Building a New PC in 2025: Windows 11 Cometh

Have a look at this:

It's a fourteen-year-old PC. I built it from parts back in 2011. Have you ever taken a photograph of your living room, and when you look at the picture everything look ugly and you can see all the dirt? I have the same feeling right now. The PC looks grotty on the outside, and it also looks pretty grotty on the inside, but for most of the last fourteen years it has been the tool of my writing trade.

Back in 2011 I wanted to leap into the world of 64-bit computing, so I bought a bunch of parts and stuffed them into a case. Goodbye 32-bit Windows XP and 4gb of memory and postage-stamp-sized videos of top internet porn legend Aria Giovanni, hello 64-bit Windows 7 Pro, 16gb, and full-screen 1080p videos of top internet porn legend Ana Foxxx, or Foxx, I'm not sure how many X's she has in her name.

Back in 2011 it had an Intel i5-2500K, which was an excellent CPU that remained relevant for years afterwards. At the time AMD was going through a bad patch and Intel was top dog. I replaced it in 2020 with an Intel Xeon E3-1275 V2, so I could eke a little bit more life out of the machine. As of today it runs Microsoft Flight Simulator 2020 fast enough at 1080 that it's fun to play.

Foxxx. Three Xs.

During the 2010s I replaced the hard drive with an SSD, upgraded the graphics card to a middle-of-the-road Geforce GTX 1650, added some black tape to cover up the lack of blanking plates, and smeared some paint on the case because I had some spare paint.

I also upgraded Windows 7 to Windows 10, which was free. I had nothing against Windows 8 - it ran perfectly fine on my ThinkPad X61 - but it didn't make a lot of sense in a desktop machine. The result is a PC that would have been state of the art in 2013, roughly on the same level of performance as one of the cylindrical Mac Pro models.

It's still usable nowadays, but there's a problem. The 2011-era motherboard doesn't have the necessary security chip to run Windows 11, so the machine is forever stuck with Windows 10, and on 14 October 2025 Microsoft stopped supporting Windows 10:

 

It feels strange, downloading the last update of an operating system. I have a couple of old PowerPC Macintoshes that are forever stuck with OS X 10.5, and occasionally I power them up and check to see if there are any updates. There are none, although apparently there is a way to run an early build of OS X 10.6 on them.

 

There was a time when a fourteen-year-old PC would have been obsolete junk. The original IBM PC would have been fourteen years old in 1995. There may have been a few machines still running Lotus 123 and WordStar in offices here and there, but Duke Nukem 3D or Quake would have been beyond them. In the age of Unreal and Half-Life the 286, 386, and 486 would have been similarly adrift.

A fast original Pentium machine from 1995 would have been far behind the curve in 2009, which was the age of Call of Duty: Modern Warfare 2 and high-def video. But the pace of change slowed down in the multi-core age. A PC built in the early 2010s isn't all that old-fashioned. My fourteen-year-old PC is still perfectly capable of running modern games at modest settings. I reviewed The Talos Principle 2 and Microsoft Flight Simulator 2020 on this very machine, and had fun while doing so. I even started to review The Outer Worlds, which came out in 2019, but mid-way through I switched to my new PC.

 

Back in 2011 I chose 64-bit Windows 7 Pro because the Pro version had XP Mode, which was a virtual machine that ran XP. I never used it, but it might have been handy. I upgraded in place to Windows 10 Pro when it came out in 2015. Microsoft originally announced that Windows 10 would be the last version of Windows, and that in future they would continually update it, along the lines of Apple's MacOS. But the company obviously changed its mind, because in 2021 it released Windows 11, and in 2023 it announced the end of support for Windows 10 as of October 2025.

 

Which was awkward, because as of this writing - mid-October 2025 - Windows 10 is still installed on around 45% of PCs running Windows. Back when Windows XP was a thing Microsoft had to push back the retirement date because it was so popular - netbooks gave it a renewed lease of life - but this time the company has held up its hand and said "enough".

Ordinarily I would upgrade, but my Asrock H67M-GE motherboard is just too old. With no way to install Windows 11 my PC has suddenly reached the end. It still works, and will continue to work, but Windows 10 is the end.

Windows 7 Pro cost me £162 in 2011, and Windows 10 was a free upgrade, so for £11 a year I got to play XCOM: Enemy Within as much as I liked, which I suppose was worth it, but the forced obsolescence is still annoying, and yes I realise that Enemy Within was actually released in 2013, but humour me.

What are the options? Amazon has a bunch of tiny little mini PCs for around £160 that run Windows 11. They're based around the Intel N95 low-power CPU. They have roughly the same CPU performance as my aging, power-hungry Intel Xeon, which is humbling, but they're reliant on the N95's built-in graphics processor, which is less powerful than the discrete card in my PC. It strikes me that a mini PC would be a dead end. It'll be junk in fourteen years.

I could of course switch entirely to a Mac Mini, and the new M4 Minis are very tempting, but it would be nice to play XCOM: Enemy Within every now and again. So, new PC.

Intel or AMD? Over the years the two companies have tussled. The first PC I built myself had an Intel Celeron 300A, which could be overclocked to run as fast as a 450mhz Pentium II by putting sellotape on one of the pins. It was sweet! I used it throughout the Pentium II and III eras, but replaced it with an AMD Duron in the early 2000s, because the Pentium 4 was a poor deal. The Pentium 4 relied on expensive, proprietary RAMBUS memory and was less efficient than AMD's designs.

AMD was hot in the early 2000s. The modern 64-bit version of the x86 architecture was AMD's doing. For a while it seemed that AMD was The Future. But Intel came back in the mid-2000s with the Pentium M and Core Duo, and my next machine was an Intel Core II Duo, I forget the model. I upgraded it towards the end of its life. I replaced it with the i5-2500k mentioned above. AMD went through a terrible patch in the 2010s with its Bulldozer architecture, which was inefficient and power-hungry, but over time Intel lost its mojo and AMD rose again in 2011 with Ryzen.

