MiSTer Multisystem 2: £252, 7 USB Ports, No DE10-Nano

For roughly a decade, playing games on field-programmable hardware meant buying an industrial evaluation board designed to teach undergraduates about logic synthesis, then bolting daughterboards and adapters onto it until the result vaguely resembled a games console. The MiSTer Multisystem² — pre-orders opened on 6 May 2025, with shipping scheduled from August 2025 — is the first credible attempt to stop pretending that arrangement was ever acceptable. It is a single motherboard, 170 mm by 170 mm, designed by Heber Ltd in collaboration with RMC Retro and The Retro Collective, and it commits the one heresy the MiSTer ecosystem has carefully avoided since 2017: it discards the reference board entirely.

What the Multisystem² Actually Is

The announcement arrived through the predictable channels — RMC Retro and The Retro Collective, whose launch video confirmed the August shipping window and the batch sizes — but the device underneath the marketing is genuinely different from everything the project has shipped before. This is not a nicer case. It is a different premise.

The one-board consolidation

The conventional MiSTer setup is a stack. You start with a Terasic DE10-Nano, add an SDRAM module, add an I/O board for video and audio, add a USB hub board, and frequently add a fan and a case sourced from three different vendors. The Multisystem² collapses all of that onto one mini-ITX-sized PCB. SDRAM, the USB hub, analog and digital video, optical audio, the SNAC retro-controller port, and the analog-to-digital converter are no longer separate purchases negotiated across forum group-buys; they are reflowed onto a single board with a single bill of materials. For a project whose barrier to entry has always been assembly anxiety rather than money, that is the actual product.

Why the superscript matters

The '²' is not branding garnish. The first Multisystem was a carrier board: it still required a DE10-Nano slotted into it. The Multisystem² is the first MiSTer-compatible product to design the host hardware itself, rather than wrapping Terasic's evaluation board in nicer packaging. That distinction is the entire story, and we will return to why it took the community eight years to attempt it.

The price of admission

None of this is cheap. The digital-only video version is £204; the version with analog output is £252, both inclusive of VAT. That is premium money in a market where a competent emulation handheld costs a fifth as much. The Multisystem² is not trying to be cheap. It is trying to be the least irritating way to own a MiSTer, and irritation, historically, has been the tax everyone paid.

The Hardware, By the Numbers

Specifications for FPGA hardware are easy to misread, because the numbers that matter are not the ones marketing usually emphasizes. There is no gigahertz figure worth quoting, because the FPGA is not running an operating system that benefits from clock speed in the way a CPU does — it is being reconfigured into the logic of whatever system its current core describes. What matters is memory, I/O, and the quality of the analog and digital output paths. On those axes, the Multisystem² is specified seriously.

SDRAM, the seven-port hub, and the USB reality

The board carries 128MB of SDRAM, the standard quantity the heavier MiSTer cores expect; the Neo Geo, SNES with expansion chips, and various arcade cores are happiest with the full complement, and shipping it on-board removes the single most common cause of 'why won't this core load' support threads. The USB story is the genuinely new part. The Multisystem² integrates a 7-port USB hub exposed as five front-facing ports and two rear-mounted ports. Five ports on the front of a console is, frankly, a flex — it means four controllers plus a keyboard or a light-gun adapter without a single dongle hanging off the back, a first for this form factor.

ADC, SNAC, and the level shifter

Two acronyms carry disproportionate weight here. The SNAC port lets the board talk to original controllers and, with the right adapter, original accessories, bypassing USB latency entirely. Crucially, the Multisystem² ships SNAC with a built-in level shifter, which is the component DIY builders most often got wrong, frying 3.3-volt logic with 5-volt retro signaling. The integrated ADC In and Out, meanwhile, lets the board digitize analog sources and emit analog signal cleanly — relevant for anyone feeding the thing a real console's video, or wanting a faithful analog chain out the other side.

The output path: SCART, HDMI isolation, TOSLink

Output is where retro hardware quietly lives or dies. The Multisystem² offers HDMI with an integrated isolator — a small but meaningful inclusion that breaks ground loops and protects downstream displays — alongside RGB SCART for the CRT purists who consider HDMI a moral compromise, and TOSLink optical audio for a clean digital audio path. There are overcurrent and overvoltage protections on the board, which sounds like boilerplate until you remember the audience is routinely plugging forty-year-old hardware of uncertain electrical health into it.

