0

I'm designing an audio mixer and took good care to create a relatively low THD signal path. At the final output stage I'm measuring a a THD of 0.06%, (10 Vpp, sine @ 1 kHz.)

This is measured directly at the op-amp output. After that I added a resistor to protect the output, a 360 Ω 0805 SMD resistor.

When I measure after the resistor I found that the THD jumped up to 0.7%.

First I thought it was because of a bad resistor, so I replaced it with a high-quality Susumu thin film (RG2012P-361-B-T5). This did not fix the issue, unfortunately.

Lowering the resistance does make it better, I'm at 0.15% with a 51 Ω resistor, but I would really like to add some output resistance.

The output after the resistor is floating when nothing is connected. There is a nominal 10 kΩ - 100 kΩ input impedance in the next device.

I'm really scratching my head as to what could cause this.

Updated schematic and measurement

edit:

Turned out to be a small non-linearity by a component in the signal path I initially overlooked.. Moving the RHS of R2 to the other side of R34 fixed the issue (ie zero output impedance config), as proposed by danmcb.

Matt
  • 1
  • 1
  • Exactly how are you measuring it? – Bruce Abbott Oct 05 '22 at 06:54
  • Look into the input impedance of whatever the measuring device is. Sounds like it is a non-linear impedance. If necessary, use a low distortion amp as a buffer ahead of it. –  Oct 05 '22 at 07:03
  • I've added my measuring set up to the question. Thanks for the suggestion will try with a buffer – Matt Oct 05 '22 at 07:18
  • Is the distortion analyzer grounded in the same way in both cases? Does the box labeled "device under test" include capacitance between the opamp's negative input and ground that could make it unstable? – bobflux Oct 05 '22 at 07:47
  • yes the distortion analyzer is grounded, with the positive probe measuring before and after the resistor. I'll look into the capacitance, maybe I need to add a bit more in the feedback loop to compensate – Matt Oct 05 '22 at 07:54
  • try moving the RHS of R2 to the other side of R34. BTW 360R is very high for a resistor setting output impedance. 37 to 75 is typical. By the way, you could save yourself a lot of time and invest in Doug Self's book on small signal audio design. He has figured all this stuff out (and his designs are in many pro consoles). – danmcb Oct 05 '22 at 07:56
  • 1
    Thanks, will try it! Yes I've read Self's book, really amazing book – Matt Oct 05 '22 at 08:04
  • ok, then you already have some excellent reference designs. I'm at a loss as to why you would want 360R there. BTW this resistor is typically added in most designs to guarantee stability when driving a capacitive load (i.e. cable). Modern opamps (5532 etc, ok not even modern) do not need it for protection. They will not be damaged even driving into a dead short generally. Adding a higher resistor is just degrading your MOL into 600R which is considered a key spec in a mixer. – danmcb Oct 05 '22 at 09:56
  • If you want "only" to protect your output (opamp), just include the 360 Resistor within the feedback loop ... between opamp ouput and R2//C2. – Antonio51 Oct 05 '22 at 10:50
  • Is the signal generator connected directly to the opamp ? If so, signal gen and opamp will **disagree** about the voltage at that node and **saturate**. This will obviously lead to distortion, but I don't see how R34 would fix it. – tobalt Oct 05 '22 at 10:58
  • Ok yes, the relatively high output resistance is to add some protection, because in the use environment it is possible to accidentally short to GND, but also to VEE or VCC rails. It also needs to be able to drive a 2m cable without a problem. I think the best solution is to move the resistor within the feedback loop – Matt Oct 05 '22 at 11:30
  • @tobalt no it's not directly connected, I didn't include all the other circuitry that's within DUT – Matt Oct 05 '22 at 11:33

2 Answers2

2

Well, without anything connected, the voltage would be identical. What is different is your measuring device and/or "floating". One guess would be that your THD is due to overshoot of the opamp, and your measuring device adds enough additional capacitance to squash it when connected directly.

user107063
  • 2,507
  • 1
  • 13
  • Ok good suggestion, I'll try adding additional capacitance to see if that does anything. I'd think 100p is already quite a lot, but there's a VCA stage before it that requires quite a bit of capacitance to function correctly. – Matt Oct 05 '22 at 07:21
1

If your scope is grounding the right hand side of R34 then you have applied a 360 ohm load to U1B and at 5 V peak that will be demanding 5 / 360 = 14 mA from the op-amp. That might explain it.

Instead connect a 10k resistor from the right hand side of R34 to ground and measure the voltage across it. That will resemble a real load more accurately.

Transistor
  • 168,990
  • 12
  • 186
  • 385
  • If the input probe for my analyzer has 1MΩ impedance wouldn't that prevent loading? Or am I misunderstanding how that works? – Matt Oct 05 '22 at 07:55
  • It wasn't clear from your diagram what the split in the probe wiring is. If it's not probe and earth then forget my suggestion. I was just concerned that you had grounded the right-hand side of R34 through the scope while the probe was connected to the left. – Transistor Oct 05 '22 at 10:23