I have tried but been unable to find an elegant solution on how to use fully differential amplifiers to drive feedbacked output common collectors and other output transistor combinations.
In case anyone can correct the mistakes I may have made or suggest new ideas, anyone is welcome.
I have some more schematics, some of which I can publish when I update the file.
However, what I try to achieve is to make people think of whether fully differential amplifiers can be elegantly be used to drive power stages.
See attached. Whoever wants to skip the introduction can directly go to point 2. Some Schematics.
I am sick and tired of these for now as I have tried to find an elegant solution for a week. On the top of everything, I cannot publish the doc file because the forum does not allow big files. I had to delete the pictures and upload them separately. The names of the figures, however, stay in the file. Just the figures are outside.
In case anyone can correct the mistakes I may have made or suggest new ideas, anyone is welcome.
I have some more schematics, some of which I can publish when I update the file.
However, what I try to achieve is to make people think of whether fully differential amplifiers can be elegantly be used to drive power stages.
See attached. Whoever wants to skip the introduction can directly go to point 2. Some Schematics.
I am sick and tired of these for now as I have tried to find an elegant solution for a week. On the top of everything, I cannot publish the doc file because the forum does not allow big files. I had to delete the pictures and upload them separately. The names of the figures, however, stay in the file. Just the figures are outside.
Why you doesn’t try bridged connection of the two independent output stages?
Yes. This is OK too. However, the speaker is not referenced to ground. But the differential approach makes this OK with the standard advantages such as low noise, etcetera.
The idea to use two speakers is not a joke, although very strange and not practical. The idea with one speaker and four Darlingtons is not very practical either but is OK logically. With one speaker, the differential amplifier is really not needed except for some advantages outlined in the TI document of which I have posted a link. Thus, with one speaker, a better approach is to use only one simple amplifier. I tried to justify the use of a differential amplifier and drive two speakers to increase the power. The problem with two speakers is they are not the same because of manufacturer's tolerances, neither are the parts before them. Thus, when two speakers are used, there would be a difference between them. However, these differences are added outside of the electronics in the human head and will be negligible and unnoticeable. Even in case some human or animal can notice them, they ( mainly phase shifts ) will look like a 3D sound positioning.
Even though the differential approach is for noise reduction, I would prefer to ground the speaker or speakers as this would decrease the noise even more, I think.
Again, I have clearly stated these are just strange schematics and are not practical and should not be used. Neither should be the one speaker with four Darlingtons ( or even single transistors ).
I hesitated whether to publish them or not and decided to wait for the main point to be published first and then to publish the unimportant schematics after, in a separate chapter. The important point is to find an ELEGANT solution on how to use the differential amplifier to drive power transistors and, as I have stated, I have not been able to do so.
Another point was to raise awareness to make other people, such as yourself, for which I do thank you, think on this problem too.
Can you post the link to the thread. I may have a look in a while when I can but I think other people can read the said post too when they have a link.
Thanks.
F5X -- the EUVL Approach
Had to use google to find: F5X + diyaudio.com
The Forum's search engine could not find f5x, nor F5x, nor F5X, nor f5X,
even when telling it that EUVL started it.
Had to use google to find: F5X + diyaudio.com
The Forum's search engine could not find f5x, nor F5x, nor F5X, nor f5X,
even when telling it that EUVL started it.
No, the main problem not to drive OPS.
Main problem is a thermal runaway of the OPS devices since load halved.
Heavy load to OPS devices makes them really slower and causes beta drop, which in case adds troubles.
The best way is to use complementary MOSFETs in a pair of cascoded OPSes.
You can try OPS from famous AMB Beta24 project:
https://www.amb.org/audio/beta24/
In your case we can remove all marked blue, add four current sources, say 10-20 mA, marked green and use FDA to drive it from points marked red.
You can use straightforward “f5x site:diyaudio.com”
Last edited:
About two main types of fully differential amplifier I think there is:
1.) with 2 negative inputs, feedback is connected to input
2.) with 2 positive inputs, feedback is not connected to input but is connected somewhere else.
There is existing amplifiers done in both ways.
TI article talking about type 1.)
1.) with 2 negative inputs, feedback is connected to input
2.) with 2 positive inputs, feedback is not connected to input but is connected somewhere else.
There is existing amplifiers done in both ways.
TI article talking about type 1.)
