Problem with DRV134 output signal balance

Discussion in 'The Projects Forum' started by MrShhh, Sep 27, 2009.

  1. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Hi, I have made a simple impedance matching circuit for changing high Z audio signals to low Z (DI box):

    http://i985.photobucket.com/albums/ae333/MrShhh_2009/DRV134DIbox.jpg

    It uses an OPA134 unity gain buffer stage driving a DRV134 balanced line driver.

    All op-amp power supplies are bypassed with 4.7uF tantalum and 100nF ceramic caps and the power supply is bypassed with 100uF electrolytic & 100nF ceramic caps where it enters the pcb (not shown).

    When I input various sine wave test signals at -6dB (0.388 Vrms), from a Zo=50ohm source, the signals at the DRV134 outputs (Vo+ and Vo-) are way out of balance at frequencies below 10k. They go out again above 15k but the low end is worse:

    .........(mV rms)
    Hz.....Vo+...Vo-
    250....404....430
    1k......366....430
    5k......379....398
    10k....391....391
    15k....391....391
    17.5k..404...398
    20k.....391...385
    The figures were collected with the outputs unloaded. When I connect the outputs to my active monitors (Zin = 20kohm) the above figures get way, way worse. It doesn't matter what input signal strength I use, if I double it to 0dB for instance, the imbalance doubles.

    Is this normal behavior for the DRV134?

    The worst case quoted signal balance ratio for the DRV134 is 35dB @ 1kHz using the test rig on page 10 of the following datsheet:

    http://www.datasheetarchive.com/dow...om/pdf-datasheets/Datasheets-7/DSA-125103.pdf

    I have tried to the following mods with zero change in the imbalance:

    1. Direct couple the OPA134 to the DRV134 (remove C2)
    2. Remove the 10uF non-polar caps at the sense/output pins
    3. Use bigger caps at the sense/output pins
    4. Change the OPA134 to NE5534, TL071 etc.

    Is there a way to improve the situation? (Other than sending the output of the DRV134 to an audio transformer to re-balance it).

    Cheers, Mark
     
  2. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,373
    1,157
    What is the load you are using on the output?

    Have you tried with a resistive load on the outputs? Say a couple of 20k or 10k resistors, for testing purposes.
     
    Last edited: Sep 27, 2009
  3. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Hi JoeJester,

    The figures in the table were taken with the output unloaded.

    When I load the output by attaching it to the balanced input of my active monitors (Zin = 20kohm) the figures get a lot worse. I don't understand why because the output impedance of the DRV134 is only 50ohm. 50ohm into 20kohm should not present a problem. The XLR cable is short, only 3 feet or so.

    I will try the signal balance ratio test circuit on p10 of the DRV134 datasheet later today and post the results.

    Thanks
     
  4. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Something else I noticed:

    If I plug an XLR cable attached to my monitors into the output, the signal balance improves while the monitors are switched off. Soon as I turn them on, the signal balance goes badly out. So it seems there is some kind of loading issue...
     
  5. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Ok I have the following test circuit:

    http://i985.photobucket.com/albums/ae333/MrShhh_2009/DRV134SBRtestcircuit.jpg

    Signal balance ratio (SBR) = -20log(Vo/Vi)

    And the results (which seem to interest only myself :)) are:

    Hz.....Vo (mV rms)....SBR (dB)
    250..........6...............42.2
    1k............5.7............42.7
    5k............5.9............42.4
    10k..........6.2............41.9
    15k..........6.7............41.5
    17.5k.......6.9............41.0
    A 0dB (0.775V rms) sinewave test signal was used, not 10Vpp as suggested in the datasheet, so I expect the actual SBR according to Burr Brown's test standards at least, is maybe a bit better. I am happy with 42dB. It is not as good as 54dB and not as bad as 35dB. I will leave you in peace.

    BTW: I have learnt something today! Even though the unloaded output signals were the most unbalanced at 1k (see first post), when they are loaded and summed, they do not sum to give the worst SBR. In fact it is the opposite. This seems to be counter-intuitive, which is why I love doing this kind of stuff :D :D

    PS This is almost certainly the reason why Burr Brown quote the SBR at 1kHz (marketing is everything right?)
     
