TDA2030A as Headphone Amp Gain problem

Discussion in 'The Projects Forum' started by oneoldude, Oct 15, 2013.

  1. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    I have a TDA2030A based amplifier that should make a decent headphone amp except that its gain, at about 32 dB, is way too high.

    It is easy to change the gain setting resistor or even put in a switch to change the gain from say 10 to 15 to 20 dB for proper Headphone use. But there is a problem.

    The TDA2030A data sheet says the chip needs to run at a gain in excess of 24 dB to maintain stability. That is too high for low impedance high sensitivity headphones. The volume control will never be in a linear tracking region.

    A resistor could be placed in series with the output. But that destroys damping factor and sound quality. That is not an acceptable solution.

    However, if you can get stable operation at lower gains not only can the amp be used for headphones but the necessary addition of feedback should improve distortion specs as well. It is a nice goal for a very available OPA.

    Meier Audio uses the TDA2030A in their Corda Brick. It has switchable gain of -1 to +14 dB. So how do they do that with the TDA2030A?

    Anybody got a working circuit that gets the TDA2030A down to a stable (switchable?) gain of 10-15-20 dB?

    Suggestions?

    Thanks
     
    Last edited: Oct 15, 2013
  2. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,232
    Can't you just put a pot at the input?
     
  3. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I see a contradiction in terms: you are starting out with a medium junky amplifier (2030) that was designed for cheap, low performance home stereos and then claiming you will be able to hear the "degradation" in damping factor from a series resistor.... if you can, you have superb ears and some very high end phones.

    The 2030 was never intended to be a headphone amp, it's noisy and has mediocre bandwidth and distortion performance and as you said, has a min gain requirement. You are trying to turn Rosie O'Donnel into Angelina Jolie.
     
  4. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    It is not compensated to be stable at that gain.

    FYI: I designed the Fairchild version of the TDA 2030 at Fairchild in 1978. It is not a high performance audio part, it was pretty junky then and there are much better solutions if you are looking for top sound from headphones.
     
  5. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    You use a resistive divider or pot at the front end to reduce the amplitude of the source signal (which ultimately degrades noise factor).
     
  6. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    FYI, I designed a pretty good headphone amp with adjustable gain for my Sennheisers. Noise is so low it is inaudible and sound is perfect. You can substitute TLO71 op amps for the LF356 shown but doesn't improve the sound. The output has a DC offset so the headphones can be driven without a blocking cap and there will be no DC on them.
     
  7. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    Hi ronv,

    Not the solution I was going after. Noise and adjustment capability are problems.
     
  8. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    Hi bountyhunter,

    Oh no! Not Rosie O'Donnel!

    I thought it was a mite better than that. I assumed that with the increase in feedback (lower gain) the distortion would be cleaned up quite a bit. But I did not realize they were very noisy. Since I have the amp (complete with nice PCB) I thought I would give it a try. Sounds like a dead issue.

    As for the degradation caused by a resistor in series with the output, I am most concerned with bloated bass caused by high output resistance. I hate excessive, fat, flabby bass. Hello Rosie!

    Yes, that is the problem I was after. I have read that you can trade signal gain for noise gain (like a resistor from + to - input, etc.) and get around the compensation problem. But the math is a bit much for me.

    You meet the nicest people in forums. And talented too! Thanks for your input.

    Yes, I wanted to avoid that if possible.

    Nice design and I think I have about 90% of the parts on hand, including the rotary switch. I avoid Tant caps though and use electrolytics bypassed with film or ceramic instead.

    I am going to sim your design. I have been using LTS to sim diamond buffers and current mirrors somewhat like yours. I have found that a 100u cap from base to base on the output Transistors (as suggested by Jung) can have a very salutary effect on distortion. I guess your circuit does not need the help.

    A few questions though:

    1. What does "(2 PL)" mean in your schematic? If it is "two paralleled" do you mean two emitters share one emitter resistor?

    2. What is the max current you would suggest for a 2N3904/06 at 12V? I guess I am asking what is the max dissipation you allow for the output transistors?

    3. Did you thermally connect any of the transistors?

    4. Do the outputs need any heat sinking?

    5. Did you match any of the transistors? (I have about one hundred of each so I could if I had to.)

    6. Whew, and last, what alternative devices do you suggest for headphone amplification?

    Thanks for your help.
     
  9. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    Read the schematic in more detail.

    Please forget questions 4 and 5 and describe the physical nature of the thermal connections and heatsinking. I presume the driver/output 04s are together and the driver/output 06s are together and the current source 04s are together but are all of them on one sink?

    Thanks.
     
  10. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    should be OK.

    I took THD data on it way back when, it was very low.

    it means "2 places". Lazy way to avoid writing a part number twice.

    I am guessing maybe 300 mA peak. For headphones, it's not going to be an issue.

    I did. I used plastic T0-92 devices on the output transistors and the bias diode connected transistors. I got a thin piece of alum maybe 1" x 1/2" and Krazy glued them to it for thermal tracking. One for each channel.


    I don't think so for headphones, I did it for thermal tracking and stable bias current.

    I don't think it is necessary because I used degeneration resistors in the emitters. But, it would be better if you do.

    I haven't built any in a while but I know that National Semi (now TI) has some very high performance fully integrated amplifier IC's now that make the old 2030 look like a Model T. The selector on the TI site probably has a bunch.
     
  11. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    Oops, I think I asked the question improperly. I am examining operating point collector current on the outputs of about 25 mA. That will give better than 40 mA current swing before leaving Class A which should cover all loads the amp will ever likely see. At that bias level the outputs are dissipating about 300 mW with a 12V supply and getting really hot. Junction temp will be about 100C at a 40C ambient inside the case with no HS on the transistor. Will they take that abuse? Will a glued on sink help much?

