I was asking myself same question and this article gave me an answer. Must read!
http://tangentsoft.net/audio/opamp-bias.html
http://tangentsoft.net/audio/opamp-bias.html
kanwar,
You can force any op-amp to flow current in one side of the output stage therefore that part is always class-a, by simply connecting a resistor or constant-current device from the output to the opposite supply rail. It seems to be more common to make the upper transistor the class-a one but I think really it depends on the PSRR for that rail and the intrinsic linearity, i.e. a lot of op-amps use quasi output.
You can force any op-amp to flow current in one side of the output stage therefore that part is always class-a, by simply connecting a resistor or constant-current device from the output to the opposite supply rail. It seems to be more common to make the upper transistor the class-a one but I think really it depends on the PSRR for that rail and the intrinsic linearity, i.e. a lot of op-amps use quasi output.
Hi
Thanks very much for the help.
Ritchie you are simply great man. yes you are right in case of some quasi-opamps.
I want to convert OPA2134 best professional high audio opamp from texas instrument.but i dont know that iis it comp. or quasi.
Now i will do the job on it and give the results soon.
here is the link to opamp http://focus.ti.com/docs/prod/folders/print/opa2134.html
Regards,
kanwar
Hi shusha,shusha said:
Thanks very much for the help.
richie00boy said:
Ritchie you are simply great man. yes you are right in case of some quasi-opamps.
I want to convert OPA2134 best professional high audio opamp from texas instrument.but i dont know that iis it comp. or quasi.
Now i will do the job on it and give the results soon.
here is the link to opamp http://focus.ti.com/docs/prod/folders/print/opa2134.html
Regards,
kanwar
I've been forcing opamps in class A since the late 80s.
For opamp preamp's, opamp driven Mosfet class A amplifiers, currently in rebuilt partly class A 4-channel car power amplifiers with OPA627 and sanken1216/2922, and in the Stokes SDS class A headphone amplifier with RFP12N20/RFP12P10 Mosfet outputs.
With many different opamps from 5532/5534, OP27, LT1028, now OPA627 and OPA637.
Also the 134, and 2134.
Difference was staggering on every attempt.
Not surprising as mosfets only sound any good driven in class A(imo), think of the number of fets in an opamp.
Think of glueing a heatsink on top of the opamp, if you are planning to stretch it.
For opamp preamp's, opamp driven Mosfet class A amplifiers, currently in rebuilt partly class A 4-channel car power amplifiers with OPA627 and sanken1216/2922, and in the Stokes SDS class A headphone amplifier with RFP12N20/RFP12P10 Mosfet outputs.
With many different opamps from 5532/5534, OP27, LT1028, now OPA627 and OPA637.
Also the 134, and 2134.
Difference was staggering on every attempt.
Not surprising as mosfets only sound any good driven in class A(imo), think of the number of fets in an opamp.
Think of glueing a heatsink on top of the opamp, if you are planning to stretch it.
kanwar,
I use the OPA134 and OPA2134 myself as it's a really nice chip. I believe it has a quasi-complementary output as the max swing available to each rail is not symmetrical (see output voltage swing vs output current graph). Because it can swing closer to the lower rail I believe that one is the quasi side. No PSRR data is given for the individual rails so I have to base my choice on the output stage symmetry alone. Therefore I hang a current source made from 2N3819 JFET and resistor off +ve rail and down to the output as I believe the cfp/quasi side works better in class-a and also there is more swing available.
I hope this brings you success and nice sound
I use the OPA134 and OPA2134 myself as it's a really nice chip. I believe it has a quasi-complementary output as the max swing available to each rail is not symmetrical (see output voltage swing vs output current graph). Because it can swing closer to the lower rail I believe that one is the quasi side. No PSRR data is given for the individual rails so I have to base my choice on the output stage symmetry alone. Therefore I hang a current source made from 2N3819 JFET and resistor off +ve rail and down to the output as I believe the cfp/quasi side works better in class-a and also there is more swing available.
I hope this brings you success and nice sound
Extrapolating from experience with discrete amps, I think the asymetic swing is more likely to from use of a single differential stage and non-symetric intermediate stage. I.e., non-"mirror image" topology. I don't think quasi-comp output stage contribute to this.
My thoughts are definately non-expert and extrapolations at that so give them no more than their due. Others with more specific knowledge may care to correct this.
My thoughts are definately non-expert and extrapolations at that so give them no more than their due. Others with more specific knowledge may care to correct this.
I too have done much testing and application of Class A output stage forcing of chips particularly the FET input chips LF357,( decompensated), TL081, and OPA627 - while viewing distortion residual waveform.
The results varied widely with chips from different mfgrs giving different results. So there is no rule of thumb each type and make will have a different optimum and benefit. You need a quality distortion analyser to do anything definitive.
The results varied widely with chips from different mfgrs giving different results. So there is no rule of thumb each type and make will have a different optimum and benefit. You need a quality distortion analyser to do anything definitive.
lumanauw
You cannot easily determine theoutput operating state of many opamps. You can insert resistors in the supply leads and raise the signal into a load to determine a point of transition for the output stage but it's very rough.
regarding power dissipation it's the supply voltage x the standing current consumption. The data sheet will tell you the case dissipation and derate for temp rise. Remember your constant current only forces half output stage dissipation. So determine your max demanded current to the load then add a couple of mA and set the current source to this. Then to finesse you need a THD analyser.
Hope that helps.