As of 2025 AMD is the internet favourite, and their chips tend to be slightly better value, but ironically their popularity is such that I couldn't find a good motherboard at a cheap price, so I picked Intel.

 

The general process of building a PC hasn't changed much since 2011. The main difference is the normalisation of SSDs. In 2011 they were exotic. Then they were super-fast boot drives. Then M2 came along and SSDs became tiny. Now it's entirely feasible to have an all-SSD system. Otherwise SATA, PCI, the general concept of RAM sticks, the CPU mounts, are all at least conceptually the same. I settled on the following system:

 

- Intel i5-12600K CPU

- MSI H610M-E motherboard

- Crucial P310 M2 1TB SSD

- Corsair RM750x PSU, which is ludicrously overspecified but was cheap

- 16gb Silicon Power DD4

- Cooler Master Hyper 411 heatsink

- iONZ KZ19 Spin case

This came to a total cost of £461.40. Looking back through my old emails, that's about £50 cheaper than the i5-2500K machine I built in 2011. The new machine has twice the memory, four times the CPU power, and an SSD that has the same capacity but is much faster.

 

Here's the case:

It has a bunch of built-in fans. Thankfully no coloured lights. My plan was to set up the machine with the CPU's integral graphics, and then think about a graphics card later. I like a big empty case with lots of airflow. Could I have saved a bit of money? I could have reused my existing Corsair RM550x PSU (-£109.99), picked a cheaper CPU (-£50 or so), and settled for a 500GB SSD (-£10) for around £290 in total. That's not much more than a pre-built mini PC.

 

Assembling the components was easy, although I hit a stumbling block early on with the M2 SSD. M2 drives have to be screwed into place, otherwise they flip up, but it took me an age to find the tiny tiny standoff screw. After that the process was M2 SSD, RAM sticks, CPU, CPU cooler in that order.

 

Then there was the rigmarole of fitting this into the case, and connecting up the power and reset switches. In the following image the back of the PC is towards the right, and the PSU fits into the basement section, sucking air in through the back and ejecting it downwards onto the table.

 

 

Physical hard drives also go in the basement, but on the left. The case has screw mounts for SSDs underneath the motherboard, either screwed into the back of the machine, or mounted in a little bracket, viz:

 

It wasn't long before it was all assembled. The same case is also available with a wraparound mesh cover. In retrospect I have no idea why I picked glass. Was it slightly cheaper? I can't remember.

 

Then, the moment of truth. I turned it on, then held down DEL to enter the bios and check out the temperatures, which were nice and low:

 

Bear in mind that my regular PC is a 2012 Mac Mini, which runs normally at around 50 degrees centigrade, up to 60-70 when playing video.

 

I then installed Ubuntu, just to see if it all worked, and also to check out the temperatures while running, which again were fine:

 

There was one curious thing, though. It froze. As in, completely locked up, out of the blue. Frozen mouse cursor, no response. Linux doesn't usually crash that badly.

 

At this point I'll share something with you. There appears to be a CPU shortage, so I picked a used i5-12600K from Amazon's warehouse. CPUs don't break, do they? They're sealed up. In all the years I've been tinkering with PCs, the only outright hardware failures I've had were a hard drive and a motherboard. Ubuntu freezing up must have been a one-off glitch.

 

With the PC assembled I bought a physical copy of Windows 11. On the one hand Microsoft's decision to give up on Windows 10 is annoying, but on the other hand I like Microsoft Flight Simulator, so there is that. Microsoft has treated me badly, but perhaps I can change Microsoft.

 

 

Windows 11 Professional has a bunch of enterprise management tools but otherwise doesn't have a compelling raison d'etre for a home user, so I picked Windows 10 Home. It comes on a USB stick, which is fair enough given that my new PC doesn't even have an optical drive, but no amount of packaging will ever make USB sticks feel like a premium product. There is apparently an DVD version, but it's only available for companies, not individuals.

 

Installation was a breeze - until it wasn't:

 

To my surprise Windows bluescreened during the installation process. Then recovered, then bluescreened again. Each time with a different error message. IRQL NOT LESS OR EQUAL. Something about memory management. Something about the kernel. Something about drivers. With persistence I managed to install Windows 11, but connecting to the internet and downloading the service pack caused it to bluescreen again and again. I could reach the desktop, but it was flaky.

 

There followed a frustrating weekend. I swapped PSUs, to no avail. It appeared that both PSUs were fine. I ran Memtest86, but the RAM passed. The SSD appeared to be a-ok. I tried installing Windows 11 with a discrete graphics card, but it still froze. I installed Windows 10 and got as far as playing XCOM: Enemy Within, but after a while it froze. I updated the motherboard firmware and BIOS, then tried running the memory at a lower clock speed, but it froze, each time with a new bluescreen code. I downloaded the Windows 11 installation media and tried installing it that way, but no dice.

 

The problem must have been the motherboard, so I returned it for a replacement. Two days later I reassembled the PC, installed Windows 11... and it froze again. What jiggery-pokery is this? Was something touching the side of the case and shorting out? Did I have fake RAM chips that passed the tests but only had half the capacity?

 

Could it be the CPU? The CPU? Et tu, CPU? Really? Some of Intel's more recent CPUs have a bug that causes them to draw too much power, which slowly damages them, but the i5-12600K seems to be unproblematic. Nonetheless after eliminating everything else I was perilously close to blaming it on bad electricity or bad air, so I returned the CPU and bought a considerably less capable i3-12100F instead.

