logo elektroda
logo elektroda
X
logo elektroda
Advanced Search
logo elektroda

Power amplifier on NE5532 - ty.ytka

tytka 17055 79
ADVERTISEMENT
Treść została przetłumaczona polish » english Zobacz oryginalną wersję tematu
Reply Cool? Ranking DIY | New topic
Notify about new articles
  • #31 20653115
    austriackimalarz
    Level 26  
    marweg1967 wrote:
    One channel for about 4 k PLN? Well, Lord, it's almost high-end. Something just in time for another block to the resurrected Unitra.
    By the way, I wonder how it would look during blind listening compared to the work of the author of the topic.

    Unfortunately, the right OP-AMPs are expensive :c
    Without offending other designs of the mentioned author, it seems to me that such a device would work quite well. Think about it - is it better to use single power transistors with medium parameters, but with adequate power, or to use a bucket of small transistors in parallel? Is it better to put one large 10000uF capacitor somewhere on the edge of the board, or instead have a thousand 10uF ceramic SMDs, one per op-amp?
    Is it better to run currents on narrow paths with high inductance, or use copper fields with 20 PCBs with power elements as paths?
    The author's design is stupid, incredibly impractical when it comes to the approach to building an amplifier. This is the extreme of all the principles of building this type of system. But I want one too. I'm at the design stage :) Assuming +/-18V and 120W into 8 ohms :D
  • ADVERTISEMENT
  • #32 20653457
    nyquist
    Level 26  
    austriackimalarz wrote:
    (...) incredibly impractical when it comes to the approach to building an amplifier (...)

    That's right.
    austriackimalarz wrote:
    The author's design is stupid (...)

    But I personally don't agree with that. The unconventional solution used in the project is actually impractical and unprofitable, which makes it rather a curiosity, but by no means stupid.
  • #33 20653467
    austriackimalarz
    Level 26  
    nyquist wrote:
    but by no means is it stupid

    It's stupid. But it works. So it's not stupid.
    I ordered 2,000 SOP8 amplifiers yesterday, and 4,000 2.2uF capacitors yesterday. You won't guess why :)
  • #34 20653475
    nyquist
    Level 26  
    austriackimalarz wrote:
    It's stupid. But it works. So it's not stupid.

    Exactly ;)
    Sorry for the little OT, but it reminded me of...
    " When everyone knows something is impossible, someone who doesn't know it comes along and he does it - Albert Einstein
  • #35 20653536
    yogi009
    Level 43  
    austriackimalarz wrote:

    I ordered 2,000 SOP8 amplifiers yesterday, and 4,000 2.2uF capacitors yesterday. You won't guess why :)


    You'll be making a radio station for a Russian alligator :-)
  • #36 20653537
    austriackimalarz
    Level 26  
    yogi009 wrote:
    You'll be making a radio station for a Russian alligator

    Exactly :) I will connect the amplifiers in series to withstand 400V power supply :P

    EDIT:
    Hello hello, stop for a moment. Not in SOP8, just plain SOT-23. I'm going to connect 500 of them in parallel, and then the whole thing into a stereo bridge connection :)
  • #37 20653543
    yogi009
    Level 43  
    This and the tuning capacitor are made of 300 small air capacitors connected in series :-) About 5 meters of the axis will come out :-)
  • #38 20653549
    austriackimalarz
    Level 26  
    yogi009 wrote:
    This and the tuning capacitor make 300 small air capacitors connected in series About 5 meters of axis will come out

    ABOUT! And I'll pull the whole thing into a plastic hose and fill it with oil. Then I will be able to roll up such a hose-capacitor, I will immediately solve the choke issue :)
  • ADVERTISEMENT
  • #39 20653558
    yogi009
    Level 43  
    austriackimalarz wrote:

    ABOUT! And I'll pull the whole thing into a plastic hose and fill it with oil. Then I will be able to roll up such a hose-capacitor, I will immediately solve the choke issue :)


    I had a different plan. 5 meters of capacitor just enters the alligator and does not interfere with the propellers :-)
  • #40 20653573
    Urgon
    Level 38  
    AVE...

    Better to take a few hundred 1N4007 diodes and use them as varicaps. acc. notes capacitance 5-~30pF per diode at voltages from 100mV to 100V. It is enough to connect them in parallel in a large enough number. Make a ladder out of it, then fold it into a double helix - it'll be electronic DNA...
  • #41 20653575
    yogi009
    Level 43  
    Urgon wrote:
    Make a ladder out of it, then fold it into a double helix - it'll be electronic DNA...


    Ba, but how to carve 21 amino acids out of it?
  • #42 20653704
    dipol
    Level 34  
    Similar wcz modules / cubes are in radio and television transmitters - commercial systems. I think our Program 1 / Warsaw I-Solec Kujawski has such lower power modules that after summing up - the combiner - give power to the antenna / en many / hundred kilowatts. Ease of servicing, automatic shuts down the damaged one turns on the reserve one, etc.
    73!
  • ADVERTISEMENT
  • #43 20653708
    Urgon
    Level 38  
    AVE...

