1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Pseudo Balanced Amplifier (PBA)

  1. earfonia
    Earfonia PBA
    (Pseudo Balanced Amplifier)

    2 channels, 3 channels, or 4 channels amplifier (balanced), how much improvement will we get by having more channels? I started a proof of concept project last year to test it. This project is not about making an amp to be compared with other amp, but a proof of concept project to compare the sound characteristic of 2, 3, and 4 channels topology within the same amplifier.
     
    P1080401w_info.jpg

    To simplified the circuit, I use Op-Amp with simple non-inverting circuit (a.k.a CMoy) for every channel.  I initially tried to make a balanced amp with 4 non-inverting channels (not bridge configuration), using resistor voltage divider to get the inverted signal, but failed on the grounding.  Then I make use the board and the power supply to make 4 channels amplifier, to make what I called a Pseudo Balanced Amplifier.  It is a 4 channels amp, looks like a balanced amp, but not a balanced amplifier because there is no inverted signal, only 2 independent ground channels for each driver, as shown in the diagram below. The 4 channels output is what I called the Pseudo Balanced Amplifier (since I haven't seen this topology on the web yet, so let me coin the term for it, Pseudo Balanced Amplifier, or PBA).
     
    EarfoniaPBASimplifiedSchematic.png
     
    EarfoniaPBASchematic.png
     
    Some information about the amplifier circuit:
    DC Coupled
    Op-Amp: OPA552 (200 mA output current, 60 volt max supply)
    Resistors: Vishay Dale
    Gain: 13
    Volume Pot: ALPS RK27 20k log.
    Power Supply: Dual supply +/- 28 volt regulated (LM317 & LM337), 375 mA per channel.
    Electrolytic Capacitors: Panasonic FC series, except the biggest cap is panasonic EC series.
     
    The schematic for the power supply:
    EarfoniaPBAPowerSupply13.png

    So it is a kind of high voltage, high power, 4 channels CMoy :)

    OPA552 is a cranky op-amp, rather high DC offset at the output. I've tried many resistor combinations, and I found that to lower the DC offset, I have to use low value resistors. I also found that in the same circuit, different OP552 chip will have different DC offset value by a lot of margin. I bought 8x OPA552 chips to select 4 chips with DC offset less than 4 mV in the PBA circuit.

    Since OPA552 is stable for gain 5 or higher, I set the gain at 13, rather high, my purpose is to drive the Orthos :)  The ground channels gain is also set at the same gain, 13. So far It sounds alright with gain 13 on the ground channels :)  It seems it is not necessary to set unity gain on the ground channel :)
     
    P1080389w.jpg
     
    The 2 empty op-amp sockets in the middle of the op-amp row, are from the previous project (SBA Project), and not used in the PBA circuit.
     
    The setup I use to test my PBA:
    Balanced headphone, Hifiman HE5-LE with DIY Silver Dragon v2 cable.
    Computer with Dr. DAC2 DX (function as DAC), mostly playing 24bit / 96kHz or 24bit / 88kHz FLAC files.
    FYI, my Dr. DAC2 DX Op-Amps have been upgraded with OPA2107 for the FDO section, and OPA1642 for the Line Out and Headphone Amp section.
     
    P1080407w.jpg
     
    I use DIY 4 pin XLR to TRS jack adapter to test the 2 channels and 3 channels output.
    P1080334w.jpg
     
    Sound Characteristic

    The overall OPA552 sound characteristic in this high voltage circuit, Burr Brown sound, a tad midcentric, with nice liquid sounding, full bodied midrange. Midrange is thicker than OPA1642, more neutral and not as warm as OPA2604. Bass is proportionally good, full bodied with good control and texture. Treble is smooth, not as bright and transparent as most of the AD op-amps, also less transparent than the OPA1642, but overall sound is not dark, still quite neutral. Detail is good, better than OPA2604, but not as detailed as OPA1642.  Presentation also quite neutral, not as laid back as OPA2107 or OPA627/637.

    The overall sound differences between the 2, 3, and 4 channels are minimal, not much.  Untrained ears might not be able to distinguish the difference.  But when listened carefully, with some songs, the differences is quite noticeable.

    2 Channels
    The 2 channels output ground connected directly to the power supply ground.
    The tonal balance is slightly more forward, less smooth, more grain compared to the 3 & 4 channels.  Vocal sounds thinner and less warm.  The Xylophone sound in song 'Nows The Time' from the album Jazz At The Pawnshop, sounds less weight and dry. The overall sound is also less airy and spacious.

    3 Channels
    The 3 channels use one active ground channel.  One channel of the non-inverting OPA552 circuit is used for the ground channel. It sounds more natural overall. More body to the midrange, vocal sounds fuller and slightly (just slightly) warmer. The Xylophone sound in song 'Nows The Time' from the album Jazz At The Pawnshop, sounds weightier with more body. It sounds like the Xylophone metal is thicker.  Imaging and instruments separation is about the same as the 2 channels, maybe slightly better in 3 channels, but not much.
     
    4 Channels
    The 4 channels output (PBA output) use independent active ground for each driver.  The PBA output sounds best IMO.  Vocal and instruments sounds fuller.  Tonal balance is similar as the 3 channels, but bass is tighter with better control, imaging is noticeably better, more 3D sounding, instruments separation is also clearer, better focus. For comparison, The Xylophone sound in song 'Nows The Time' from the album Jazz At The Pawnshop, sounds more realistic than the 2 and 3 channels output, with better body to the sound.
     