 

The i3-12100 has four cores vs ten in the 12600, and it doesn't have a built-in graphics chip. It's still over twice as powerful as my Xeon 1275 and uses less power, but it's a definite step down from a 12600K. It will however play XCOM: Enemy Within at max settings.

 

After reassembling everything again the i3-12100F chugged into life. Windows 11 installed, didn't crash, it installed the updates, it continued to not crash. I downloaded Firefox, Steam, Speedfan, iCloud, a bunch of other stuff, and Windows continued to work.

 

Activating Windows 11 on this new machine was annoying, because activation was tied to the previous motherboard, which I had returned, but after entering a huge string of numbers into my mobile phone's keypad I managed to sort that out. For some reason Windows 11's "activate by phone" menu doesn't have the United Kingdom. It stops at Palau. For the record the correct telephone number is 0800 0188 354, as of October 2025, and you need to select one of the countries in the drop-down menu in order to show your device ID.

 

The proof of the pudding is in the eating, so I installed The Outer Worlds and played it for a bit, then I wrote this blog post, and Windows 11 continued to work. I'll write about Windows 11 separately. It has a curious mixture of pointless-but-hip features that put me in mind of Active Desktop and push technology, remember that? From Windows 95? Push technology? You could have news feeds on the desktop. It didn't take off in the 1990s, but the industry has now decided to make it happen again. Also, why in the name of living heck does Notepad have embedded AI? It's a text editor.

 

Some of the menus date back to Windows 95. Technically the following menu is the Screen Saver dialogue, but don't tell me they didn't just take Windows 95's display settings screen and tweak it a bit:

 

Oh, the fonts are new, but I recognise Windows 95 when I see it. Right-click a hard drive, select "properties", and look at all those tabs. Two layers of them. That's Windows 95 all right.

 

XCOM: Enemy Within, running on Windows 11 on my new PC. 

 

It's ironic. When Apple released OS X fans of Windows laughed at the trendy Aqua interface, with its buttons that looked like droplets of water. But modern MacOS is visually clean, and functionally almost indistinguishable from mid-2000s OS X 10.4, which is great, because it gets out of the way. Meanwhile Windows is still messy, with traces of Windows 95 lurking here and there, and features spread across different control panels. The mouse speed is controlled with one panel, but the mouse cursor is controlled with another, and fixing some problems requires delving into the Windows 95-era, perhaps even Windows 3.1-era device manager.

 

What year will it be, fourteen years from now? No-one knows. In 2011 smartphones were here to stay, the iPad was a year old, and it seemed that the x86 architecture and Windows were safe for the foreseeable future. However fourteen years later Apple has successfully demonstrated that ARM will work on the desktop, Intel is in a terminal death spiral, Windows itself seems to be drifting ARM-wards, laptops are mostly on a par with desktop machines, held back only by weak graphics, and it's entirely possible that computers fourteen years hence will be dumb terminals streaming everything from the internet.

 

I'm reasonably confident that my PC will still be relevant - if Microsoft drops x86 support tomorrow it'll still take several years for it to fade away - and of course my PC is ripe for upgrades, because it has an i3. In the next post, let's see how Microsoft Flight Simulator 2020 works on this new machine. There's a sequel, Microsoft Flight Simulator 2024, but I'm not made of money. Let's try Microsoft Flight Simulator 2020. If the machine can survive that, it'll survive anything.

Every Jetliner Ever Made: Embraer

Let's have a look at Every Jetliner Ever Made, in alphabetical order, but with Airbus and Boeing at the end because the list would be incredibly front-loaded otherwise. Today we're going to look at Embraer, of Brazil, where we stood beneath an amber moon, and softy murmured "some day soon".

Embraer (Brazil)

Up until the mid 1960s Brazil's aviation industry relied on foreign designs, such as the Savoia-Marchetti S55 flying boat and the Lockheed Electra pictured on the stamps at the top of the page. The Brazilian government decided it was high time to kickstart the country's domestic aviation industry, so it issued a specification for a light turboprop transport with a view to military and commercial sales.

A team led by French import Max Holste came up with a sleek, low-winged design, which was eventually called the Bandeirante, or Pioneer. It was named after the 16th- and 17th-century bandeirantes, who had ventured into the Brazilian interior in order to peacefully persuade the local inhabitants to work on farms and worship the correct God. The bandeirante are controversial nowadays and I suspect that if the aircraft had been designed in the modern age it would have a different name.

The aircraft's design process was similar to the Soviet method, whereby the prototype was constructed by Brazil's leading technical bureau, with a plan for a separate company to carry out series production. However the private sector failed to step up, and so a chap called Ozires Silva was asked to set up a state-owned aviation manufacturer, which became the Empresa Brasileira de Aeronáutica, or Embraer for short.

The resulting EMB 110 Bandeirante was a sleek, low-wing turboprop, not a million miles from the Handley-Page Jetstream or the Fairchild Metroliner. It went on to sell around five hundred units internationally. It was the very first aircraft of Irish low-cost airline Ryanair, who flew it from Waterford in Ireland to London Gatwick. Embraer continued to have success with turboprops over the coming decades, but in 1989 the company decided to take things up with a notch with a regional jet, the Embraer Regional Jet (ERJ).

Initially the design was very conservative. It was essentially a stretched derivative of the EMB 110, retaining the straight wings and T-tail, but with podded jet engines instead of turboprops. The result was aerodynamically fussy, so Embraer gave it rear-mounted engines instead and a gently swept wing, which ended up mirroring the configuration of the Bombardier CRJ and the Douglas DC-9. The complicated development meant that, although the ERJ had been announced at roughly the same time as the Bombardier CRJ, it entered service four years later, in 1996.

