68kMLA Classic Interface

This is a version of the 68kMLA forums for viewing on your favorite old mac. Visitors on modern platforms may prefer the main site.

Click here to select a new forum.
Overclock LC475/P475/Q605 without soldering or spicy-o-clock
Posted by: Joopmac on 2024-10-09 00:35:55
Does anybody dare to classify these as original or fake??
www.ebay.com

MC88916DW80 SOP-20 IC DRIVER CLK PLL 80MHZ 20-SOIC RH #A1 | eBay

Find many great new & used options and get the best deals for MC88916DW80 SOP-20 IC DRIVER CLK PLL 80MHZ 20-SOIC RH #A1 at the best online prices at eBay! Free shipping for many products!
www.ebay.com
I always am cautious from ordering from China, as a fact i don't do it at all normally but this one i cannot find anywhere
Posted by: Mustermann on 2024-10-09 03:08:36
I propose to split the MC88916 topic to this thread

Upgrade MC88920 to MC88916 for LC475/P475/Q605 (advanced soldering)

In the Overclock LC475/P475/Q605 without soldering or spicy-o-clock thread I learned that LC475 can be pushed to the limit of the MC88920 by software, but to exceed this limit MC88920 need to be replaced by MC88916. There were some questions about availability and sources that can be discussed here.
68kmla.org 68kmla.org
Posted by: Bolle on 2024-10-09 03:11:37
System does not fail but frequency does not rise above the limit.
This is not how the MC88920/88916 work. It's impossible that it caps the speed at some value and just won't go any higher than that.
It will always try to output the input frequency times 2 and times 4 to supply those multiplied clocks to the CPU and the rest of the logicboard.
When run out of spec the quality of the output clock signals will decrease... more jitter, not in phase to each other, etc.
Eventually the signals will become bad enough to not meet the requirements of the CPU anymore and the system will become unstable/crash.

I briefly ran a 475 board with a stock MC88920 in place at 40MHz doing one of the older hardware mods and it seemed to do that just fine.
Posted by: MrFahrenheit on 2024-10-10 18:17:40
There is a known limitation:

MC88920 is involved into clock generation as well. Each MC88920 has an individual maximum frequency. This is usually around 36-40MHz.

No matter what you set in Soft Overclock extension this frequency cannot be exceeded.
System does not fail but frequency does not rise above the limit.

Only way out is replacing the MC88920 by a MC88916W80 (require advanced soldering).
@Mustermann

I can confirm using MC88920, it appears you are on to something.

One of my 475 which usss the 88920 caps at 43.24mhz no matter the setting above 43mhz.
Posted by: Powerbase on 2024-10-11 17:54:52
@Mustermann

I can confirm using MC88920, it appears you are on to something.

One of my 475 which usss the 88920 caps at 43.24mhz no matter the setting above 43mhz.
So, basically no matter how high you set your wanted frequency it just tops out at what that 88920 can do?

Fiddling around with mine tonight and was testing how high I can get. At 42MHz right now, going by benchmarks compared to my 475 at its "stock" 33MHz and such.
Posted by: MrFahrenheit on 2024-10-12 04:56:53
So, basically no matter how high you set your wanted frequency it just tops out at what that 88920 can do?

Fiddling around with mine tonight and was testing how high I can get. At 42MHz right now, going by benchmarks compared to my 475 at its "stock" 33MHz and such.

It appears so. Something is going on that limits the maximum speed on some boards. The 88920 chip being the culprit is the leading theory.

Boards with another chip in its place do not seem to have the same limit.

However: is 50mhz really necessary? It’s cool and all, but kind of extreme. A 40+ MHz LC 475 with a full 68040 chip is more than capable of running pretty much any and every 68k app and game at good speeds. So for most people, clocking up to the maximum the board can handle at ~42mhz is probably good enough.

Technically speaking, RAM and VRAM might start to become the limiting factors at 50mhz and above. RAM is both soldered on the board and in a SIMM socket. I would imagine removing onboard chips and using a 50ns SIMM might help with stability a bit at above 50mhz. And using 70ns or even 60ns VRAM would help with video artifacts.
Posted by: stynx on 2024-10-12 05:22:05
However: is 50mhz really necessary?
I think that most things in retro computing are more about possibility than usability. I doubt that anyone will use a 68K machine for anything productive. How far can such a computer be driven, that is the main question here.
Posted by: MrFahrenheit on 2024-10-12 06:55:17
I think that most things in retro computing are more about possibility than usability. I doubt that anyone will use a 68K machine for anything productive. How far can such a computer be driven, that is the main question here.

Yes, I agree. However, what I'm saying is that if one wants a 475 to use on a regular basis as a 68k "go to" Mac, and their machine only clocks as high as 43Mhz, the extra 7Mhz isn't "necessary", and 43Mhz is quite sufficient for a good user experience. It runs games well, and Mac OS 8 actually flies on the thing.

