Thermal dissipation, RF shielding and ways of improving the lot

Disk drives, Monitors, SuperCPU etc.
Stealth
Member
Member
Posts: 19
Joined: Mon Mar 27, 2017 3:27 pm
Location: Croatia - Zagreb

Thermal dissipation, RF shielding and ways of improving the lot

Post by Stealth » Fri Apr 07, 2017 9:26 am

A few days ago a relatively new YouTube channel MindFlareRetro ran a test of how much the RF shielding and its heatsink tabs actually help dissipate the heat from a PLA.

I knew the PLA in particular is a very busy chip, which makes it draw a lot of current and naturally dissipate a lot of heat - but I had no idea the RF shield did so little to draw heat away from it. That got me thinking about both the heat issues and the RF shielding and how to solve it. I
Again, my two current patients are a black-screen breadbox (250469) with a dead PLA and a black-screen 64C (250425) with an unknown cause. The breadbox is about to receive a PLAnkton so I'm hoping that'll both revive it and lower the heat a bit.

----

So as far as passive thermals go, would removing the RF shield tabs and using thermal pads and small 8mm × 8 mm × 5 mm heatsinks be a good option? From what I've seen, each IC would need two of those to get sufficiently covered?

I've also seen people use active cooling on their VIC-II and other chips - am I right to assume this should only be done if a heatsinked IC is still exceeding 40°C or thereabouts? Would it be logical to just connect the fan to the nearest +5V/GND lines with an parallel fly-back diode, or would I be better off finding a different source for the fan's power?

For the RF shielding, I know there's a good reason for it, but the metal plate appears to be overkill and the cardboard-metal foil appears to be underkill. I'm a bassist who'd already done RF shielding to a bass with excellent results so I'm wondering if it would also be a good idea to cover the inside of the 64's case in self-adhesive copper tape (and an insulating plastic layer to prevent shorts) and solve it that way.

Finally, what of the RF shielding over the VIC-II and the RF modulator? I know the VIC-II is sensitive to noise and I understand the logic behind shielding it, but would using something like a LumaFix board solve that (even though I suspect RF shielding and LumaFix solve different types of EMI). However, I don't plan to use the RF modulator at all, I will only be need the Y/C output - is there even a way to remove it and clean up the signal that way?



Ready.
Member
Member
Posts: 10
Joined: Wed Mar 29, 2017 2:42 pm
Location: Trieste - Italy

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by Ready. » Fri Apr 07, 2017 4:40 pm

in 2 of my old breadbins I added some custom cooling, but since I HATE to drill the case to put fans, in this mod I used some special arrangement taking advantage of existing venting holes. So no C64 case was harmed. I just used hot plastic glue (easily removable) to put things in place. Runs very silent. It cools rather good, check the SId temp graph:
https://ibb.co/deGWFk
https://ibb.co/mux5ak
https://ibb.co/gdNtT5
https://ibb.co/eNyYT5
https://ibb.co/hTwyvk
https://ibb.co/bvB91Q
https://ibb.co/bwxSo5

User avatar
eslapion
Member
Member
Posts: 702
Joined: Mon Jul 20, 2015 10:11 am
Location: Canada

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by eslapion » Fri Apr 07, 2017 5:44 pm

The large RF shield in a C64 actually traps heat. It doesn't help dissipate at all, quite the contrary.

The reason why 82S100 based PLAs go bust is specifically because the technology they employ forces them to use a lot of energy (around 100mA) to perform the task they do, all of which is turned to heat. When going to PLAnkton, you go from antiquated NMOS technology to very high speed CMOS technology which runs at a small fraction of its real limit.

It's pretty much the same difference between the MOS 6502 CPU and Rockwell's R65C02. At 1MHz, the first draws about 60mA and the second draws about 6mA.

PLAnkton actually draws close to 22mA so the gain in energy efficacy isn't as great. That's because it uses an RC filter as a delay component which causes losses but also because it's a programmable logic chip with 2 independent 54V18 grids (36 macrocells) all of which has to be powered for the chip to operate but most of which remains unused because it's not needed to perform the functions of a PLA.

