So - there’s a lot of combinations! To keep this post reasonably short, I’m focussing on connecting the Dwarf to a typical amp input (unbalanced, high impedance), since that seems to be where the biggest problems lie. Rather than going through an actual amp which might have introduced its own noise, I plugged directly into my audio interface with instrument mode on. I wanted to focus only on noise on the Dwarf’s outputs, so I left the inputs unconnected and input gain at 0%.
The short version
A re-amping box works well for connecting the Dwarf to an amp without getting digital noise induced on the outputs.
The long version
I tested three different types of connection:
- a jack lead directly from the Dwarf’s output to the interface’s input, with:
a. TS connectors
b. TRS connectors (note that this does not create a balanced connection since the input is unbalanced) - a passive DI box connected in reverse between the Dwarf and the interface, i.e.
Dwarf → (TRS jack to XLR) → DI box → (TS jack lead) → interface - a re-amping box connected in between the Dwarf and the interface, i.e.
Dwarf → (TRS jack lead) → re-amping box → (TS jack lead) → interface
For both 1a and 1b, the jack lead’s shield was connected at both ends (i.e. grounded). 2 and 3 were both tested with ground connected and ground lifted.
Test equipment and conditions:
MOD Dwarf
- Output 1 to audio interface
- Output 2 not connected
- Outputs 1+2 gain 70%
- Headphone volume 0%, headphones not connected
- Inputs 1+2 not connected, gain 0%
- USB not connected
- Ground loop compensation ON
- Buffer length 128 samples
- Pedalboard loaded with test tone generator (440Hz sine wave at -6dB, muted during recordings)
- Powered by the stock, earthed power supply
Audio Interface: Focusrite Scarlett 4i4 connected to laptop.
All recordings made through Input 1 with:
- Instrument mode ON (unbalanced input, 1.5MΩ input impedance)
- Gain adjusted before each test so the test tone registers -6.0 dB on the input meter, to ensure recordings are comparable
- Air mode OFF
- Pad OFF
- 48V phantom power OFF (irrelevant in instrument mode anyway)
DI Box: Samson MD1 Mono Passive Direct Box
Reamping box: Orchid Electronics Amp Interface (following a tip from @QuestionMarc in another thread)
| Connection | Peak noise level (dB) | Description | Recording (boosted 48dB) |
|---|---|---|---|
| TS jack lead | -49.6 | Digital (inc. 375Hz) + hiss | MOD Dwarf output noise tests - TS jack lead INST.wav |
| TRS jack lead | -53.5 | Digital (inc. 375Hz) + hiss | MOD Dwarf output noise tests - TRS jack lead INST.wav |
| Reverse DI box, grounded | -77.5 | Hiss, high frequencies attenuated | MOD Dwarf output noise tests - Reverse DI grounded INST.wav |
| Reverse DI box, ground lifted | -77.0 | Hiss, high frequencies attenuated | MOD Dwarf output noise tests - Reverse DI ground lift INST.wav |
| Re-amp box, grounded | -48.3 | Digital (inc. 375Hz) + hiss | MOD Dwarf output noise tests - ReAmp grounded INST.wav |
| Re-amp box, ground lifted | -64.7 | Hiss, full range | MOD Dwarf output noise tests - ReAmp ground lift INST.wav |
Based on the numbers alone, the Reverse DI box looks best, however numbers don’t tell the whole story, as the object here is not just to get the lowest possible overall noise level but to eliminate, if possible, digital noise induced on the outputs while affecting the output signal as little as possible. The reverse DI recording has noticeably less high-frequency hiss than the re-amped signal - the question is, will the high frequencies of the actual signal be reduced as well?
To test this, I recorded the output of the CAPS White Noise plugin using the reverse DI and re-amp methods, and also using a TRS jack cable direct to the interface but this time with the input in line mode (and therefore balanced). Again, the gain was set for each recording so that the 440Hz test tone measured -6.0 dB on the input meter, to ensure the recordings are comparable:
- 0 - 3 secs : TRS jack lead, line level, balanced
- 3 - 6 secs : Re-amp box, ground lifted
- 6 - 9 secs : Reverse DI box, ground lifted
MOD Dwarf output noise tests - white noise - TRS line, ReAmp, Reverse DI.wav
This sounds a lot like the previous recordings of hiss but remember that now we’re listening to the output of the White Noise plugin, i.e. this is the signal, not the noise. The re-amping box sounds indistinguishable from the balanced line level recording (to my well-worn ears, at least), whereas the high frequencies of the signal are noticeable attenuated for the reverse DI box method. This isn’t really surprising, since the re-amping box was designed to be used exactly like this and will have had its transformer selected accordingly. Meanwhile, the DI box is getting the signal pushed through in the opposite direction to what it was designed for, so who knows what impedance mismatches are going on.
This high-frequency attenuation explains the apparently better noise figure for the reverse DI box - since it’s attenuating all high frequencies, signal and noise alike, the overall recorded noise will be lower in volume. However, the high frequencies of the signal will also be similarly attenuated, so the re-amping box is actually giving the better performance. Most importantly, the digital noise is gone.