Hello, I came across this schematic.
https://www.electroschematics.com/mosfet-audio-amplifier-irf9530-irf530/
To my understanding, the opamp is loaded with the 240ohm resistor, and the signal is taken from the current consumption of the opamp ? Is my understanding correct ? Is that really smart or really ugly ?
Thank you 🙂
https://www.electroschematics.com/mosfet-audio-amplifier-irf9530-irf530/
To my understanding, the opamp is loaded with the 240ohm resistor, and the signal is taken from the current consumption of the opamp ? Is my understanding correct ? Is that really smart or really ugly ?
Thank you 🙂
Attachments
Looks like the 240 ohm resistor is used to intentionally draw current from the opamp supply rails that will develop a voltage through R8 and R13 to drive the gates of the MOSFETs.
Yes, it's an old circuit idea that sort of works, but not very well.
Last edited:
Elektor, the hobby electronics publication that design was lifted from, presented a couple more, somewhat improved versions of that in the OP. It became a popular project, not least because it included a JFET differential input stage at no significant cost!
There's an informative description of the approach at Rod Eliott's ESP website: Audio Designs With Opamps - 3
There's an informative description of the approach at Rod Eliott's ESP website: Audio Designs With Opamps - 3
Last edited:
R6-R5 belong to a local negative feedback loop including the output Mosfets.
Point 6 is a perfect example of a low impedance inverting input.
If I remember correctly, the circuit was initially proposed by Siliconix.
Point 6 is a perfect example of a low impedance inverting input.
If I remember correctly, the circuit was initially proposed by Siliconix.
My SoundCraft mixer used this circuit as a headphone amplifier. IT SUCKED! I replaced it with Rod's headphone amp:
Headphone Amplifier
Headphone Amplifier
Yes I recall seeing it in the Siliconix MOSFET applications book going back to the 1980s. I turned it into a commercial subwoofer amp in the mid 1990s.
I tried it a couple times as a low power (whatever +/-15V rails can give you) amplifier.
Worked, soft of, always felt insecure about open loop gain and stability, because power transistors do have voltage gain but it´s not clear how much and besides that undocumented extra gain appears inside the NFB loop
Might have solved it but not worth for a seldom used power range and whose job could be done much better and in a more predictable way by chipamps so not worth pursuing that avenue.
Worked, soft of, always felt insecure about open loop gain and stability, because power transistors do have voltage gain but it´s not clear how much and besides that undocumented extra gain appears inside the NFB loop
Might have solved it but not worth for a seldom used power range and whose job could be done much better and in a more predictable way by chipamps so not worth pursuing that avenue.
It's a trick circuit with the downsides of a trick circuit - particularly, being very parts-specific. The basic idea is cascoding the output devices in the opamp. A part of low idle current draw but decent output current capability is required for best results.
The original simple application shown in Databooks used a couple plain complementary pair transistors , to get a few watts.
I used that to drive reverb tanks, much harder than what a puny Op Amp could.
Acceptable performance in that realm.
I could hit them with, say, 60mA peak from a couple TO92 transistors instead of typical 5mA from a generalpurpose Op Amp or 15/20mA from an NE553x one.
Then I tried using TO220 mid power Darlingtons, think TIP121/126 family, for small practice amps, it worked but again, a TDA20x0 gave better results and was simpler/cheaper to use, so why bother?
Doubling the bet by adding level shifters to stand higher voltage as shown above makes it as complex as a "real" discrete amplifier, so again, why bother?
I used that to drive reverb tanks, much harder than what a puny Op Amp could.
Acceptable performance in that realm.
I could hit them with, say, 60mA peak from a couple TO92 transistors instead of typical 5mA from a generalpurpose Op Amp or 15/20mA from an NE553x one.
Then I tried using TO220 mid power Darlingtons, think TIP121/126 family, for small practice amps, it worked but again, a TDA20x0 gave better results and was simpler/cheaper to use, so why bother?