  6. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Your input capacitor will make the OPA134 output vary about 0 volts. Is there some reason for the 100Uf coupling cap at the input of the DRV134? You are also running the op amp at the extreme of its absolute maximum power inputs. Is there some reason for this?
     
  7. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Hi beenthere, the cap at the input of the DRV134 was to remove any offset voltage at the output of the OPA134 (which admittedly is tiny). It could be removed altogether. I tried this, but it has no impact on the signal balance at the DRV134 outputs.

    I chose the maximum power supply values as both the op amps perform (according to their datasheets) at their best at +/- 18v (greater headroom, less distortion etc.)

    How do you mean the input capacitor will make the OPA134 output vary about 0 volts? Is this a problem that needs resolving?

    Many thanks :)
     
  8. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    I simply do not understand the purpose of your circuit. The DRV134 is a single-ended to balanced line driver. This is handy for audio work where you have to drive a signal down a long cable and not pick up noise. What is your application?

    The data sheet for the device indicates the feedback pins only need the capacitive coupling when you have "large DC cable offset errors". Do you?

    While the DRV134 is rated at +/-18 volts, that is the absolute maximum for the OPA134. You speak of line level signals, and there is no gain through the op amp. Use regulators and reduce the voltage to the op amp to something more like +/-12 volts. The device is on the edge of failure.

    What is the purpose of the 1000 Uf coupling capacitor? If it is electrolytic, it is likely to cause problems if the signal in is AC in nature. The insulating film in the capacitor only functions if the voltage across it is of the correct polarity. The data sheet for the DRV134 does not indicate any need for such coupling.

    You might be better to construct your circuit according to the data sheet's simplest configuration, and see if there is any need at all for further elaboration.
     
  9. MrShhh

    Thread Starter Member

    Jul 7, 2009
    24
    0
    Thanks again beenthere, here is my reasoning behind everything. If anything doesn't hold up to scrutiny please let me know :):

    Say you have an electric guitar with a nominal output impedance of 50kohm. You cannot then plug that into the Zi=600ohm mic-pre amp of say, a mixing desk and expect all of your signal to survive. The bass will disappear and it will sound thin and have a very poor signal to noise ratio because the guitar signal loads down the mic-pre.

    The purpose of the circuit is to take a z=50kohm signal and convert it to a z=50ohm signal suitable for driving a 600ohm load. It is also for balancing an unbalanced signal which gives much better noise rejection, even on short cable runs.

    The Zin of the DRV134 is only 10kohm, so you cannot simply plug the 50kohm guitar signal into the DRV134 without loading down the DRV134 and again causing signal loss, so you need a very high Zin, very low Zo buffer stage to drive the DRV134, hence the OPA134. P10 of the DRV134 datasheet under "Layout Considerations" states: "the DRV134 input should be driven by a low impedance source such as an op-amp or buffer".

    I used the coupling caps just to see if they make an audible difference with different cables that I have. Curiosity is a good thing, yes? I have jumpers on the pcb to take them in or out of the circuit.

    I am using 7818 & 7918 reg's. According to the datasheet, the OPA134 THD+noise doubles if you drop the supply from +/- 18v to +/-16v. I do not want double the THD+noise. I am trying to make an ultra-low distortion DI box that passes as clean a signal as possible. Using a lower voltage means the output signal will also distort earlier when driven hard (ie with large a input signal from our z=50k guitar). The datasheets clearly state both op-amps can be run from +/-18v. I don't understand why Burr Brown would state this if it isn't true? They make it seem like a selling point...

    There is a small DC offset voltage present at the output of the OPA134 which is unavoidable due to limitations in the manufacture of the op-amp. The 1000uF cap blocks this DC offset preventing if from causing any instability in the DRV134, while allowing the AC audio signal to pass. It can be left out altogether, but for the sake of completeness, I wanted to try it out to see if I could hear any audible benefit. I chose 1000uF because it is large enough to be all but invisible to audio frequencies. If the cap is small enough, it will cause signal distortion due to the capacitative reactance of the cap becoming large at low frequencies (Xc = 1/[2.pi.f.C])

    It is a bipolar electrolytic cap, polarity is not an issue, apologies, I mistakenly marked it as polar on the schematic.
     
  10. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    What happens when you try terminating the signal line (shielded twisted pair) with 600 ohms?
     
Loading...