Again, as previously mentioned, this is not a solution.
The question is not to build an amplifier but to use already built fully differential amplifier to drive the transistors, in other words, just a fully differential amplifier and two common collectors.
The schematics provided in chapter two are just ideas and are not for practical use.
Yes, there are some schematics which are beta dependent and there are some which are not. There are some where the beta dependency is in the low power level too.
Yes, quiescent current should be avoided.
The common emitter and common collector example with a center current is to show the logical inversion of Uo+. The disadvantages of this combination, with or without beta dependency, has been clearly stated.
As I have clearly stated, the schematics with outputs which are buffered by their own transistors are not the subject of this conversation and should be avoided whether with bipolar or any other transistors. They are just for a show and are not to be used in any shape and form, the ones suggested by you too.
I do not know how better to explain this as I have been trying and people do not seem to understand. I will try again :
I am interested in a possibility to drive two common emitters with offsets which offsets prevent deep closure of the transistors at Uo = 0. I am interested in a LOGICAL solution and not so much in physical calculations which are not part of the design but are just to tune a logical design.
I will try to express this again, even simpler :
There is one fully differential amplifier and two common collectors transistors, one NPN and one PNP, with their emitters connected. Design a LOGICAL circuit with as fewer parts as possible to drive these transistors without a jump. Do NOT provide solutions which can be done with normal amplifiers and not with fully differential ones as you have suggested. ONLY ONE FULLY DIFFERENTIAL AMPLIFIER ( which must be used in a way in which a normal amplifier cannot ) AND TWO COMMON COLLECTOR TRANSISTORS and AS FEWER PARTS AS POSSIBLE.
Whether bipolars are used or MOSFET's or JFET's or tubes is irrelevant and does not change the logical design. The design must be logical and then everyone can use whatever they want. Of course, other than bipolar transistors are noisy and have big parasitic capacitors while tubes are slow and need to be heated but these are physical characteristics and not logical.
I will try to provide you with an example of the difference between logical and physical : a transistor is nothing but a device which takes current or voltage and provides a possibility to conduct a bigger current or voltage. This is logic. Whether beta is temperature dependent or not and whether local feedback is used for a center current stabilisation is not important. This is a physical description.
The idea I have tried to employ is to use an independent control of the bases of the two feedbacked common collectors. One of them would get Ui + Uoff ( with any gain, gain is not important : not logic ), the other one : Ui - Uoff. The signal is the same, the offsets are different ( opposite polarities ). When Uo nears zero, the transistors will see 2Uoff between their bases. They can be made not to be fully closed. As the voltage changes and is different than 0, eventually, one of the transistors will close and the other one will be open. However, because audio signals are slow, as Uo nears 0, the closed transistor will be able to open ( or not be so much closed ) and the open transistor will remain open ( or will become not so much closed ) because of the offset difference of 2Uoff which came over the two base emitter junctions as Uo neared 0.
And this is a LOGICAL objective. The PHYSICAL is not important.
I do not know how to explain this. Thus, disregard those schematics which are not part of this design and I should not have published at all but have done this just for entertainment. As I have clearly stated, what you discuss can be done with a normal amplifier and a differential one is not necessary. The differential amplifier, in this case, is just for lower noise and not for LOGICAL design.
In case you think the fully differential amplifier cannot be used FOR A LOGICAL DESIGN to drive the proposed two common collectors, just say so and do not chew the circuit with two buffered and opposite outputs, which, again, can be LOGICALLY done with a simple amplifier and there is no need for a differential one. The idea to use a fully differential amplifier is because there are two outputs connected to the differential " wheel " of the input. The problem is, because of the figuratively presented " wheel " these perform subtraction ( differentiation ) of the inputs and provide signals with opposite polarities while the same polarities of the signal are necessary. I tried to use the fully differential amplifier as a NON DIFFERENTIAL amplifier where the differential amplifier is used to provide offsets and the non differential ( internal error ) amplifier is used to provide the signal with the disadvantages I have outlined. Thus, I am interested in a possibility to use the availability of two outputs and do NOT care at all whether the amplifier is differential or integral or not an amplifier at all.
I think I have explained what LOGIC means in an electronics design. In case anyone has any idea for a LOGICAL design which can utilise the INPUTS and OUTPUTS of something which is called a fully differential amplifier and which I do not care at all of, please, inform. Otherwise, I am not interested in designs in which the fully differential amplifier is used as a PHYSICAL device and the LOGICAL equivalent of which is a simple amplifier.