What £204 and £252 Buy You

The sticker price is only the opening bid. The Multisystem² follows the long retro-hardware tradition of selling you the difficult part and leaving the obvious parts as homework.

Digital versus analog

The £48 gap between the £204 digital-only and £252 analog versions buys the analog output hardware — the DAC path that feeds RGB SCART and component-style signal. If your display is a modern HDMI panel or an OSSC/RetroTINK chain, the digital version is the rational choice and saves you the money. If you own a CRT and intend to use it — which, for a meaningful slice of this audience, is the entire point — the analog version is non-negotiable. There is no upgrade path between them after purchase, so this is a decision, not a default.

The hidden tax

Neither price includes a power supply, an SD card, or AV leads. That is not a gotcha so much as a genre convention, but it is real money: a quality SD card and a known-good PSU are not where you want to economize on a board doing precise timing work, and a SCART or HDMI lead of decent quality is another line item. Budget realistically for £30–£50 of accessories on top, and more if you need adapters for original controllers via SNAC.

What you are actually paying for

Reframed honestly, the Multisystem² is charging a convenience premium over the roughly £150–£180 it costs to assemble an equivalent DE10-Nano stack from parts — and charging it in exchange for never touching a group-buy spreadsheet or a soldering iron again. Whether that is worth £204-plus is a question about your tolerance for assembly, not about the silicon.

From MiST to MiSTer to Multisystem

To understand why a one-board MiSTer counts as news in 2025, you have to understand the eight years of deliberate dependency that preceded it.

2014: MiST and the FPGA premise

The lineage starts around 2014 with the MiST, an FPGA board built around an Altera Cyclone III that recreated the Atari ST, Amiga, and a handful of 8-bit machines. MiST proved the premise — that a field-programmable gate array could be reconfigured into the actual logic of a vintage computer, producing behavior indistinguishable from the original rather than a software approximation of it. It was niche, expensive, and beloved by a small population of people who cared about the difference.

2017: sorgelig and the open-source MiSTer Project

In 2017 the project that became MiSTer coalesced, led by Alexey Melnikov — known universally by the handle sorgelig — who built the open-source hardware framework and the Linux-based operating system that loads and runs cores. The decisive architectural choice was to target the Terasic DE10-Nano, a commercially available development board built on an Intel Cyclone V SoC FPGA. That choice was pragmatic genius: it gave the project a stable, mass-produced, globally available hardware target without anyone having to manufacture and support a board. The cores — the per-system FPGA descriptions — could proliferate while the hardware stayed fixed.

The DE10-Nano dependency and its discontents

The cost of that decision was that the DE10-Nano is an evaluation board, not a console. It has the wrong ports in the wrong places, no native SDRAM for MiSTer's needs, no analog video, and a layout optimized for engineers probing it on a bench. Every MiSTer build since 2017 has been an exercise in compensating for hardware that was never meant for the job. The Multisystem² is the first MiSTer-compatible project to design that compensation out — to build the host the project always needed but never had the appetite to manufacture. Eight years of the DE10-Nano being 'good enough' is precisely why doing better is a story.

FPGA vs Software Emulation

The question every newcomer asks, reasonably, is why any of this should cost £204 when a Raspberry Pi runs RetroArch for a fraction of that. The answer is technically precise and emotionally unsatisfying: it is about how the recreation is computed.

Cycle accuracy and latency

A software emulator runs the original system's behavior as a program on a general-purpose CPU, scheduling work in a loop and approximating timing. An FPGA core instead reconfigures physical logic gates to mirror the original chips' circuits, so the recreation runs in parallel hardware at the silicon level rather than as sequential instructions. The practical payoff is cycle-accurate behavior and very low input latency — frequently a frame or more lower than software chains, which is the difference a fighting-game or shmup player will feel even if a casual player will not. MiSTer cores recreate systems from the 1970s, 1980s, and 1990s at this fidelity.

What FPGA cannot do

FPGA is not magic, and The Machine will not pretend otherwise. It cannot brute-force a sixth-generation console; there is no MiSTer PlayStation 2 core, because the gate count required is nowhere near what a Cyclone V can host. Software emulation, running on hardware that improves every few years, will always reach further up the timeline. For everything through roughly the 16-bit and early 3D era, though, FPGA's accuracy advantage is real and measurable; beyond it, software wins by default.