An ERJ 145

The ERJ was offered in three basic models, the ERJ 135, ERJ 140, and ERJ 145, which carried 37-50 passengers depending on the model. Embraer also built long-range variants with larger fuel tanks. The 50-seat ERJ145 accounted for the majority of sales. Despite the delay in production it managed to keep pace with the CRJ, sales-wise.

Partially this was the result of great timing. In 1992, when the CRJ entered service, smaller airliners were still wary of regional jets, but by the mid-90s turboprops were increasing perceived as being old-hat, and passengers routinely expressed a preference for jets.

The launch airline, Continental Express, was typical of ERJ's customers. In the mid-90s Continental had a fleet of aging turboprops of different types that had been accumulated over the years. Rather than replace them like-for-like the airline decided to switch to regional jets instead, with an order for 200 ERJs. Over time the ERJ also developed an excellent safety record, with ten crashes but no fatalities to date, which is impressive for an airliner designed to make a lot of short flights. ERJs are still routinely used nowadays, with production only ending with the COVID pandemic in 2020.

The ERJ (right) was visually similar to the Bombardier CRJ (left), but with a sleeker nose and different engines. Bombardier used the General Electric CF34 while Embraer used the Allison / Rolls-Royce AE 3007.

As is the fashion for regional jet manufacturers, Embraer decided to follow up the success of the ERJ by chasing a slightly larger segment. In 2002 the company launched its next airliner, the E-Jet, which was closer in size and specification to the hugely popular Airbus A320 and Boeing 737, although around two-thirds the size. The original plan was to stretch and re-engineer the ERJ, but perhaps mindful of the delays incurred during the early development of that aircraft Embraer decided on a clean-sheet design instead.

An E-190

The E-Jet has the same configuration as a full-sized airliner, with a pair of engines under the wings and a conventional tail. It was at the time something of a gamble. Bombardier had toyed with the idea of a miniature full-sized jet in the late 1990s, but decided against it, which meant that the E-Jet was unique when it entered service in 2004. The most popular model turned out to be the Bombardier E-175, which carries 78 passengers, with other models - the stretched E-190 and E-195, and the original E-170 - carrying anything from around 60-100 passengers depending on configuration. Embraer also sold freighter conversions of the E-190.

The E-Jet benefited from a peculiar, US-specific law whereby airlines that have a regional subsidiary are legally obliged to use smaller airliners on that subsidiary, topping out at 76 seats, which was a perfect fit for the E-175. For that reason there was even a specific submodel of the E-175 that eliminated two seats to bring it to the 76-seat maximum.

On the whole the E-Jet benefited from a general inflationary trend in the airliner market, whereby former short-range jets such as the A-320 and 737 have been stretched and re-engined to a point where they can cross the Atlantic, while smaller regional jets have been redeployed as workhorses on trans-European routes.

Another E-190

Production of the original E-Jet continues as of this writing. In 2019 Embraer embarked on talks with Boeing for a possible deal whereby the two companies would team up to make a second-generation E-Jet. The proposal mirrored a similar deal brokered between Airbus and Bombardier that resulted in Bombardier C-Series becoming the Airbus A-220.

However a combination of COVID and Boeing's troubles with a new model of 737 eventually caused the deal to fall apart, at which point Embraer continued as an independent entity. Bombardier, meanwhile, sold its C-Series jet to Airbus, and left the passenger aviation market to concentrate on business jets.

An E-195, which is visually almost identical to the E-190, but eight feet longer and with two more rows of windows. The E2 model has two overwing exit doors instead of just one.

The reason for the talks was the E-Jet E2, which was Embraer's successor to the E-Jet. It was announced in the 2010s and entered service in 2018. The E2 has a similar two-engines-under-the-wing configuration, but with improved aerodynamics and more efficient powerplants. The two models currently in service, the E-190E2 and E-195E2, carry 96 and 120 passengers respectively. Such has been the inflation in aircraft capability over the years that the E2 is actually larger and carries more passengers than the original Douglas DC-9 and Boeing 737, which was why Boeing was so keen on having something that could slot into that section of the market.

Unfortunately for Embraer sales were sluggish, not helped by the COVID pandemic. As of this writing Embraer has sold around one hundred and fifty E2s, much less than the more capacious Airbus A220, but there are around 350 more on the order books, and production is still ongoing. It has yet to land a big order with one of the more glamorous airlines, and only time will tell if it was a stretch too far.

And that's Embraer. Next, Fokker. And VFW-Fokker. And Fairchild. And Fairchild-Dornier. Or possibly just the first two, and then the next two. Or maybe all of them.

Novation Circuit Mono Station: Fat Bread

Let's have a look at the Novation Circuit Mono Station. It looks like a nightclub, doesn't it? There's a dancefloor with glowing floor panels, and a DJ's mixing desk. But it's not a nightclub. Oh no. It's actually a surprisingly fully-featured synthesiser with a built-in pad-based sequencer.

Just look at all those colours:

Novation released the Mono Station in 2017 as part of their Circuit range of groovebox sequencers. Sound on Sound liked it, but while the other Circuits were updated or replaced circa 2020 the Mono Station was quietly discontinued. I think the problem is that it was marketed as an all-in-one analogue sequencer workstation, but it's really just a synthesiser with an admittedly smart sequencer, a kind of super-sized Korg Volca. Unless you do some clever programming it can only play a single bassline with no drums.

At heart the Mono Station is a Novation Bass Station 2 synthesiser combined with a Novation Circuit sequencer. You can in theory just use it as a Bass Station 2, ignoring the sequencer entirely, in which case it's actually cheaper on the used market.