And here is another LOGICAL difference between the fully differential amplifier and the simple one : the simple one has only one output while the fully differential one has TWO. Can we use this difference to make a LOGICAL design which takes this into account?
I hope you now understand what I look for.
And this is an excellent answer. I would like to provide your answer as an example to other people who talk of other things which are not subject to the topic. I look for a logical design solution. Obviously, your answer is connected with this as new type amplifiers may have the capabilities I look for. I continue to be interested in logical solutions with type 1, but I am also interested in type2. Thank you very much.
I have only found three fully differential amplifiers, all made by TI, which can work with +-15V supply. The 30 and 31 ones have a slew rate of 51 V / us, The others, 650 V / us. All of these are type 1.
I now learn from you there is a type 2 and I am extremely interested. Can you, please, post a link of to Type 2 overviews or datasheets?
Hi Steve, have you read this AES paper? AES Journal Forum >> Fully Balanced Bridge Amplifier
"Fully Balanced Bridge Amplifier" by Takahashi, and Tanaka published in JAES Volume 32 Issue 6 pp. 415-421; June 1984.
Circuit:
It uses a floating power supply E3 for the power stage and a non-floating split supply (E1/E2) for the driver and input stages. The VAS is a Class-AB current drive stage. Ingenious but way ahead of it's time IMHO. I haven't made this topology but would like to.
Another approach might be the way the TDA2003 and TDA2005 bridge circuits were done with cross-coupled inverting inputs...
The TDA2003 internal circuit (attached) contains some interesting twists that DIY'ers might like to know about. Notice the input stage looks like single ended CFA. I made this bridge amp and sold about 20 as a kit to students back in the 90's. I used the 2nd non-inverting input as a differential input for ground loop noise reduction (but as shown it can't handle large common mode offsets).
Cheers
Attachments
Thank you for this excellent answer. This is the best answer so far as far as I can see.
I would look into the possibility.
I wanted to avoid complications because I may be able to make own differential or anti differential amplifier which defeats the purpose. I wanted a more elegant solution which, so far, has not been found.
Also, I realise the dual amplifier example is for a better understanding. There are other solutions which do not include differential and bridge amplifiers with two amplifiers to drive the power transistors, some of which have been discussed and are not a subject of this thread.
Thank you, again.
I would look into the possibility.
I wanted to avoid complications because I may be able to make own differential or anti differential amplifier which defeats the purpose. I wanted a more elegant solution which, so far, has not been found.
Also, I realise the dual amplifier example is for a better understanding. There are other solutions which do not include differential and bridge amplifiers with two amplifiers to drive the power transistors, some of which have been discussed and are not a subject of this thread.
Thank you, again.
Last edited:
OK. I have just spent a few seconds with the transistor essential of what is called a bridge amplifier and I was unable to find any solution which had not been found before here. I am sure the other people would love this but I do not. I cannot see anything I can do with the so called bridge amplifier transistors which I cannot do with just a simple single amplifier and two or three transistors or a transistor and two Darlingtons.
I am unable to dedicate more effort unless I am sure I can get somewhere, sorry, I have spent enough.
However, in case anyone can find any ELEGANT logical solution with the bridge amplifier or without, please, by all means, do publish this as an answer. Any other solutions which I do not like but most people do ( or don't ) are very welcome too. This thread is for all not only for whoever started the thread. I, personally, am not interested. There were a few people who showed interest, however. Most of them do.
In case anyone can do something which I like, I would be happy. In case not, I would remain unhappy which is not a big deal.
I am unable to dedicate more effort unless I am sure I can get somewhere, sorry, I have spent enough.
However, in case anyone can find any ELEGANT logical solution with the bridge amplifier or without, please, by all means, do publish this as an answer. Any other solutions which I do not like but most people do ( or don't ) are very welcome too. This thread is for all not only for whoever started the thread. I, personally, am not interested. There were a few people who showed interest, however. Most of them do.
In case anyone can do something which I like, I would be happy. In case not, I would remain unhappy which is not a big deal.
Yes. I can’t understand why you want to scratch behind your left ear with your right leg.
Difference amp are easily to understand as two inverting amps having one reference.
Why you want to drive one output stage from amp demanding two phase reversed feedbacks?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.