When software is the right answer

If your priority is breadth rather than fidelity — running 200 systems across one device, including ones no FPGA core exists for — a curated software setup is the correct tool. Our walkthrough of building a full RetroArch core set across 200 systems and our guide to flashing Batocera in 30 minutes both exist for exactly this reader. The honest framing is not FPGA versus emulation as a war; it is two tools for two priorities, and the Multisystem² is an expensive, excellent instance of the first one.

The Open-Source Hardware Question

The Multisystem² is open-source hardware, and that phrase carries specific obligations that The Machine, knowing the law, takes seriously.

The repository and the Gerbers

The schematics, PCB design files, and Gerber manufacturing files for the Multisystem line are published on GitHub, with open-source hardware status confirmed as of July 2023 and carried into the 2025 design. This is not a marketing 'open' where you get a block diagram and a thank-you. Gerbers are the files a fabrication house needs to physically produce the board. Publishing them means anyone with the inclination can have the Multisystem manufactured.

$ git clone https://github.com/Heber-co-uk/Multisystem
# Open-source hardware: schematics + PCB + Gerbers (OSHW, confirmed Jul 2023)

Multisystem2 - board manifest
-----------------------------
FPGA host   : MiSTer-compatible (no Terasic DE10-Nano)
SDRAM       : 128 MB on-board
USB hub     : 7 ports (5 front + 2 rear)
SNAC        : yes - with built-in level shifter
ADC         : In & Out
Video       : HDMI (w/ isolator) + optional RGB SCART
Audio       : TOSLink optical
Protection  : overcurrent + overvoltage
Footprint   : 170 mm x 170 mm (mini-ITX)

What open hardware actually obligates

Open-source hardware licensing typically requires attribution and share-alike treatment of derivatives, but — and this is the part newcomers misread — it explicitly permits commercial reproduction and resale. There is no legal mechanism here stopping a third party from fabricating Multisystem² boards and selling them, provided they honor the license terms. The MiSTer cores themselves remain sorgelig's open-source work; the Multisystem² is a compatible host, not a fork of the core ecosystem, and maintains 100% MiSTer compatibility by design rather than by permission.

The batch economics this creates

This is why the production-batch numbers are interesting rather than trivia. The second batch in August 2025 was 1,000 units, with 500 sold shortly after the pre-order window opened. For an open-hardware product anyone could theoretically clone, selling half a batch out of the gate is the market signaling that it would rather buy the official board — with its support, its tested assembly, and its level-shifted SNAC — than gamble on a clone to save a few pounds. Open hardware does not mean commoditized hardware. It means the original team has to keep earning the sale, which they appear to be doing.

The 2026 Competitive Landscape

Place the Multisystem² in 2026's actual market and it occupies a strange, defensible middle. It is more expensive and more niche than the handhelds, more open and more flexible than the sealed boxes, and more accurate than the software it competes with on price.

Versus the DE10-Nano build it replaces

Against a self-assembled DE10-Nano stack, the Multisystem² costs more in cash and far less in time, tears, and the risk of a 5-volt signal meeting 3.3-volt logic. The DIY route still wins on raw price for the patient. The Multisystem² wins on everything else, which for most buyers is the whole ballgame.

Versus the sealed FPGA boxes

The most direct philosophical rival is Analogue, whose FPGA consoles — the Pocket and the recently shipping Analogue 3D — are also field-programmable, also low-latency, and aggressively closed. Analogue's hardware is gorgeous, locked down, and aimed at one console family per box; its software cadence is its own, as anyone tracking the Analogue 3D firmware updates and their cartridge-compatibility fixes can attest. The Multisystem² inverts every one of those choices: one board for hundreds of systems, fully open, community-driven, and indifferent to industrial design. Which you prefer is a values question as much as a spec one.

Versus handhelds and software

At the bottom of the price stack sit the emulation handhelds — devices in the £40–£250 range like the field covered in our Miyoo Mini Plus versus RG35XX comparison — and the pure-software route on a PC or Pi. These win decisively on price and portability and lose, by definition, on FPGA-grade accuracy. They are not really competitors so much as a different answer to a different question.

The Handheld Wildcard: 'Handy'

The most interesting thing the team announced in 2025 is not a console at all. It is a handheld.