Let's talk about the bits. The original Novation Bass Station came out in 1993. It was a two-oscillator monosynth housed in a cute little case, with a two-octave keyboard. At the time it was sold as a clone of the Roland TB-303 acid house bassline synthesiser, but it was much more versatile. It could do analogue lead lines as well. For a long time it was the only affordable analogue synthesiser on the market and it sold like hot cakes.

Novation upgraded the engine in the following years. The Super Bass Station rackmount of 1997 added a sub-oscillator, ring modulation, white noise, and a distortion effect, which meant that it could also double as an analogue percussion generator. For the next few years Novation concentrated on the Nova digital synth, but in 2013 they launched the Bass Station 2, which added a bunch of upgrades including oscillator sync and (with a firmware upgrade) duophony, essentially turning the Bass Station into a modern analogue of the ARP Odyssey or Sequential Pro-One.

The Mono Station's synthesiser engine is a Bass Station 2 minus one of the LFOs, minus one of the envelopes, minus the TB-303 filter emulation, minus oscillator cross-modulation, but what remains is still pretty versatile.

It has two mixable oscillators, sine-triangle-saw-square, with variable pulse wide on the square, although you can't set this manually. Pulse width has to be modulated. There's a slightly underwhelming sub-oscillator, plus white noise and an audio input. The audio input goes through the envelopes and the filter.

It also has ring modulation and oscillator sync. The Mono Station's manual does a poor job of explaining how to make oscillator sync sound good - it works best if you apply a bit of pitch modulation to the sync source. At that point the sync effect makes the distinctive beoooo sound.

There's a single, multi-envelope, sync-able, key-syncable LFO that can be assigned to pitch, pulse width, filter cutoff, and the distortion effect. The distortion has three types, all of which have a digital-sounding crackly crunch. There's also a relatively subtle overdrive effect. As pictured above the Mono Station uses glowing LED lights to show the depth of a modulation source.

The six-part, button-driven modulation matrix is easy enough to use, although it would be nice if there was a way to quickly zero out the modulation. Sometimes I found that my patch sounded odd, at which point I had to click through each of the six modulation destinations to make sure that they weren't interfering with the sound. The modulation depth knob doesn't have detents, so you have to be really precise to zero out the modulation.

The Mono Station has one envelope and one LFO, which is reminiscent of the original Roland SH-101. Sadly this means that you can't combine a slow-running filter sweep with fast-moving pulse width modulation, for example. But it's work-around-able.

There are two filters. The 12dB filter has a bright sound. The 24dB filter is darker-sounding, more Moogy, very powerful. I've always had a penchant for Doctor Who-style chunk-a-chunk bass sounds, and the 24dB filter is good at that.

Unusually the two oscillators are independently addressable. Ordinarily the Mono Station acts as a two-oscillator mono synth, but if there are any notes in the second sequencer channel they're assigned to the second oscillator. This also works with incoming MIDI information, with the two oscillators using two selectable MIDI channels. By default the Mono Station's single envelope is only triggered by the first sequencer channel, so oscillator two only appears in the shadow cast by oscillator one, but there is an option to trigger the envelope from both channels.

As a MIDI synth module the Circuit Mono has one problem that annoyed me. It transposes incoming MIDI notes to the selected scale. It puzzled me for ages until I realised I needed to set the scale to chromatic if I wanted to address every note. A subsidiary issue is that the scales aren't labelled. It would have been nice if Novation had simply cut the range down to eight scales - there are sixteen scales, which feels excessive - and labelled them underneath the lower row of pads.

The clock output is compatible with the sync ports on the Korg Volcas. The CV/gate output is tied to sequencer channel one.

Which brings me to the sequencer. There are three channels. Channel one controls oscillators one and two. The second channel controls oscillator two. And there's a third channel that contains modulation data, which is useful if you want to layer a pattern of accents on top of several other sequences. A later firmware update turn this into a fully-fledged third sequencer channel that can drive external gear from the Aux CV output.

The sequencer has some clever ideas, but also some limitations. Each channel can have an independent pattern chain. You can set channel one to loop after four patterns while channel two loops through three patterns, and the patterns themselves can be shortened from sixteen steps. Chaining them together is a simple matter of holding down the first step and then pressing the other steps in sequence, but as with the Korg Volcas you can only step through the patterns in order. You can't program a song that goes 1-1-1-2-1-2-3-4, for example.

The process of programming the actual notes reminds me of the Arturia BeatStep Pro, but it's clunkier. With the BeatStep, you can cue up a sequence and then selectively mute notes while the sequence plays, or add notes transparently, which is great for live sequencing. On paper you can also do that with the Circuit Mono, but in practice pressing the keyboard immediately triggers a note, which interrupts the sequence. There doesn't seem to be a way to cue up notes silently, or at least if there is I couldn't find it in the manual.

A second issue is that the current view is tied to the playing pattern. If you want to edit pattern 1, and you're playing a chain of patterns that goes 1-2-3-4, you have to wait for the Circuit Mono to loop back to pattern 1 before you can edit it, and you have to implement your edit before it moves to pattern 2. In general I find the BeatStep superior as a live sequencer.

The Mono Station will synchronise up with the BeatStep Pro via clock sync, and vice-versa.

The sequencer has one major limitation. The two melody channels are hardwired to oscillators 1+2 and 2 respectively, as if Novation really wanted to interest you in paraphony. In practice this means that anything you put in sequencer channel 2 triggers the second oscillator. It would have been nice if channel 2 could have been used purely to drive external gear. The aforementioned firmware update essentially does this with the modulation channel, which is nice, and it's less of a limitation if you just use the Mono Station as a clock source, but it's a problem if you intend to use the device as the heart of an analogue studio.