    Standard construction method for high power transmitters. The only "problem" is that you have to assemble the whole thing in such a way that all the power amplifiers are connected "in phase" - the transmission lines, both those going to the inputs of the amplifiers and the output ones, must have the same length. It also works the other way around - in radio astronomy, many smaller antennas are combined into one phase-synchronized assembly - the apparent diameter of such an assembly is much larger than the diameter of a single antenna...
  • #44 20653757
    dipol
    Level 34  
    Urgon - as you write - m/other power adders are used for this, I have a simple solution in my radio/shortwave transceiver.
  • #45 20654167
    marweg1967
    Level 14  
    Urgon wrote:
    You can use the SOIC version and do it on a double sided board. All circuits on one side, other components in the SMD variant on the other. Then, grease the amplifiers with thermal paste and screw the entire plate to the heatsink ...

    tytka wrote:
    If a commercial version were to be created, such a solution should be seriously considered.


    All in all, I'm thinking about such a solution (on NE5532, not OPA), although mainly due to the lower cost of ICs, and that it would also solve the cooling problem, so much the better.

    Added after 12 [minutes]:

    austriackimalarz wrote:
    Without offending other designs of the mentioned author, it seems to me that such a device would work quite well. Think about it - is it better to use single power transistors with medium parameters, but with adequate power, or to use a bucket of small transistors in parallel?
    ...
    The author's design is stupid, incredibly impractical when it comes to the approach to building an amplifier. This is the extreme of all the principles of building this type of system.


    Well, in higher power amplifiers, output transistors were quite commonly connected in parallel, both BJTs and MOSFETs.

    I wouldn't call the project stupid by any means. Unconventional - yes. Is it incredibly impractical? On the one hand, the construction itself can, even if so many legs are connected in parallel, on the other - there is no need to regulate the quiescent current, for example :)
  • #46 20654363
    austriackimalarz
    Level 26  
    marweg1967 wrote:
    on the other - there is no need to regulate the quiescent current, for example

    And this is a huge disadvantage of this solution. It is true that in a small scale, as in the case of a titty, the sum of the quiescent currents of the amplifiers will not be large, although larger than in the typical AB class;
    NE5532P has Icc with parameters No load, Vo=0, Vcc=+-15V around 8mA (max 16mA). They are connected 64 in parallel, so the total quiescent current of the final stage is as much as half an amp, and a maximum of 1A in the worst case scenario.
    Meanwhile, in my case, one op-amp with the same parameters has catalog currents of 1.4-2.5mA, and since there will be 2,000 of them, we get a beautiful 2.8-5A!!!
  • #47 20654385
    nyquist
    Level 26  
    austriackimalarz wrote:
    Yesterday I ordered 2,000 amplifiers in SOP8 (...)

    Just out of curiosity, also 5532 or something else?
  • #48 20654388
    austriackimalarz
    Level 26  
    nyquist wrote:
    Just out of curiosity, also 5532 or something else?

    austriackimalarz wrote:
    Hello hello, stop for a moment. Not in SOP8, just plain SOT-23. I'm going to connect 500 of them in parallel, and then the whole thing into a stereo bridge connection

    We settled on the TL071HIDBVR, which is the well-known TL071 in the SOT-23 housing. On the TI website, 2000 units cost about 100 euros, also bearable.
  • #49 20654398
    Urgon
    Level 38  
    AVE...

    It's only 150W, don't whine. The stereo amplifier on the GU81M takes more for the glow current alone ...

    By the way, generally parallel connection of op-amps is practiced to reduce self-noise, not for greater current efficiency. From the latter we have transistors...
  • #50 20654409
    austriackimalarz
    Level 26  
    Urgon wrote:
    By the way, generally parallel connection of op-amps is practiced to reduce self-noise,

    Meanwhile, in post 12:
    austriackimalarz wrote:
    Generally, connecting several amplifiers in parallel eliminates noise, but ... without exaggeration ;P

    Urgon wrote:
    It's only 150W, don't whine. The stereo amplifier on the GU81M takes more for the glow current alone ...

    150W is enough for one lamp to glow :)
  • ADVERTISEMENT
  • #51 20654463
    Urgon
    Level 38  
    AVE...

    If you connect N amplifiers in parallel, the noise power of a single amplifier is divided by N, and the equivalent input noise voltage is divided by √N.
    Example: by connecting two TL071s in parallel you will reduce the noise floor from 15nV/√Hz to 10.6nV/√Hz, so by connecting 2000 devices you will gain 0.335nV/√Hz.
    For the audio band, this will give a noise voltage of 47.4nVp-p, bearing in mind that a single TL072 is 2.12µVp-p...
  • #52 20654573
    marweg1967
    Level 14  
    austriackimalarz wrote:
    NE5532P has Icc with parameters No load, Vo=0, Vcc=+-15V around 8mA (max 16mA). They are connected 64 in parallel, so the total quiescent current of the final stage is as much as half an amp, and a maximum of 1A in the worst case scenario.
    Meanwhile, in my case, one op-amp with the same parameters has catalog currents of 1.4-2.5mA, and since there will be 2,000 of them, we get a beautiful 2.8-5A!!!