You cannot easily determine theoutput operating state of many opamps. You can insert resistors in the supply leads and raise the signal into a load to determine a point of transition for the output stage but it's very rough.
regarding power dissipation it's the supply voltage x the standing current consumption. The data sheet will tell you the case dissipation and derate for temp rise. Remember your constant current only forces half output stage dissipation. So determine your max demanded current to the load then add a couple of mA and set the current source to this. Then to finesse you need a THD analyser.
Hope that helps.
I seem to recall from an appnote somewhere that using a resistor btween output anv -V can afefct the input impedance of the next stage as seen by the opamp you are lifting into class-a. This doersn't necessarily mean you have a problem, but you need to be aware of it in the context of the particular opamp you are using and what you are doing with it.
LM359 output stage is NPN darlington with NPN current sink, pure class A. The sink current can be programmed with an external resistor or current source. This is an early 80's "Norton" op amp like LM3900 but with video bandwidth, needs a discrete first stage to be useful with high impedance sources. Max voltage rail-to-rail is 22V. It's still available from National Semi.
NE5534 output stage is totem pole, Q13 is upper NPN, Q17 is lower NPN, with diode D5 in series with lower NPN collector, driven by NPN phase splitter Q18, assuming I've read the "simplified schematic" correctly. A PNP emitter follower Q22 completes a feedback loop from Q17 collector to Q18 base. Q17 collector is connected to pin 5 which is normally used for compensation but can be used with a current source for a single-ended class A output that can sink at least 15mA and has no current limiting if output pin 6 is open. The schematic I refer to is available here .
For NE5534, a single op amp, in DIP package, maximum dissipation is about 500mW at 25C with thermal resistance 160C/W and 105C chip temperature. With 4mA typical supply current and +/- 15V rails, dissipation is 120mW. With no output load, up to 380mW can be taken by a class A current sink, or 25mA at 15V. 10mA sink at 15V causes 150mW dissipation, leaving 230mW or 15mA output current available. 5534 output pin can sink/source 50mA so its current limit will not activate.
Op amp output stages are class A up to a certain current depending on the design, if an op amp has supply current of 5mA, this does not mean that its output is class A up to 5mA. As amplifierguru says, if you really wanted to you could measure the class A-B transition current by applying a load to ground and measuring the current in a supply pin while applying an input, if the output is positive the negative supply pin current will decrease only a certain amount, this decrease should match the built-in class A bias current.
NE5534 output stage is totem pole, Q13 is upper NPN, Q17 is lower NPN, with diode D5 in series with lower NPN collector, driven by NPN phase splitter Q18, assuming I've read the "simplified schematic" correctly. A PNP emitter follower Q22 completes a feedback loop from Q17 collector to Q18 base. Q17 collector is connected to pin 5 which is normally used for compensation but can be used with a current source for a single-ended class A output that can sink at least 15mA and has no current limiting if output pin 6 is open. The schematic I refer to is available here .
For NE5534, a single op amp, in DIP package, maximum dissipation is about 500mW at 25C with thermal resistance 160C/W and 105C chip temperature. With 4mA typical supply current and +/- 15V rails, dissipation is 120mW. With no output load, up to 380mW can be taken by a class A current sink, or 25mA at 15V. 10mA sink at 15V causes 150mW dissipation, leaving 230mW or 15mA output current available. 5534 output pin can sink/source 50mA so its current limit will not activate.
Op amp output stages are class A up to a certain current depending on the design, if an op amp has supply current of 5mA, this does not mean that its output is class A up to 5mA. As amplifierguru says, if you really wanted to you could measure the class A-B transition current by applying a load to ground and measuring the current in a supply pin while applying an input, if the output is positive the negative supply pin current will decrease only a certain amount, this decrease should match the built-in class A bias current.
Hi buddy
Hi Ritchie,
I have applied the constant current source with a transistor 2N5551 at about 10mA, with +-18V supplies. The result was astonishing and the sonic detail was something to heard off, really this chip in class-A works extremely well.
I also think that this chip is quasi comp.
The max. output swing i get is 10Vrms with 1kiloOhm load, and that too at frequency at 30kHz.
There is also another Chip IN134, IN2134 from texas instruments which has internally laser trimmed resistors which result in differential gain of one. there is no need to put extra resistors which saves space and of course cost of resistor......HaHa.
Now the sound of our workhorse amp would be much much better!, ThanX alot.
This discussion is getting high slew rate.........
regards,
Kanwar
richie00boy said:kanwar,
I use the OPA134 and OPA2134 myself as it's a really nice chip. I believe it has a quasi-complementary output as the max swing available to each rail is not symmetrical (see output voltage swing vs output current graph). Because it can swing closer to the lower rail I believe that one is the quasi side. No PSRR data is given for the individual rails so I have to base my choice on the output stage symmetry alone. Therefore I hang a current source made from 2N3819 JFET and resistor off +ve rail and down to the output as I believe the cfp/quasi side works better in class-a and also there is more swing available.
I hope this brings you success and nice sound
Hi Ritchie,
I have applied the constant current source with a transistor 2N5551 at about 10mA, with +-18V supplies. The result was astonishing and the sonic detail was something to heard off, really this chip in class-A works extremely well.
I also think that this chip is quasi comp.
The max. output swing i get is 10Vrms with 1kiloOhm load, and that too at frequency at 30kHz.
There is also another Chip IN134, IN2134 from texas instruments which has internally laser trimmed resistors which result in differential gain of one. there is no need to put extra resistors which saves space and of course cost of resistor......HaHa.
Now the sound of our workhorse amp would be much much better!, ThanX alot.
This discussion is getting high slew rate.........
regards,
Kanwar
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.