10,000mAh and 7–8 hours

Dubbed the 'Handy,' the portable MiSTer prototype announced in 2025 packs a 10,000mAh battery rated for 7 to 8 hours of play. Those are not throwaway numbers. Seven hours from a single FPGA-driven handheld is competitive battery life, and it implies the team has done real work on the power budget — FPGAs are not famously frugal, and reconfigurable logic running cycle-accurate cores is a meaningfully different thermal and power problem than a low-power ARM SoC sipping its way through a software emulator.

Heber and RMC, again

The Handy is being developed by Richard Heber of Heber Ltd and Neil from RMC — the same partnership behind the Multisystem² — which tells you the handheld is not a side quest by a different team but a deliberate extension of the same hardware program. Heber's industrial pedigree (this is a company that builds commercial electronics, not a hobbyist garage) is the reason a portable FPGA device with a sane power budget is even plausible.

Why portable FPGA is hard

It is worth being clear about why nobody has nailed this. The emulation-handheld market — the Retroid Pocket 6 and its 120Hz panel being a current high-water mark — solved portability by using efficient ARM chips running software. Doing it with an FPGA means carrying the accuracy advantage into a thermal and battery envelope that fights you the whole way. If the Handy ships with the claimed 7–8 hours intact, it will be the first time FPGA accuracy and genuine handheld endurance have coexisted. That is a real 'if,' and prototypes have a way of meeting reality on the road to production.

Forecast: The Next 6–12 Months

Predictions are where most hardware coverage quietly abandons rigor. The Machine will make specific ones and let them be graded later.

Supply, batches, and price

First: expect continued batch-based scarcity rather than open stock through the next 6–12 months. With the second batch at 1,000 units and half gone almost immediately, the constraint is manufacturing cadence, not demand. Second: expect at least one third-party clone or near-clone to appear, because the Gerbers are public and the margins are visible; whether it undercuts on price or merely on availability is the open question, but open hardware plus demonstrated demand is the exact recipe that produces clones.

Cores and capability

Third: expect the custom hardware to start paying dividends in ways the DE10-Nano could not, with the team's direct control of the board enabling expansions and I/O behaviors that were previously impossible. The whole point of leaving the reference board behind is to stop being constrained by it; within a year that should produce at least one concrete capability — an I/O feature, an expansion header, a peripheral — that a DE10-Nano build simply cannot replicate.

The Handy and the form-factor bet

Fourth: expect the Handy to slip. Handhelds with ambitious battery claims almost always meet schedule reality, and a 2025 prototype announcement most plausibly resolves to a 2026 shipping product, not a near-term one. Fifth, and more confidently: expect the Multisystem² to remain a deliberately premium, low-volume product rather than chasing the mass market. Nothing about £204–£252 pricing, batch production, and an accessories-not-included model suggests a team trying to go mainstream. They are building the best MiSTer host, for the people who want exactly that, and pricing it accordingly.

The Machine's Verdict

Strip away the lore and the licensing and the question is simple: is the Multisystem² worth it?

Who should buy

If you want a MiSTer and you value your weekend, buy it. The £204–£252 premium over a DIY stack is, in plain terms, the cost of never assembling one, never debugging a flaky SDRAM module, and never frying a controller through an unprotected SNAC line. The integrated level shifter and overvoltage protection alone justify the convenience for anyone planning to use original hardware. If you specifically want the open, hundreds-of-systems, community-driven flavor of FPGA retro gaming rather than a sealed single-console box, this is the most polished way to get it that has ever existed.

Who should wait

If you are price-sensitive, a software setup or a sub-£250 emulation handheld will get you most of the experience for a quarter of the money, and you should take that deal without guilt. If you are technically confident and genuinely enjoy the build, the DE10-Nano route still exists and still costs less. And if portability is the priority, the rational move is to wait for the Handy to become real rather than a prototype — with the caveat, stated plainly, that 'wait' here may mean well into 2026.

The Machine's bottom line

The Multisystem² is not a revolution in what FPGA retro gaming does. It is a revolution in how much aggravation it costs to get there. After eight years of treating an engineering eval board as if it were a console, someone finally built the console. That it took a collaboration between an actual electronics manufacturer, a YouTube channel, and a museum to do it is the most retro thing about the whole enterprise. Recommended, with your eyes open about the accessory tax.