Doubling the bet by adding level shifters to stand higher voltage as shown above makes it as complex as a "real" discrete amplifier, so again, why bother?
Properly implemented it works extremely well. At two former employers I helped design or designed very similar power amps with current mirrors in the supply rails to drive the output devices. (Same operating theory, better implementation) Millions of amplifiers based on that design were sold. I have a home theater system that uses a 75W class G variant of this design (that I did not design) that has operated flawlessly for the past 23 years.
There are pitfalls to this design however for the unwary.. 😛
If you took the time to debug and successfully implement it, more power to you 🙂
If you made money, even better 😀
If you made money, even better 😀
Unfortunately someone else made all the money.. I designed a number of amplifiers based on this topology, up to about 250W, but I was nothing more than a salary man.. 😀
These days I'd think for most of those applications a class D amplifier would fit the bill, be much simpler, more efficient and less costly.
These days I'd think for most of those applications a class D amplifier would fit the bill, be much simpler, more efficient and less costly.
What's interesting is the bias adjustment. It only feeds the top FET and relies on the op-amp to match that bias to the lower FET in order to maintain DC offset.
I think a better plan is to boot-strap the op-amp to extend it's range above +/-15VDC, but, there are problems and limitations.
As others have said, if you want a cheap little amp, best to buy a chip-amp where thermal and short protection comes ~free.
I think a better plan is to boot-strap the op-amp to extend it's range above +/-15VDC, but, there are problems and limitations.
As others have said, if you want a cheap little amp, best to buy a chip-amp where thermal and short protection comes ~free.
I guess the Alexander current feedback power amp uses the similar idea.
https://www.analog.com/media/en/tec...tes/58052492001115525484056221917334AN211.pdf
https://www.analog.com/media/en/tec...tes/58052492001115525484056221917334AN211.pdf
I should mention that many op-amp SPICE models, notably most 5534 models do not model the supply pins well and therefore do not simulate this circuit properly. Only single op-amps, not 5532 or other duals and quads can do this because they share supply pins.
-Yes, feedback and loading to the op-amp output is reflected in the supply pin currents.
-The amount of feedback to the op-amp output complicates stability because too little voltage swing at the op-amp output cancels the op-amp compensation.
-And the op-amp output A/AB mode impacts mode of the added stages, which means that different op-amps behave differently in such circuits.
-The supply pins carry other current beside the output current, which may be an issue, both as undesirable output and as a PSRR issue for the input section.
-Perhaps Alexander was first to think of this topology and perhaps not. Patents are given to the first person vain enough to think their idea is new and worth patenting and has the resources to file a patent. Usually the actual first user doesn't bother with a patent because they think it trivial and/or they do not have the time and money. I used the idea in the 70's and I know it didn't come from Alexander.
-Yes, feedback and loading to the op-amp output is reflected in the supply pin currents.
-The amount of feedback to the op-amp output complicates stability because too little voltage swing at the op-amp output cancels the op-amp compensation.
-And the op-amp output A/AB mode impacts mode of the added stages, which means that different op-amps behave differently in such circuits.
-The supply pins carry other current beside the output current, which may be an issue, both as undesirable output and as a PSRR issue for the input section.
-Perhaps Alexander was first to think of this topology and perhaps not. Patents are given to the first person vain enough to think their idea is new and worth patenting and has the resources to file a patent. Usually the actual first user doesn't bother with a patent because they think it trivial and/or they do not have the time and money. I used the idea in the 70's and I know it didn't come from Alexander.
This is the oldest instance of the circuit I've seen. I saw a much more refined version in Wireless World (UK) in the late 1980s - I believe this is where I saw current mirrors employed. In most designs there was a follower inserted below the mirror in the op-amp current leg to set the op-amp supply voltage to the desired value.
Devices with low quiescent current and high output current capability relatively speaking are desirable. We used LF351 and LF411 from Analog Devices and had some problems with second sources when the originals starting going EOL.
Simulating them in spice can be problematic if the op-amp's supply behaviors are not modeled accurately.