There are actually 3 different factors which limit the operating speed of the XC9536XL (and reduce the required power) at the core of PLAnkton: 1. The version we buy is the slowest one 2. It is programmed to operate in low power mode 3. All outputs have the slew rate limiter enabled

The gain in energy efficiency is also significant with TOLB. While the genuine 8701 consumes 30mA on PAL C64 and 25mA on NTSC computers, TOLB operates with 9mA of power.

On my own C64 with board 250466, I have replaced the 6567R9 with a 8562 (and rerouted the 12Vdc input to the 5Vdc). Going fron NMOS to HMOS cut the heat dissipation to about 1/3rd. I also replaced the 6510 with a 8500 CPU.
Wealth, like happiness, is never attained directly. It comes as a by-product of providing a useful service. -Harland D. Sanders

Stealth
Member
Member
Posts: 19
Joined: Mon Mar 27, 2017 3:27 pm
Location: Croatia - Zagreb

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by Stealth » Sat Apr 08, 2017 5:03 pm

Ready. wrote:
Fri Apr 07, 2017 4:40 pm
in 2 of my old breadbins I added some custom cooling, but since I HATE to drill the case to put fans, in this mod I used some special arrangement taking advantage of existing venting holes. So no C64 case was harmed. I just used hot plastic glue (easily removable) to put things in place. Runs very silent.
This looks very interesting, indeed. I couldn't really gauge that there was all that much space in the breadbin. Wonder if something similar would be applicable to a flatbox.
eslapion wrote:
Fri Apr 07, 2017 5:44 pm
The large RF shield in a C64 actually traps heat. It doesn't help dissipate at all, quite the contrary.
Yipe. Definitely sounds like the bottom RF shield needs to go right out.

I just took apart the 64C (doing a retrobright on the case tomorrow) and I noticed the metal keyboard supports look like they're riveted to the top RF plate, or at least a rivet-like flange. I know I need the keyboard supports, so, is the only way to separate them to dremel the flange off or use a slightly larger bore to eat away at the "channel"? Is the top RF shield a problem-maker anyway?

I don't want to sink any cash into a replacement keyboard support for the stock 64 just yet - I'll probably be getting a Return Magazine replicated case for a future C64R and I know I'll need a 3D printed keystand.
eslapion wrote:
Fri Apr 07, 2017 5:44 pm
PLAnkton actually draws close to 22mA so the gain in energy efficacy isn't as great. That's because it uses an RC filter as a delay component which causes losses but also because it's a programmable logic chip with 2 independent 54V18 grids (36 macrocells) all of which has to be powered for the chip to operate but most of which remains unused because it's not needed to perform the functions of a PLA.

There are actually 3 different factors which limit the operating speed of the XC9536XL (and reduce the required power) at the core of PLAnkton: 1. The version we buy is the slowest one 2. It is programmed to operate in low power mode 3. All outputs have the slew rate limiter enabled

The gain in energy efficiency is also significant with TOLB. While the genuine 8701 consumes 30mA on PAL C64 and 25mA on NTSC computers, TOLB operates with 9mA of power.
It's a shame they don't produce an even lower-end model, it sounds like you did all you could to lower power draw - a near 80% drop in power usage is no small thing.
As for the TOLB, it's good to know it's been optimised as well - if it turns out to be a culprit on any of the machines, I'll look into it as a replacement.
eslapion wrote:
Fri Apr 07, 2017 5:44 pm
On my own C64 with board 250466, I have replaced the 6567R9 with a 8562 (and rerouted the 12Vdc input to the 5Vdc). Going fron NMOS to HMOS cut the heat dissipation to about 1/3rd. I also replaced the 6510 with a 8500 CPU.
My breadboard (250425) already has an 8500R3, amusingly enough, though the VIC-II remains a 6569R3. I have yet to see the picture from it to know what's up.