Now, this is more of a feature request than a limitation, but it would have been great if the Mono Station simply had four individually assignable sequencer channels that could be set to drive any combination of oscillators or external gear. I have the impression the designers started with the Novation Circuit's two-polyphonic-synthesisers-plus-drums arrangement and simply transposed it to the Mono Station, but it doesn't make as much sense on a monophonic synth that hasn't got a separate percussion engine.

Are there any other limitations? The Bass Station 2 has an arpeggiator. The Mono Station doesn't. You can in theory program a pattern as an arpeggio, but there's no way to transpose it, either with the keyboard or with incoming note data.

The Mono Station will also synchronise via MIDI clock, although this requires a chain of 3.5mm-to-MIDI-to-3.5mm adapters. The Mono Station and BeatStep Pro both use the same 3.5mm-to-MIDI protocol, so in theory a direct stereo 3.5mm cable connection should work. I tried this, and it worked, but it's apparently bad form.

This is why I think of the Circuit Mono as a Bass Station 2 with a built-in sequencer rather than as a fully-fledged groovebox. The original Novation Circuit had rudimentary reverb and delay effects, and a drum channel. The Circuit Mono Station doesn't have that, so if you plan to use it as a portable ideas notepad you'll probably get sick of sixteen-step basslines, at which point it becomes very limited. A simple Tangerine Dream-style delay would have been nice.

If you plan to take it on a journey with you the Mono Station is roughly the same size as two Korg Volcas. It'll synchronise with them using simple mono Eurorack patch cables, but you would of course need a small mixer as well. The Mono Station has an audio input, but it goes through the filter and envelope, which can't be bypassed. You can of course pair it with a MacBook Air, but you'd need some kind of audio interface, because the Mono Station's USB connection only carries MIDI. Of note the Mono Station relies on a 12v mains adapter for power.

How would I have implemented the Mono Station? I would have taken one of two directions. If money was tight I would have dropped the paraphonic, duophonic aspect, dropped the ring modulator, added a delay effect, and also added four individual, assignable sequencer channels, with channel one optionally driving the internal synthesiser and the others driving either the CV output or a selectable MIDI channel. If there was room for a simple drum synth on the unit, or even just a bunch of drum samples, I would have thrown that in as a bonus.

Or, if the economics were viable, I would have given the Mono Station four individual, assignable sequencer channels, and four separate single-oscillator synthesiser engines, one per channel, that could be combined in different ways. This would have turned the Mono Station into a Vermona Perfourmer with a built-in four-channel sequencer. The end result would probably have been very expensive, but I can't think of anything else like it on the market. The end result would have been Depeche Mode's Speak and Spell in a box.

It also synchronises with the original BeatStep, although you have to remember that the original BeatStep had its own MIDI adapter.

So, that's the Circuit Mono Station. On the positive side, it's a cheap way to get hold of the guts of a Novation Bass Station 2 in a compact case, and the sequencer is handy. On the negative side the sequencer has some major limitations, but on the positive side again it's much easier to use than the sequencers built into Korg's Volca modules. If you think of it as a kind of super-Volca it makes a lot more sense than it does as the sequencing heart of an analogue studio.

Korg iM1: A Slickness in the Sky

There was a famous slogan in the punk era. "This is a chord. This is another. This is a third. Now form a band."

Let's fast-forward a decade to the late 1980s. How could we update that slogan? "This is a Korg M1. This is a copy of Zero-G's Datafile One. This is a second-hand Akai S950 sampler. Now come up with a band name, but you don't need to actually form a band, just come up with a name. You will be the band. You, and samples."

You might also need an ADAT recorder and an Alesis Quadraverb. And possibly an Atari ST with a cracked copy of Cubase. Maybe a little mixing desk to tie it all up. And some old records, although Datafile One covers that. This being 1989 all of the aforementioned would have cost you several thousand pounds, but it was all you needed to make an actual chart hit. To be an actual, bona fide pop star. To appear on Top of the Pops, standing awkwardly behind a keyboard while dancers did their thing in front of you.

You might also have wanted to hire a really good singer. And a model, who would appear in the video, but not on the record. You would have one, maybe two top ten singles.

Then it would all go wrong. Your first album would reach #14 in the charts on the back of the singles. Your second album would spend a week at number sixty. The record label would void your contract. And now it is 1993, and tastes have changed. You're baffled by the new sounds of trip-hop and jungle and big beat. You can't catch up, but it was fun while it lasted.

iVCS3 running on an iPad Mini

But let's talk about the Korg M1. It was launched in 1988, but it took a few years to become ironic, and then it had a second wind. Now, whatever your opinion of Apple, you have to admit that the company has a wide range of interesting musical apps for the iPad and iPhone. Near the top of the tree is Apesoft's iVCS3 (pictured above). It's a software recreation of the impenetrable-but-fascinating EMS VCS3. There's also Moog's Model D, which recreates the Moog MiniMoog, and Olympia's Patterning, a drum machine with a novel circular interface. And iM1, a modern interpretation of the Korg M1.

Modern-ish, because it was launched in 2015, but Korg continues to update it, and it works fine with modern versions of iOS. As of 2025 it sells for £29.99, but it's occasionally on offer. For that money you get a simulation of most of a Korg M1, with the original presets, plus some more presets from the expanded M1EX, and an in-app option to buy more presets from a bunch of voice cards.

Which isn't a particularly appealing option, because the key to the M1's success was the iconic range of sounds that came with the original keyboard. Extra sounds are nice, but do you really want a batch of b-list sounds from a 1990s digital synth? Neither do I.

What is the Korg M1? It's a sample-based synthesiser with a built-in sequencer and multi-effects. For many years it was the best-selling synthesiser of all time. Why was it so popular? There were essentially two reasons.

