Buying a vintage sampler is a decision that resists justification. Your laptop can load longer samples at higher resolution with better filters and no floppy drive. The honest reasons are about constraint and character, and the useful version of that argument is specific rather than mystical.
Ian Dixon — whose free browser-based Web MIDI editors for classic synths we wrote about yesterday — has reviewed the E-Mu Emax II, and his answer covers the history, the workflow, the sound library and the filters.
What the Emax II is
Released in 1989, a 16-bit digital sampling workstation, and E-Mu’s successor to the original 12-bit Emax. The 1990 rack version has 16 voices and digital resonant filters.
And the detail that makes it more than a spec bump: it dropped the analogue filters of its predecessor.
The 12-bit-versus-16-bit question is not what people assume
The instinct is that 16-bit is straightforwardly better than 12-bit, and in measurable terms it is — roughly 24dB more dynamic range. But in the vintage sampler market the 12-bit machines are the ones with the cult, and the Emax/Emax II pair is a clean illustration of why.
The original Emax was 12-bit with analogue filters — specifically E-Mu’s own SSM-style VCF per voice. Its character came from two things stacked: the quantisation noise and aliasing of 12-bit conversion, and then analogue filtering smoothing the result. That is the same recipe as the Emulator II, the SP-1200, the Mirage and the S900 — the machines people still pay a premium for.
The Emax II improved the converter and replaced the analogue filters with digital ones. Cleaner, more accurate, more flexible — and the specific grit-plus-warmth interaction was gone. This is a recurring pattern across late-80s digital audio: the generation that “fixed” the previous one’s limitations removed the thing people later wanted.
Which makes the Emax II genuinely interesting rather than merely superseded. It sits at the hinge. 16-bit resolution with a digital resonant filter designed in 1989 — a filter implemented with the arithmetic available on period hardware, which is to say with its own character, different from both an analogue ladder and a modern software filter. Dixon’s review covering the filters specifically is the right emphasis.
The Gotek drive is the actual answer to “is it usable?”
Dixon demonstrates building a personal sound collection through the Gotek drive workflow, and this deserves to be the headline for anyone considering any late-80s or 90s sampler.
A Gotek is a floppy emulator. It is a drop-in replacement for the original 3.5” drive that reads disk images from a USB stick instead of magnetic media. They cost around $25–40, and with community firmware (FlashFloppy being the standard) they handle the non-standard disk formats these machines use.
Why this changes everything:
Original floppies are dying and not coming back. Thirty-five-year-old magnetic media fails, and the drives themselves fail more often — belts perish, heads drift, mechanisms seize. The single most common reason a vintage sampler is unusable is its drive.
Disk images are copyable and backupable. Your sound library stops being a shoebox of deteriorating plastic and becomes a folder you can back up.
Loading becomes fast and silent. No seek noise, no waiting, no failed reads halfway through a bank.
And the factory libraries are archived. Large parts of the original E-Mu sound libraries exist as disk images online, so a machine bought without disks is not a machine without sounds.
The practical consequence: the gating question on a vintage sampler is no longer the drive. It is whether the sample RAM is good, whether the outputs are quiet, and whether the display is readable.
”Vintage samplers still feel different from modern software”
Dixon’s framing, and it is the right claim to make because it is a claim about feel rather than a claim about sound quality that would not survive a null test.
What is actually different, concretely:
One sample at a time, and you have to commit. Limited RAM means every sample is a decision. You sample, you trim, you tune, you keep it — and the next one requires deleting something. That constraint produces different music from an infinite browser of pre-cleared loops, and the difference is not subtle.
The signal path is fixed. Converter, filter, output stage, and that is what you get. No plugin chain to agonise over. A fixed path you learn the sound of is faster to work with than an infinitely adjustable one.
And it does not have a screen full of options. A two-line display and a data slider is a constraint on second-guessing yourself.
None of which is an argument that you need one. It is an argument that the constraint is the product, which is the only honest case for vintage hardware, and Dixon makes it without overclaiming.