So the only question now is what to do with the VIC-II and RF shielding. The VIC-II is getting its own heatsink anyway, and I don't think I'll be keeping the RF modulator since S-Video is vastly superior (and easier to connect to a modern monitor) - can those RF boxes come off with no side-effects?

User avatar
eslapion
Member
Member
Posts: 702
Joined: Mon Jul 20, 2015 10:11 am
Location: Canada

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by eslapion » Sat Apr 08, 2017 9:59 pm

Stealth wrote:
Sat Apr 08, 2017 5:03 pm
My breadboard (250425) already has an 8500R3, amusingly enough, though the VIC-II remains a 6569R3. I have yet to see the picture from it to know what's up.
It has to be. Boards 250425 and 250466 still feed 5Vdc and 12Vdc to the VIC-II. A small hack is required to feed 5Vdc on both voltage sources for compatibility with the 856X VIC-IIs.

Board 250469 feeds 5V on both voltage sources and is therefore only compatible with 856X VIC-IIs.
So the only question now is what to do with the VIC-II and RF shielding. The VIC-II is getting its own heatsink anyway, and I don't think I'll be keeping the RF modulator since S-Video is vastly superior (and easier to connect to a modern monitor) - can those RF boxes come off with no side-effects?
You have to install some video amplifier board to replace them but there is no need to remove the RF modulator to get S-Video. Luma and chroma are separate on the 8 pin DIN video output connector, the chroma signal is just a bit too strong and can be reduced to the proper intensity with a 360 Ohms resistor.
Wealth, like happiness, is never attained directly. It comes as a by-product of providing a useful service. -Harland D. Sanders

Stealth
Member
Member
Posts: 19
Joined: Mon Mar 27, 2017 3:27 pm
Location: Croatia - Zagreb

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by Stealth » Mon Apr 10, 2017 10:12 am

eslapion wrote:
Sat Apr 08, 2017 9:59 pm
It has to be. Boards 250425 and 250466 still feed 5Vdc and 12Vdc to the VIC-II. A small hack is required to feed 5Vdc on both voltage sources for compatibility with the 856X VIC-IIs.

Board 250469 feeds 5V on both voltage sources and is therefore only compatible with 856X VIC-IIs.
I've read about the mod - all it takes is to make sure pin 13 (VDD) is rerouted to the nearest +5 VDC instead of the +12 VDC on the socket pin below it, right?
eslapion wrote:
Sat Apr 08, 2017 9:59 pm
You have to install some video amplifier board to replace them but there is no need to remove the RF modulator to get S-Video. Luma and chroma are separate on the 8 pin DIN video output connector, the chroma signal is just a bit too strong and can be reduced to the proper intensity with a 360 Ohms resistor.
Good point, not going to touch the RFM, then. I know of the chroma fix so I'm just going to fit a resistor onto the cable plug's chroma pin and call it a day.

User avatar
eslapion
Member
Member
Posts: 702
Joined: Mon Jul 20, 2015 10:11 am
Location: Canada

Re: Thermal dissipation, RF shielding and ways of improving the lot

Post by eslapion » Fri Apr 14, 2017 3:23 pm

Stealth wrote:
Mon Apr 10, 2017 10:12 am
eslapion wrote:
Sat Apr 08, 2017 9:59 pm
It has to be. Boards 250425 and 250466 still feed 5Vdc and 12Vdc to the VIC-II. A small hack is required to feed 5Vdc on both voltage sources for compatibility with the 856X VIC-IIs.

Board 250469 feeds 5V on both voltage sources and is therefore only compatible with 856X VIC-IIs.
I've read about the mod - all it takes is to make sure pin 13 (VDD) is rerouted to the nearest +5 VDC instead of the +12 VDC on the socket pin below it, right?
Correct!
Wealth, like happiness, is never attained directly. It comes as a by-product of providing a useful service. -Harland D. Sanders

Post Reply Previous topicNext topic