    The derating specs 12 mW/C above 25C indicate that the 25C 1.5 W dissipation rating be lowered to 900 mW at 100C. That suggests we are still in the ballpark if I have applied the ratings to junction temp properly.

    The datasheet shows Rθ JA to be 200C/W and R JC to be 83C/W but there are no specs for case to sink. I suspect a 1 sq in sink will be a bit worse than 20C/W on its own but I cannot estimate the actual result with what I know.

    I have no experience with pushing the little guys that far. Thoughts? Or should I back off some and let the peaks wade around in crossover distortion?

    FWIW, when I was a young lad, about 50 years ago, I worked at the Goddard Space Flight Labs in Greenbelt Maryland as a NASA contractor. I was a tech. That was a long, long time ago with several intervening professions. I have forgotten most all of what I knew then and your help has been invaluable.

    Thank you for your time, patience and kindness. It is a real pleasure getting help from a pro like yourself.


    Thanks
     
  12. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I thought the idling current down the output on mine was more like 10 mA ballpark. It's set by the diode connected current sink at the bottom fed by a 1K resistor with about 10V across it, mirrored over to the output chains. That puts maybe 120 mW of power in them at idle.

    I recall the plastic TO-92 devices are about 180 C/W. The only thing I remember is that the transistors never got hot enough to notice with their flat sides glued to the small piece of aluminum. However, plastic conducts heat poorly so the heatsinking effect is not great using plastic devices.

    Not sure about cranking it up to 25 mA? Power diss at idle would be more like 0.3W, operating power a little higher. Even if the transistors rise 60C above ambient, they will still be OK. With IC's, we run junction temps of 100 - 110C all the time and still have very long operating lives. I would use a little heatsink for them.
     
    Last edited: Oct 16, 2013
  13. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    That is exactly what my sim shows.

    Current datasheets show the devices at 200 C/W.

    Great.

    One of my concerns.

    I have been at .3W and hoped that it would be fine.

    To me that means I can run Tj at 100-110C, glue on a little HS FWIW, and be OK because the devices should be able to stand that Tj even without the HS.

    Thank you for your insight and experience.

    I do have some questions re: simple buffer topology. I hope you can help.

    Here goes:

    The schematic below shows two simple but functional versions of common buffers used in Headphone amplifiers. The one on the left is a Complementary Symmetry Push-Pull (CSPP) buffer and the one on the right is a Diamond buffer.

    The cap in each circuit is as suggested by Jung and in my simms that cap does a great job at reducing distortion in Class A mode. If either circuit will spend considerable time in Class B, an additional cap can be placed from the top of R5/R105 to the bottom of R6/R106. This will improve distortion in the Class B region. I do not want to go there so I do not include those caps here.

    It seems to me that the CSPP is better suited for discrete construction. Thermal feedback in the CSPP is between identical transistors, NPN to NPN, PNP to PNP. They are easy to match if you wish and of course the bias of the drivers is set by the diode junctions of identical transistors that are thermally connected.

    The Diamond on the other hand has its output transistors biased by unlike transistors, NPN to PNP and PNP to NPN. These are much harder to match if you wish. Also, I have seen Diamond circuits on the net where the thermal feedback is NPN to NPN and PNP to PNP which makes no sense to me because the thermal feedback is not directed to the actual transistor that provides bias to a given output. Indeed, I have seen cases where the current in the drivers exceeds the current in the outputs and each transistor is on a separate HS with no thermal feedback at all.

    Am I missing something? What is it that makes the Diamond circuit favored over the CSPP?

    Well, here are the example circuits and thanks for any help you can give.


    [​IMG]
     
    Last edited: Oct 16, 2013
  14. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    http://phonoclone.com/diy-bb.html

    What I read is that in the diamond buffer, the two NPN's have to match (each other) and the two PNP's have to match (each other). The bias current is the same in all four transistors. I have never built that one.

    The CSPP looks like the typical output stage used in most amplifiers.
     
  15. oneoldude

    Thread Starter New Member

    Aug 7, 2013
    12
    0
    Yes, that is one of the sites I was talking about. I have simmed some of his work and the bias transistors may have more IC than the outputs. That seems a bit much even if you are relying on a current mirror. Also he temp compensates like transistors. That makes no sense to me because he is not temp compensating the bias transistor with the output transistor that the bias transistor is feeding. I understand it is a bit different in an OPA where everything is on one substrate. But in a discrete circuit shouldn't the temp compensation be between the bias transistor and the output it biases?

    I was not intending to publish anything novel. I was simply asking why the Diamond buffer seems to be preferred over the old standby CSPP circuit. Clearly, it is something I do not understand. Perhaps input impedance is higher? Perhaps the current mirror aspect stiffens the front end for better distortion? I don't know. Since Diamonds are used in OPAs I thought you might know.

    Fact is, in my simms the FFT distortion products are down more than 120 dB with both topologies. Yes the Diamond is a couple dB better, but when you are down that low does it really make any difference other than bragging rights? I understand that simulation is not the same as actual measurement, but I am comparing apples to apples.

    Thanks much for your time and effort.
     
  16. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I have no expertise on the diamond buffer.

    Nope. If distortion (or noise) are so low that nobody can hear it... then it doesn't matter.

    The claim I read is that the diamond has lower distortion, but I assume only if good matching exists. Sounds like a design that would be well suited for an IC, not so much discrete parts.
     
Loading...