SX48 OverClocking
crgwbr
Posts: 614
If I use a nice big Heat sink, is it possible to run the SX48 or 28 at 100MHz?· I won't do it if I run the risk of·damaging anything, just thought it would be nice to have an extra 25MHz of Speed for one of my projects.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
NerdMaster
For
Life
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
NerdMaster
For
Life
Comments
You can try it, but you may find that certain chips work, and others don't.
An active heatsink would be even better.
P.S. Ubicom used to make a 100MHz part, I think they were just "cherry-picked" parts that passed the test.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheap used 4-digit LED display with driver IC·www.hc4led.com
Low power SD Data Logger www.sddatalogger.com
SX-Video Display Modules www.sxvm.com
Stuff I'm selling on ebay http://search.ebay.com/_W0QQsassZhittconsultingQQhtZ-1
"USA Today has come out with a new survey - apparently, three out of every four people make up 75% of the population." - David Letterman
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
NerdMaster
For
Life
There was some posts awhile back about people using coins as heatsinks on the SX48.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheap used 4-digit LED display with driver IC·www.hc4led.com
Low power SD Data Logger www.sddatalogger.com
SX-Video Display Modules www.sxvm.com
Stuff I'm selling on ebay http://search.ebay.com/_W0QQsassZhittconsultingQQhtZ-1
"USA Today has come out with a new survey - apparently, three out of every four people make up 75% of the population." - David Letterman
Hey, how about this one:
Three people enter an empty bus. Later, seven people leave the bus. How many people must enter the bus to have it empty again ?
Besides this, I don't recommend pennies as heat sinks. You should use a "real" one, fixed on the chip with heat-conducting glue. On the other hand, you might consider splitting the work-load between two or more SXes. This keeps them cool, does not generate that much EMI, and would not be much more expensive than cooling down an SX driven to its absolute heat limits.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Greetings from Germany,
G
I've been running some simple code on my SX28 very happily at 103 MHz and 5V without ANY heatsink. It gets nice and warm (45C with an I/R thermometer), but certainly not what I would call HOT to the touch.
You can readily try this too by using the SX-Key as your clock, and cranking up the speed to your heart's content!
My particular SX craps out at clocks over 103 MHz. Some will go faster... I had one at 110Mhz once with tweaking the power supply voltage.
I believe the heat generated does vary with the code you are running... probably due to the total number of transitions happening per second.
Cheers,
Peter (pjv)
i've only tried overclocking one of them so far... 80MHz no prob... 90MHz no prob 100MHz NO DICE
95 looks ok too.
Given that I was aiming for 80MHz, I'm happy.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com - a new blog about microcontrollers
I am also interested in running the SX28 at 100MHz. At this time I am using a 50MHz clock for my application. Upgrading to a 75MHz clock is not an attractive option because the clock period becomes a not so usefull 13.3333... nS for practical timing applications ! Clocking the SX at 100MHz would result in a very practical 10nS timing increment for my application.
If you need an effective heatsink for your overclocking experiments I recommend using an old PC CPU heatsink/fan assembly if you can fit it onto your board. The 12V fan should provide sufficient cooling with a reduced 9V supply. You may need to mill away some of the aluminium heatsink to fit over your PC board mounted components or use a heat spreader/spacer on top of the SX . I often use this cost effective method to cool expensive power devices in power amplifiers while under developement in the lab.
Does anybody know if Parallax is planning to release a 100MHz SX processor (like the Ubicom part) in the future ? I will buy some if they do.
Best Regards
electronic
to my knowledge, Ubicom never produced special SX versions for 100 MHz. As Bean mentioned, I also think they "cherry-picked" those that passed the 100 MHz test.
Devices like the SX get hot because each logical level change causes a logical cell on the chip to be charged or discharged, i.e. a current flows into, or out from such cells. Although the interconnections between the cells are very short, they have a certain resistance. So, there is a voltage drop across these interconnections. This voltage, multiplied by the current means dissipated power, or heat. Thus, the faster you clock an SX more heat will occur on the chip.
I agree with Peter (pjv) that the generated heat does vary with the executed code. For example, when the SX runs a simple, end-less loop (JMP $), only the state of the PC is changed, besides other states in the control unit, and on the program memory address and data bus. On the other hand, when it reads or writes a file register, additional state changes of the W register, or a file register will occur, as well as state changes on the register address and data bus. The same is true for all other controllers, or processors. When you have a PC with a temperature-controlled fan, you my notice that the fan sometimes turns on more often (or runs faster), depending on the program you are running.
A while ago, I had posted some graphs in the forum showing the SX supply current vs. various clock frequencies. These graphs show that the supply current is remarkably lower when you reduce the supply voltage from 5 to 3.3 Volts, or even lower. As lower supply current automatically means less power dissipation, or less heat, you should consider running the SX at 3.3 Volts or lower when driving it to the clock limit.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Greetings from Germany,
G
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
NerdMaster
For
Life