    I didn't say anything about the size of this current, only that it doesn't need to be regulated.
    Are you 5A you say? Well, while I described the whole project as unconventional, I would similarly define the number of cubes in your version of this project as unconventional. That is, such unconventionality squared ;)

    Added after 8 [minutes]:

    austriackimalarz wrote:
    We settled on TL071HIDBVR, which is the well-known TL071


    Why exactly TL071? If not NE, then why not TL074? It would be 4 times cheaper, at least according to Mouser's price list.
  • #53 20654593
    austriackimalarz
    Level 26  
    marweg1967 wrote:
    Why exactly TL071? If not NE, then why not TL074? It would be 4 times cheaper, at least according to Mouser's shader.

    You took up the topic of profitability in the topic of building power amplifiers on operational amplifiers.
    I don't think you thought your question through :)
  • #54 20654601
    marweg1967
    Level 14  
    austriackimalarz wrote:
    marweg1967 wrote:
    Why exactly TL071? If not NE, then why not TL074? It would be 4 times cheaper, at least according to Mouser's shader.

    You took up the topic of profitability in the topic of building power amplifiers on operational amplifiers.
    I don't think you thought your question through :)

    I thought: since you have already spared the money for the TL071, then by switching to the TL074 you would improve the noise parameters twice for free :D
  • #55 20654604
    austriackimalarz
    Level 26  
    Possible.
    But I don't want to redesign the board :P
  • #56 20654612
    marweg1967
    Level 14  
    austriackimalarz wrote:
    Possible.
    But I don't want to redesign the board :P

    And who hasn't thought this through? ;)
    I keep my fingers crossed for the success of your crazy... uh, unconventional project :D

    But I will tell you that if I did not consider TL at all, using the TL074 now seems to me an interesting option. Because of course I also want to commit this project, as I mentioned before. But I would rather close myself in dozens of integrated circuits, and electronics in a not very tight housing without transformer oil :D

    BTW: interesting nickname ;)
  • #57 20656289
    terminus
    Level 19  
    #20640831
    You should not use such a large capacitance (2200 µF) at the output of linear stabilizers, it greatly deteriorates their parameters.
    In this case (LM338) 1 µF is enough on the output.
  • #58 20657278
    yogi009
    Level 43  
    austriackimalarz wrote:
    It ended on TL071HIDBVR


    I do not know how with such an application, but listening tests showed a significant qualitative advantage (in terms of sound) of the NE5532 (TI) or even the LM833 over the TL07x series. A separate topic is the quality of the loudspeaker drivers on which we are to make the comparison. It's always the bottleneck in the puzzle that counts.

    Added after 1 [minutes]:

    Urgon wrote:
    paralleling op-amps is practiced to reduce self-noise, not to increase current efficiency


    And that's the essence of these types of projects.

    Added after 9 [minutes]:

    marweg1967 wrote:
    using the TL074 now seems to me an interesting option


    I wouldn't get into quad dice for several reasons. Cooling, crosstalk, the ability to replace the circuit with another (here double opamps give the widest possibilities). From the point of view of the lowest noise, a single circuit would seem optimal, but its application is slightly less convenient, and the surface of the PCB increases. So it seems that compromise dual op-amps are commonly chosen for a reason.

    I also think that you can compare the parameters of such a system with something dedicated and more powerful. In the scale of headphone amplifiers, the LME series from TI is used (e.g. LME49860 + LME49600 buffer - or maybe it is enough to multiply such a buffer?). The lack of capacitors separating the successive stages is tempting ... I am curious about the differences between one and the other approach.
  • #59 20658126
    katakrowa
    Level 23  
    To make it modern, it begs to be built on op amps in class D ;-)
    I wonder when these will be widely available and at prices comparable to classic ones.
  • #60 20658137
    austriackimalarz
    Level 26  
    katakrowa wrote:
    To make it modern, it begs to be built on op amps in class D
    I wonder when these will be widely available and at prices comparable to classic ones.

    Weak.
    I would suggest converting some polyphase converter from the GPU power supply to the amplifier :)
Reply Cool? Ranking DIY | New topic
Notify about new articles
Report a violation of the law

Topic summary

The discussion revolves around the construction of a power amplifier using the NE5532 operational amplifier, achieving a power output of 2×14 W at 8 ohms. The project, inspired by Douglas Self's article in Elektor magazine, is described as an experimental endeavor rather than a commercial product. Participants share insights on the design, including the use of a uPC1237 for protection and LM338 for voltage regulation. Various contributors express interest in replicating the project, discussing the challenges of sourcing components, particularly the NE5532, and the implications of using clones versus originals. The conversation also touches on measurement techniques, noise reduction through parallel connections of op-amps, and the practicality of such designs in audio applications. Some participants suggest alternative components and configurations, while others highlight the potential for high power outputs with larger setups.
Summary generated by the language model.
ADVERTISEMENT