Optical vs Coaxial vs USB

All three carry the same digital data. What differs is the maximum rate each one is specified for, whether it makes an electrical connection between your boxes, and what audio formats it will pass.

Those three differences are all published, and they decide the answer more cleanly than any argument about sound quality does — because at the level where these connections are working correctly, the bits arriving are identical.

The published limits, per input, per DAC

Every figure below is from the manufacturer’s own specification page. Notice how often the three columns disagree with each other on the same box.

Three findings drop out of that table immediately.

USB is the only input that exceeds 192 kHz on any DAC here. If a specification above 192 kHz matters to you, it is a USB-only specification.

FiiO caps optical at 96 kHz where coaxial gets 192 kHz — on both models. Same box, same data, half the rate, decided by which socket. Topping, SMSL and Schiit publish parity between the two. FiiO does not, and FiiO says so plainly on its own parameters page.

The iFi ZEN DAC 3 has USB and nothing else. It is an excellent converter that cannot connect to a television, a CD player, or anything without a computer-style USB output. That is a compatibility question, not a quality one, and it is the sort of thing worth checking before rather than after.

What each connection actually is

Optical (TOSLINK)

The signal travels as light down a plastic or glass fiber. That has one consequence that neither other connection can offer: there is no electrical connection between the two devices at all.

Ground loops — the hum you get when two mains-powered boxes are connected by a cable that also joins their grounds — are physically impossible over optical, because there is no conductor for a ground current to flow through. If you have hum you cannot chase down, optical is the one connection that eliminates an entire class of cause rather than mitigating it.

The cost is the rate ceiling, and on FiiO’s DACs a lower one than coaxial.

Coaxial (S/PDIF over RCA)

Electrical, over a 75 Ω cable with an RCA plug. Higher published rates than optical on the FiiO units and equal on the rest, no fiber to snap, and connectors that lock in more securely than TOSLINK.

It does make an electrical connection between your boxes, so it can carry a ground loop that optical would not.

USB

The only bidirectional connection of the three. The DAC tells the computer what it can do, and the computer sends data at that rate — which is why USB reaches rates the one-way S/PDIF formats do not.

It also carries power, which is worth noting on the DACs that use it. Schiit publishes 1.5 V RMS output on USB power and 2.0 V RMS with an external supply for the Modi 5 — a difference in output level decided by how you power it.

Manufacturers publish the USB interface chip because it matters: Topping uses an XMOS XU208 in the E30 II, SMSL an XMOS XU-316 in the SU-1, FiiO an XMOS XUF208 in the K7 and an SA9312L in the K11. Topping and Schiit both publish UAC2 compliance, which is what allows the higher rates without a manufacturer driver on most systems.

Bus power, and the specification it quietly changes

USB carries power as well as data, and several DACs use it — which is convenient until you read what it costs.

Schiit publishes two output levels for the Modi 5 depending on how it is powered: 1.5 V RMS on USB power, 2.0 V RMS with the external supply. That is a 2.5 dB difference in output level decided by which socket you plug into the wall — enough to make an A/B comparison against another source misleading if you are not matching levels.

The iFi ZEN DAC 3 is the pure case: USB-C is its only input and, in normal use, its only power. iFi publishes a line dynamic range of ≥ 113 dB(A) for it — the lowest figure of any DAC we cover, though still around 18.5 bits of resolution.

Meanwhile the units with separate supplies publish the strongest numbers: Topping 123 dB for the E30 II, SMSL 121 dB for the SU-1, FiiO ≥ 126 dB on the K11’s line out. That is not proof of causation — these are different chips and different designs — but the pattern is worth knowing, and the Schiit figure shows the effect directly on one product.

If your DAC offers an external supply, use it and check the published output level. It is the one specification that changes with nothing else changing.

The trap: what optical will and will not pass from a television

This is the most common real-world failure with optical and it has nothing to do with sound quality.

An optical connection carries a stereo PCM stream comfortably. It does not carry every surround format, and the devices we cover state their limits explicitly:

• WiiM Pro Plus:its optical input accepts up to 192 kHz/24-bit, “stereo PCM only” — WiiM states that Dolby Digital and DTS are explicitly unsupported.
• WiiM Ultra: optical to 192 kHz/24-bit, and separately an HDMI ARC input that WiiM publishes as accepting PCM and Dolby Digital 5.1. Same box, different socket, different format support.
• WiiM Amp: HDMI ARC input, stereo PCM only — WiiM publishes that Dolby Digital and DTS are not supported at all.
• Denon PMA-600NE: coaxial and optical inputs, published as linear PCM.

If you connect a television by optical and get silence on some content, this is usually why: the television is sending a bitstream the receiving device does not decode. The fix is normally to set the television’s audio output to PCM, or to use HDMI eARC instead, which is specified to carry the high-bitrate formats optical cannot.

The box that solves the opposite problem

The Topping D10s is worth singling out because it runs the chain backwards. Its only input is USB — and it publishes RCA line out plus optical and coaxial S/PDIF outputs.

That makes it a USB-to-S/PDIF bridge as well as a DAC. If you have a good older receiver with optical and coaxial inputs but no USB, this is the part that connects a modern computer to it without replacing anything. It is a small point that solves a specific and common problem, and it is why the D10s heads this page.

The Bluesound Node Nano is the counter-example: it publishes an optical TOSLINK output and no coaxial output at all. If you own an external DAC whose best input is coaxial, that single missing socket decides the pairing.

Does the connection change the sound?

When all three are working within their specification, they deliver identical data. There is a long-running argument about jitter — timing variation in the recovered clock — that differs between transports.

We are not going to adjudicate it, because none of the manufacturers here publish a jitter figure we could compare. Topping, Schiit, SMSL and FiiO all publish THD+N and signal-to-noise; none publishes jitter per input. In the absence of published data, we would rather say we do not know than repeat an argument as though it were a measurement.

How to choose

1. Computer or phone as the source? USB. It is the only input that goes above 192 kHz on any DAC here and the only one that negotiates rates rather than assuming them.
2. Television as the source? Optical, and set the television to PCM output — or use HDMI eARC if the formats matter.
3. Hum you cannot trace? Optical, because it physically cannot carry a ground loop.
4. A source with both S/PDIF outputs, and a FiiO DAC? Coaxial. FiiO publishes twice the sample rate on it.
5. Old receiver, new computer? A Topping D10s in between, using its S/PDIF outputs as a bridge.