    That's my impression so far. On my PBA, I found the 3 channels output sounds better than the 2 channels, and the PBA output (4 channels) sounds best.  Will test with more balanced headphone later, need to find time to re-cable some of my Headphones. My PBA is currently being tested by my friend Mike from Headfonia, he tested it with Hifiman HE6 which is famous for being difficult to drive.  He will post his impression as a short article in Headfonia later.
     
    Hopefully PBA topology can be another simple alternative for balanced headphones :)
     
     
    18 May 2011 update:
     
    Short article about PBA in Headfonia:
    http://www.headfonia.com/earfonias-pseudo-balanced-amplifier-pba/
     
    I agree with Mike:
    ...So affordable that the PBA can also stand for the “Poor-man’s Balanced Amplifier”.
     
  2. Yoga Flame
    Nice! Did you notice any difference in the amount of crosstalk?
     
  3. earfonia
    I haven't check (or measured) the crosstalk yet, but I didn't notice any crosstalk, maybe because I use balanced cable, separate ground wire for each driver.  From my experience, crosstalk mostly caused by shared single wire ground in the headphone cable.  When we use the 2 ground wires in the headphone cable, one for each driver, crosstalk usually eliminated, even with unbalanced headphone output.
     
  4. Avro_Arrow
    Crosstalk can be generated anywhere along the signal chain all the way back to the power supply.
    No topology can eliminate crosstalk generated before or after it.
     
    dhruvmeena96 likes this.
  5. earfonia
    Right, if the crosstalk happen before the output, separate ground wires on the headphone cable will not eliminate them. But usually from my experience, for headphone system, the major source of crosstalk is the single ground wire in the headphone cable. The crosstalk from source and amplifier usually very small.
     
    Some test I did last time:
    http://www.head-fi.org/forum/thread/484744/crosstalk-on-audio-technica-m50#post_6601078
     
  6. jcx
    looks like a great start
     
    how about putting the same connector on all amp configurations - since if crosstalk from common impedance is an issue then I can assure you that the TRS jack/plug contact resistance causes measurable crosstalk where a 4 pin with multicontact socket design will have much lower common gnd impedance
     
    further the differing cable connectors suggests you can't have DBT ABX  tested the 4 pin cable against different amp gnd configurations
     
    it would be good to have "wrung out" all measurable differences with high rez soundcard measurements at the far end of the (loaded) cable - you should clearly see added noise from the extra op amps in the gnd channels - balanced input like the Juli@ make the measurements easier - two tone cross channel IMD distortion may also be visible with "3-channel" active gnds in series with the outputs
     
  7. earfonia
    @jcx: Thanks for your suggestion!
    My main concern with my PBA project is not really to measure all the technicalities, but to listen with my ears, which output sounds better.  And I confirm, the PBA output, the 4 channels, sounds best from other outputs.
     
    Maybe the active ground channel adds noise, maybe measurable noise, but if I don't hear it on my Ortho, I just don't really bother with all the measurement result :)
     
    I'm EE graduate, so a bit familiar with those measurement things, but one thing I learn, my ears are the judge for the final result, that's what matter most for me, that it sounds good or not, regardless the measurement result :)
     
  8. Uncle Erik Contributor
    Really interesting project!

    I love it when people actually put theory to the test. Please keep us updated with your tests and changes!

    Also, I wonder if the results would be similar with other circuits.
     
  9. earfonia
    I found only a handful articles / threads regarding the OPA552 and how it sounds, so that's also another thing that trigger me to try OPA552.  I like the OPA552 sound, but as other Op-Amps, it requires matching with the headphones.  It sounds good with my DT880 Pro 600 and Hifiman HE5-LE, but it doesn't match very well with my Shure SRH-840 and ATH-M50, a bit too much midrange on those headphones.
     
    Currently my PBA is being tested by my friend Mike.  Once I received it back, i will do more test with more headphones, take notes of the crosstalk and noise.  I tried once with cheapo, very sensitive IEM, I guess 16 ohms, I could hear some soft AC interference noise.  But with higher impedance headphones, like M50 and SRH840, PBA sounds clean, I didn't hear that AC interference noise.  I guess I put the AC transformer too close to the audio circuit.  I will take note of this on the next project.
     
    Previously I wanted to build a balanced amp with OPA552, but failed.  I should have tried bridge topology next time I build a balanced amp.  But on TINA TI models, sometime I saw some phase distortion between the positive output and the negative output.  Only a very small degree of phase distortion, and only for high frequency like 10 kHz and above, but that makes me wonder, Is this phase distortion a common problem in designing Balanced Amplifier, that the positive output slightly out of phase, meaning not exactly 180 degree with the negative output?
     
    Pseudo Balanced in this case should not have that phase distortion problem, since there is no negative output.  And hopefully will have the same improvement of driver control ability as the fully Balanced amp.  I interested to see how PBA concept compared to fully balanced circuit such as bridge :)  Maybe the full balanced concept will win in slew rate :)
     
  10. earfonia
    Short article about PBA in Headfonia:
    http://www.headfonia.com/earfonias-pseudo-balanced-amplifier-pba/
     
  11. Packdemon
    So if the PBA does not have the invertion of a true BA, would that mean that you still hear the EMI noise?
     
  12. Rroff
    Unfortunately not an area I'm an expert in - I've built something fairly close to the PBA that has fairly decent resilience to EMI/RFI via LC filters in the power supply and RC filter on the input.
     
  13. carlmart
    Have you tried a balancing chip, like the THAT 1646, to achieve that same balanced out?

    https://www.mouser.com/datasheet/2/407/1600data-6754.pdf
     

Share This Page