Here’s a batch of audio cables and stuff I built. Unfortunately I haven’t taken time to really dive deep in the measurements yet.
As to going unbalanced output to differential, RCA tip to + and RCA shell to -, BUT, to keep the - from becoming a large antenna, you will then need to separately ground the unbalanced DUT, and use the analyzer differential input to subtract out the difference between the DUT’s ground and the analyzer’s ground. While the analyzer has a fairly high impedance from the BNC shells to USB power from the computer, all analyzer BNC shells are solidly connected together, so if your generator output is driving an unbalanced input, you’ve already connected the grounds together. It’s all part of why unbalanced sucks.
Hi @daleshirk. I find the mode of measurement you propose interesting. I have always followed the modes suggested at this link: Measurement-with-the-QA401 Download and have always obtained good results in noise reduction. But I would be curious to try the mode you propose. Could you, when time permits, make a diagram of the connections? Thank you
Hi @Claudio. The post to which I was responding by @HawkerAMpT concerned a processor with unbalanced RCA outputs, which would almost certainly be an ungrounded device, therefore my recommendation to ground it when using a differential input. 'Tis but the first step, not an alternate to the methods described in the paper you referenced.
The paper you linked looks like an excellent resource. It certainly references impeccable sources, Ott, Whitlock, Joffe, el.al. He gives an excellent analysis of the stray capacitance and coupled inductance modes of noise coupling. Oddly, he ignores shared impedance coupling and indeed his bonding scheme may (or may not) improve with a star connection layout, especially to the USB shield. Why should its current couple to the other devices. His use of a low-Z attenuator to reduce Johnson noise is excellent. His assumption that the black speaker terminal of the amplifier is connected to chassis ground is perhaps forgivable considering the date and hifi orientation, but it is in no way a given in pro sound today. Also he oddly bypasses the benefits of twisted pairs or coax in favor of tied bundles, but I lack personal experience to comment further on that. None the less, it is an excellent resource which I shall download and save. Thank you.
You’re making a smart move by shifting from the QA400 to the QA403, especially to take advantage of the differential inputs and outputs. This can significantly reduce noise and ground loop issues in your test setup, particularly when working with vintage single-ended gear like your 1970s Phase Linear preamps. Since you’re already building quality cables using Amphenol 69475 BNC connectors and Belden 83284 coax for your QA400, you’re clearly familiar with precision cable work. However, when it comes to using the QA403’s differential capabilities, a different approach is needed—primarily using shielded twisted pair (STP) cable rather than coax. STP is preferred because the twisted conductors help reject EMI via common-mode cancellation, and the shielding adds another layer of noise protection.
For connecting the QA403 differential outputs to a single-ended DUT input, use a twisted pair where the positive (hot) conductor goes to the RCA center pin, and the negative (cold/inverted) conductor is either tied to ground at the QA403 side only or left floating, depending on noise performance. The shield should also be connected only at the QA403 end. This prevents creating ground loops and keeps noise to a minimum. For input connections (i.e., DUT single-ended output into QA403’s differential input), you can connect the RCA center to the QA403’s IN+, and RCA ground to IN− if needed, or leave IN− floating. Again, connect the cable shield to the QA403 side only. This configuration creates a pseudo-differential input, which improves common-mode rejection without needing a full balanced signal. If you want to match impedance more closely, a resistor (1k–10kΩ) between IN− and ground can help, but it’s often not necessary for preamp testing.
As for cables, high-quality STP options like Mogami W2549 or Canare L-4E6S are excellent choices. These are widely used in professional audio and offer both shielding and the correct conductor arrangement for differential audio signals. Therefore, it’s best to source cable from audio-specific suppliers like Redco, Markertek, or even Canare and Mogami directly if possible.
In summary, to get the best performance from your QA403 with single-ended gear, build your cables using high-quality STP, maintain careful shield and conductor routing to avoid loops, and avoid grounding both ends of the shield. With that setup, your measurements will be cleaner and much more immune to interference.