceptable performance. It comes in 8 pin DIP package.
Here I simulate with +/-27V which is about what you get from a 2x20VAC transformer.
Result is max 33 Watt output with rather low distortion.
The bandwidth is high.
Attachments
There are several topologies that allow an opamp frontend to avoid high voltages or high slew rate of the rest of the amp, for instance using current drive, so most of those limits can be eliminated. As for linearity that's simply not the case, opamps with <0.0001% THD are readily available (semiconductor shortages aside!). The more complex frequency compensation of these topologies may well be limiting, that I do agree with, and failure modes can be problematic. Mind you if the opamp's in a socket that's handy for repair!
Mark - I was referring to open-loop linearity. I have never seen an IC op-amp designed for good open-loop linearity. This is an area in which discrete transistor designs can do better.
Ed
Ed
yes it was exactly this circuit only the transistor has changed to higher wattage type and the bias readjusted to suit
Yes, TI/BB discontinued opa2604 due to thermal related issues when used close to max voltage, I simply can not recall the specifics. Use a couple of opa604 instead.
If you can use bipolar input, OPA551/OPA552 are cheaper and work well up to ±30V.
Last edited:
I use an older version of MultiSim.
It is a very expensive program.
But I downloaded it with torrent, a pirate copy.
"Multisim - NI Circuit Design Suite Power Pro 12.0 - Cracked"
It works well.