For pushing the limits, absolutely, let's see if somehow we can get 60Mhz out of the things.
Posted by: Powerbase on 2024-10-12 07:49:42
Yes, I agree. However, what I'm saying is that if one wants a 475 to use on a regular basis as a 68k "go to" Mac, and their machine only clocks as high as 43Mhz, the extra 7Mhz isn't "necessary", and 43Mhz is quite sufficient for a good user experience. It runs games well, and Mac OS 8 actually flies on the thing.

For pushing the limits, absolutely, let's see if somehow we can get 60Mhz out of the things.
Oh, for me 42-43MHz is plenty and amazing for not having to use spicy-o-clock, particular capacitors, bodge wires, and ecetera. I need to get it recapped at some point, but I probably won't bother replacing the 88920. Amazing gains with it, as is. High resolutions work and it handles all my 68K needs.

A thousand thanks to Mustermann and everyone that helped.
Posted by: opualuan on 2024-10-12 08:23:50
My board is stuck at 36Mhz. I’m up for changing the 88920, ram ICs, etc. Most things run well as-is but I’d like to take it as far as I can. After all, Doom is still a slide-show full-screen at 36Mhz.
Posted by: Mustermann on 2024-10-12 08:56:26
...
It will always try to output the input frequency times 2 and times 4 to supply those multiplied clocks to the CPU and the rest of the logicboard.
When run out of spec the quality of the output clock signals will decrease... more jitter, not in phase to each other, etc.
...
You are right, "...It will always try to output the input frequency times 2...".
I did some measurements:
Yellow line is input of MC88920 pin 8, blue line is one of the outputs pin 10.

MC88920 TRY to output double of input.
This work well for
Input 12,2MHz -> Output 25MHz (25MHz.png)
Input 20MHz -> Output 40MHz (40MHz.png)

but does fail for any frequency above 40MHz (for the chip in my device):
Input 22,2MHz -> Output 40MHz (45MHz.png)
Input 25MHz -> Output 40MHz (50MHz.png)

Up to input 20MHz output is in sync with input (every second raise of output matches a raise of the input).
Above they are out of sync. No longer every second raise of the output matches to the input, but output is late.

If I have not overseen something, the output frequency of MC88920 is not able to raise above an individual level.
Yes, jitter increase but voltage is ok. At least my device does not crash if I set input frequency of MC88920 to 25MHz but run only at 40Mhz instead of 50MHz.
Posted by: stynx on 2024-11-03 03:06:43
I just replaced the MC88920 with a MC88916-80 an now the system can start and benchmark with 46MHz . The CPU can most likely do >50MHz but i get a black screen at over 46MHz while i hear the HDD chugging on. The graphics-hardware is most likely at its limit. The 68040 CPU is a 40M rated 2E31F Mask. I will try again with a LC-PDS graphics card.
@46MHz, the benchmark results in 2-15% higher scores than a Mac 840AV
Posted by: opualuan on 2024-11-30 22:29:25
I put together another system from spare parts, replaced the 88920 and scsi resistor, moving to the other thread.
Posted by: opualuan on 2024-12-01 22:12:23
I may have discovered a bug in the extension, unless intentional. I had 24-bit memory addressing on due to replacing the pram battery, and noticed the settings don’t work. It shows the startup speed setting in the control panel but clockometer shows 25MHz. Startup icon also doesn’t show the speed.

Switching to 32-bit and works as intended.
Posted by: Callan on 2024-12-01 22:39:51
It disables the driver with 24-bit addressing. You'll notice the extension will be (red if I remeber correctly) and show a 24-bit logo.
Posted by: opualuan on 2024-12-01 23:15:41
Ah. Next time I should read the docs more thoroughly.
Posted by: pizzigri on 2024-12-02 08:59:52
I just replaced the MC88920 with a MC88916-80 an now the system can start and benchmark with 46MHz . The CPU can most likely do >50MHz but i get a black screen at over 46MHz while i hear the HDD chugging on. The graphics-hardware is most likely at its limit. The 68040 CPU is a 40M rated 2E31F Mask. I will try again with a LC-PDS graphics card.
@46MHz, the benchmark results in 2-15% higher scores than a Mac 840AV
Can you share the source of the MC88916-80?
Posted by: Joopmac on 2024-12-02 09:06:07
I bought these https://www.ebay.nl/itm/175963018948
Posted by: zigzagjoe on 2024-12-02 09:32:17
I'm not quite sure why @Phipli extension would disable in 24 bit addressing mode. It should work just fine so long as the extension wraps the register pokes in SwapMMUMode calls. I simply didn't put this in the crude extension I made as I forgot.
Posted by: stynx on 2024-12-04 08:59:46
I bought these https://www.ebay.nl/itm/175963018948
I bought these here:
https://www.ebay.com/itm/186473581042
The chips were in original packaging and unused.
The one chip i have used, is working fine.

-Jonas
< 6 >