Overclocking this Mac to the limit using scrap parts and period correct mods

Overclocking is a strange, inexact science. As you increase the overclock, you'll eventually hit instability where the system fails to POST or crashes shortly after, but sometimes if you continue to push the overclock you may hit an island of stability where it works just fine. This is true with these old machines and still applies to modern systems as well.

The 565 color format uses 6 bits for green. Our eyes are most sensitive to the green part of the spectrum, so it made sense to give green the additional color info.

On SMD resistors, '100' means 10 ohms with no (zero) multiplier. A 100 ohm resistor would read '101', or 10 ohms with a single 10x multiplier.

That is officially the first time in my life I've ever seen a computer overclocked making a visible difference in normal use. I can now die happy...

Coincidentally, I have a commercial overclocking add-on for the 610 (and similar) sitting right in front of me. 😆 It used a PLL block synthesizer and a sixteen position rotary switch to choose different clock speeds— there’s a little injection molded piece that just press-fits over the SMD crystal. Probably circa ~1996. If I recall correctly, some models of Mac (and OS) would notice the upgraded clock speeds and change their product names — e.g. Centris would become Quadra, etc.

Oh man, that old netscape load screen brings back memories!

I'm always learning new things from Adrian's videos - I had no idea about needing a managed switch for my old machines. Great stuff.

Reminds me of the ancient days of overclocking my old Power PC 60 to 80mhz with a little piggyback clock crystal that sat over the original clock crystal. Really helped with Doom 2.

@28:20 The original IBM PC/XT didn't just use that crystal for the CGA card, but as the primary clock source for the whole system. IBM was cost cutting the PC design at every corner possible, while still maintaining the sturdy business look of the case. The 14.318 clock signal is used to derive the 4.77MHz processor clock (/3), the 1.193MHz timer base clock (/12) and (as you already mentioned) the CGA pixel clock (1:1 in 640 pixel modes, /2 in 320 pixel modes). Those oscillators were produced in high volume (i.e. at low cost) already when IBM designed the PC, and by being part of the PC architecture, they got the industry standard frequency for cheap crystal oscillators later on. The original IBM PC in a standard configuration did have a second oscillator on the "monochrome display and printer adapter" (MDA/MDPA) at 16.257MHz a third one on the floppy controller (at 8MHz) and finally a fourth one on the serial interface card (1.8432 MHz), because those clocks couldn't be derived from the 14.318MHz reference. Note how all the options I mentioned were unnecessary if you used the originally intended "base configuration" of 16K RAM, no floppy drive, CGA card and a tape drive as BASIC program storage, running only ROM basic or games loaded from tape, so that kind of machine really fully ran of the one 14.318MHz clock. This configuration never hit the market (although IBM employees could get it), so I am unsure whether any IBM software was ever released on tape. The actual base config a normal consumer could get included 32K of RAM and a floppy drive. The IBM floppy boot process is incompatible with 16K machines. In PC evolution, the 14.318MHz clock stayed the "reference clock" for a very long time: Most PLL-based clock generator chips, at least up to the Pentium III generation (i.e. the end of the century) use a 14.318MHz clock input. Many clock synthesizers on Super VGA cards also use a 14.318MHz refence clock (although some VGA pixel clock synthesizers use a 40MHz reference clock, as 40MHz initially was a common SVGA memory clock when the split RAM/pixel clock designs started to appear - so a 40MHz clock needed to be provided on the SVGA chip anyhow).

The memory usage reminds me of my beloved Risc PC on RISC OS. Same limitations with the VRAM. Nice upload :)

SMD resistor marked "100" is 10 Ohm, not 100 Ohm like you might think. Marking is XXY where XX is 2-digit value and Y is the power of 10. So 100 means 10 * (10 ^ 0) = 10 Ohm.

Excellent video, loved the screen background colours too, as well as the indepth advice. A bit of a nostalgia hit from seeing the Netscape browser, and logo. I recall using video and audio calling with some Netscape app back in 1994/5 (PC). Love this Mac! good job.

I think the third digit on SMD resistors is the same as colour codes on THT ones. So a 0 for the third digit means no zeros so that was a 10 ohm resistor. I’ve made that mistake before too.

I have fond memories of this computer as I was the system and network administrator for a computer lab of this and similar models in my high school as part of an “independent study” class in computers. This included upgrading from LocalTalk networking to Ethernet to connect to the school district’s first non-dialup internet connection, a leased 56k data line. I remember using crumpets to make all of the Ethernet cables ourselves to keep the cost down.

I have an in-box Centris 610 overclocking kit from 1994 sold by KS Labs as the "Alacrity 610". It's similar to the RocketSocket pictured in your video in that it's a socket for a crystal oscillator that slips ontop of the computer's built-in oscillator. It also came with multiple crystal oscillators, a heatsink/fan, and an installation manual.

Nothing fell better than making old things works better by modding with spare or scraps parts, thank you for your video.

This was fantastic. We forget about the speed of video memory/video cards, RAM, and CPU power when we look back on the old and slow dialup days.

@Adrian, it's called AAUI, "Apple AUI" Apple commissioned Amphenol for a custom connector, and paid for all the custom tooling and ordered a bazillion of them.

If you want to know for sure if it's EDO and it's a 3.3V module and do not have a memory tester: get a ohm/signal meter and check if pin 66 is connected to VSS (aka ground, on pin 72). It's not 100% fool proof as some modules are mismatched but it's in spec for only EDO to have pin 66 connected to ground. On Fast page mode ram it should not be connected. This does NOT work for 5v modules.

With surface mounted resistors 100 = 10 ohms, 101 = 100 ohms. Separate the the first 2 digits from the third one. If third digit is a zero then drop it the resister is the value of the first 2 digits. if the third is 1 - 9 then it's 10 to the power of the third digit then that value times the first 2 digits. 104 = 100,000. the quick method is to add the number of zeros of third digit to the first two digits. If you want to put a 50 ohms resistor there use one marked 500.