Firstly, it was a miniature studio in keyboard form. It had an eight-track sequencer that could play eight separate instruments at once, including drums. The architecture was built on a pool of samples, including guitars, electric basses, pianos etc, which meant that it was one of the first synthesisers that didn't necessarily sound electronic. It could do rock or orchestral arrangements, not just techno.

It also had a built-in multi-effects unit with a mixture of reverbs and delays - standard stuff at the time - but also distortion, EQ, an exciter, a phaser, even a rotary speaker simulator. As a result the M1's presets sounded unusually slick, as if they were part of a finished record. Korg's preset designers didn't just splash the effects onto the sounds arbitrarily, they knew how to make the built-in samples sound good.

The second and most important reason was the M1's range of preset sounds. As a synthesiser - as a means of generating new tones - the M1 was very limited. Nowadays it's often called a ROMPler, because it was really just a sample playback unit with a bunch of waveforms stored in a built-in ROM chip. The synthesis engine had a bank of four megabytes of 16-bit, 44khz samples that could be layered and fed into the multi-effects unit. There was a simple non-resonant digital filter, although it was more of a muffler than a proper filter. As if to compensate for its stiffness the envelopes were unusually complex, partially to enable a simple form of wave sequencing and partially to disguise the lack of real-time control.

But very few people minded the limited synth engine, because the preset sounds were fantastic. For the first time, a relatively affordable keyboard synthesiser had lengthy, professionally-recorded 16-bit sampled presets, instead of compressed little sound snippets. The traditional instruments were impressive enough to use as stand-ins for their real-life cousins, while the sci-fi sounds were perfect for ambient techno. As far as I can tell literally the first sound on The Orb's "Back Side of the Moon" is a Korg M1, specially the Universe preset:

Unfortunately the M1's sounds quickly became overused. I don't know exactly which instruments Livin' Joy used to make "Dreamer", but to my untutored ears the whole track sounds like Korg M1 with some drum samples layered on top:

Beyond the house piano and heavenly choir the M1's other popular preset was a boop-boop, boopy-boopy organ sound. It was bassy, but it also had enough treble that it didn't get buried by the drums. It was used prominently on The Nightcrawlers' "Push the Feeling On", which is another record that seems to have been created entirely with a Korg M1 and drum samples. The M1 also coincided with the Nintendo SNES, which used sample playback for its music, so a lot a SNES games had M1 samples because the composer owned an M1.

Apropos of nothing here's a recording of Japanese ambient radio station St. GIGA, which might not have any M1 on it as far as I can tell, but it reminds me of the era (nb there's a really good track at the 24:30 mark):

On the whole the M1's mixture of clean samples and digital reverb have a slickness that got old quickly. If music was a big circle, the Korg M1 would be on one side, and Johnny Cash's American Recordings would be on the other. Even in the field of dance music it dated badly. But some of the synth strings and pads have a timeless quality, and it has a lot more character than the General MIDI keyboards that followed it.

But what of iM1? It doesn't have the sequencer, presumably taken out because there are better options. I admit I haven't tried sequencing iM1 with my iPad. Instead I used Logic running on a Macintosh. The iPad has terrific integration with a Mac. It acts as both an instrument and a digital audio input, so I don't have to run a cable from the headphone jack to my audio interface. When hooked up in the fashion iM1 essentially operates as a virtual instrument within Logic, but running on an iPad.

As far as I can tell it's 44khz, stereo only, but so was the original M1. What if you don't have Logic, or a Macintosh? That's a good question. Technically it's two questions. Two good questions.

As with the Korg Volca FM there's a certain pointlessness to iM1. Korg also sells a VST version that can run directly in a sequencer, although at $49.99 it's not an impulse buy. For the record I paid £12.99 for the iPad version, which is slightly more awkward to use than the VST version and doesn't save any money if you don't already have an iPad. Compared to an actual M1, however, iM1 has a much nicer interface, and real-time parameter control. It has limited support for automation.

As a proof of concept I recorded the following piece of music using the iM1, plus Nils Schneider's free VST recreation of the Kawai K1. I was going for an early-90s SNES soundtrack / synthesiser demo song feel.

The Kawai K1 was a blend of M1 and Korg Wavestation. Each patch could be created from four samples layered on top of each other, mixed with a joystick in real time. The samples were 8-bit and very muffled, and there was no filter at all, and some things still baffle me. There's an LFO, but seemingly no way to assign it to anything. A complex modulation section, but no way to assign the envelopes to anything except volume, which is a shame because the amplitude modulation feature would have benefited from pitch modulation of the amplitude source. But it's free, so I shouldn't grumble.

While playing with iM1 it struck me that if you didn't grow up in the 1990s its sounds probably don't come across as cheesy and dated. And for an early ROMPler the recording quality of the samples is surprisingly good, so in isolation the M1 doesn't sound all that old-fashioned today. And perhaps you do want to evoke the sounds of Culture Beat or Whigfield.

Anything else? As with the original keyboard iM1 is eight-part multitimbral, but it only supports a single stereo output. This is one thing the original M1 has over iM1, because the original M1 had four separate audio outputs. It has limited support for automation, which is undocumented in the manual, but it will respond to MIDI control codes. The M1 was 16-voice polyphonic with dual-sample single patches, whereas iM1 raises this to 64 voices. On my first-generation iPad Pro it never slowed down, but then again the application is quite old and is at heart only playing back a bunch of samples.

As mentioned up the page Korg sells extra sounds as in-app DLC. The two expansions are £2.99 each. I'm sure there are some gems, but a large part of iM1's appeal is the M1's ironic, original set of preset sounds. The Korg T1 was probably fantastic, but what does it mean in 2025? The M1 has meaning, that counts for a lot.

Resurrecting a MOTU 2408 with ADAT

This is MOTU 2408 MkIII, an audio interface. It's a frustrating thing. It was originally released in 2003, but despite being over twenty years old it's still pretty good. It has eight balanced 1/4" audio inputs, a bunch of ADAT, TDIF, and SPDIF ports, and it records 24-bit audio at 44, 48, 82, and 96khz.

But it's a pain to get working with modern computers. The original MOTU 2408 MkI came out in 1999, slightly before FireWire, at a time when USB was restricted to USB 1.0, so it connected up to contemporary PCs and Macintoshes with a PCI card. What's a PCI card? Ask your parents. They might remember their own parents fiddling around with PCI cards.

The 2408 MkII and MkIII had the same system. To their credit MOTU continued to update the drivers, but as of 2025 the only Apple Macintosh with PCI slots is the Mac Pro, and MOTU's drivers haven't been updated for modern Apple Silicon processors. MOTU's next interface was the 828, but that used Firewire, which is another story entirely.

It would be nice to still use the 2408. And there is hope! Even without a computer connection the 2408 still works as an audio interface. In standalone mode analogue audio that comes in through the 1/4" jacks is routed out to the ADAT, TDIF, and SPDIF ports. SPDIF folds everything down to two-channel stereo, but ADAT and TDIF transmit eight channels of audio at 44 or 48khz, or four channels at 82 and 96khz.

What would be nice is a simple dongle that could pump ADAT to a computer, preferably something cheaper than an entirely new modern audio interface with an ADAT input. There are lots of USB interfaces that output ADAT, but a dearth of interfaces with ADAT inputs.

There is one exception, the MiniDSP MCHStreamer, which is available either as an unclad circuit board or in a little box. I'm not a farmer, so I bought the version that comes in a box. This is what it looks like:

NB I bought it with my own money and have no commercial relationship with MiniDSP. It shipped from Hong Kong and is, as far as I can tell, exempt from UK customs duty, although even if it isn't the price isn't onerous. The boxed version is currently listed at $115, shipping $35, which is about £110. The kit version (which is assembled, it just doesn't have a case) is $10 cheaper. Is it just a Raspberry Pi or something, with special firmware? Could you make one yourself? I have no idea, and possibly, in that order.

The MCHStreamer is essentially a tiny little computer board that converts a bunch of digital audio formats in real time. The boxed version only supports ADAT and SPDIF, although I assume that's only because the internal headers are covered by the case. It's powered by USB, and it's small enough to rest on top of the 2408.

Getting it to work isn't straightforward, although it's still easy. I have a Macintosh. The first step is to download the firmware bundle and plug the MCHStreamer into the computer. At that point you have to upload the correct firmware into the box. In the following screenshot I've picked ADAT:

Step two involves plugging the included optical cable - the package includes two optical cables and a USB cable, which is a nice touch - into the 2408's BANK A optical ADAT output:

Step three involves booting up the 2408 and fiddling with the SELECT and SET buttons. For the SOURCE I picked the audio inputs. Confusingly the LEDs imply that the audio is routed to ADAT BANK C, but no, the audio is output to all three ADAT banks.

For the CLOCK I picked 44khz, internal. Baby steps. I then plugged a radio into input 3, just to see if it worked, and also because I wanted to make sure that it was transmitting on discrete channels and not just 1+2 as a stereo pair.

Then I popped open Audio MIDI Setup and set the MCHStreamer to pick up clock from the 2408's ADAT signal:

Why do you have to do this? If you don't - and I tried it - the signal is crackly. My understanding is that the optical audio protocol isn't like computer networking. It doesn't send packets of data, it just sends a string of bytes, and the receiver has no idea where each byte begins unless it has a clock signal from the source. The result is a kind of audio chaos that sounds awful. Does that sound like a plausible explanation? That's how my mind envisages it. Sometimes my mind plays tricks on me, but deep down there is a brave heart. In my mind.

Synchronising the MCHStreamer to ADAT clock was flaky. At first. I can't tell if it just takes a long time, or if there's a quasi-random factor, but at first it didn't work. After cycling through the 2408's clock options it did eventually hook up, and remained hooked up, although curiously after the first run (pictured above) it only synchronised at 48khz. And yet once it worked, it stayed working. Maybe it's a first-run thing. Perhaps I just didn't have the optical cables plugged in all the way.

After setting all this up I opened Logic, and lo and behold, MCHStreamer ADAT was an input source, with eight channels of audio:

I clicked the input monitoring icon and voila, I could hear a signal coming from the 2408, and I could select each of the eight channels and record audio from them.

As a proof of concept I used it to record the following piece of music, which has a mixture of audio tracks from my modular synthesiser and some virtual instruments:

Sadly one thing missing from this setup is the 2408's original monitoring hardware, which was housed on the PCI card and acted as a mini-mixer.

Still, if you happen to have an ancient MOTU 2408 lying around doing nothing, it can be returned to service as an eight-channel audio interface with a simple ADAT to USB box that powers itself from the USB port. Alternatively, if you want to use the 2408 as a crude mixer you could power the MCHStreamer with a powered USB hub.

In theory you could bypass all of this malarky by plugging the 2408 into the ADAT input of another audio interface. But the cheaper good-quality audio interfaces don't have ADAT ports, and the more expensive interfaces are much more capable than the 2408, so why bother?

If the MCHStreamer was $35 and just a tiny little USB dongle with USB at one end and a pair of optical ports at the other it would be even more superb, although it has to be said that the population of people who have a twenty-year-old MOTU 2408 lurking in a cupboard probably isn't large enough to justify the research and development outlay. If it had a built-in DAC and a 3.5" headphone output it would also be useful for the tiny, tiny population of people who want to use a PlayStation 3 with a computer monitor that doesn't have built-in speakers, but again that's a small constituency. There are dozens of us. Dozens.