Overclocking the SX?
peterj
Posts: 24
So, I just discovered that the clock generator on the SK-Key goes to over 100 Mhz, and I must say that I was quite impressed when I was able to push it up to 96 Mhz before there were visible signs of problems. It seemed as if the problem was simply in the switching speed of the clock itself at that point and, the chip only saw a clock edge maybe 1 in 10 times. Most amazingly, when I backed the speed off, the program seemed to resume normal operation -- no reset or data corruption. (ok, my app *is* pretty small and simple...)
So I read in one of the other threads that there was never anything special about the 75 Mhz and 100 Mhz versions of the SX; they were from the same production line as the rest and merely verified at those speeds. What are the issues, if any, of clocking the chips this fast? I ran at 95 Mhz for several minutes and while the chip did get warm, it did not get hot -- I could put my lip to the chip and it wasn't overly warm. However, I did notice that the chip draws a LOT of power when run this fast -- I nearly burned my lip on the LM2940's heat sink! I'm still working on the SX-Tech breadboard so I can't (easily) take any real current readings, but the lip test is not an unreasonable one... (I think!)
Anyways, I'm just curious to know what I should be aware of when running this fast. As an aside, what's the fastest that anyone has been able to reliably clock the SX?
-p.
So I read in one of the other threads that there was never anything special about the 75 Mhz and 100 Mhz versions of the SX; they were from the same production line as the rest and merely verified at those speeds. What are the issues, if any, of clocking the chips this fast? I ran at 95 Mhz for several minutes and while the chip did get warm, it did not get hot -- I could put my lip to the chip and it wasn't overly warm. However, I did notice that the chip draws a LOT of power when run this fast -- I nearly burned my lip on the LM2940's heat sink! I'm still working on the SX-Tech breadboard so I can't (easily) take any real current readings, but the lip test is not an unreasonable one... (I think!)
Anyways, I'm just curious to know what I should be aware of when running this fast. As an aside, what's the fastest that anyone has been able to reliably clock the SX?
-p.
Comments
I don't know about overclocking the SX. I use 20 and 50 mhz usually so there's never an issue. Like PC CPU's, the manufacture puts 100 mhz to the chip, and if it performs as designed and doesn't smoke, then they throw it into the 100mhz bin for sale.
I think I know that you can safely hold your finger on something that is around 150 degrees F for an extended amount of time.
What I want to know is at what temperature does a human being burn their lips? Is this with or without chapstick?
After I burned my lips and other unmentionable appendages, I got me one of those non-contact IR thermometers. It's safer and it's got calibrated, repeatable display.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
There are 10 kinds of people in the world.... those that know binary, and those that don't.
FWIW, all SX devices should be able to overclock at room temp to ~95MHz. Operation over the entire temp range spec'd for the device is another story. You will start to see failures at hot temp in an over clocked device... So watch out if you plan to use them overclocked in automotive or outdoor apps.
Does the SX reliably start every time if it has had to time to stablize its·temperature near an extreem ambient temp?
I am using the SX in an automotive application where the SX inside the vehicle cabin receives power when the engine starts and then goes about its program.
The vehicle cabin can be upwards·of 70C (or -20C overnight) when the car sits with all the windows up for more than an hour in Death Valley (or Minnesota overnight).·I am not sure if the SX is actually reaching that temp however. (also need to check if they are the industrial grade or not)
Sometimes the SX will not run when it is first powered up.....but later (3 ~5 min) the circuit tends to start working after the vehicle cabin starts to cool off (or Heat up).
Ignoring all other possibilities (solder joints etc) for the SX not running, I am simply curious if anybody has had the SX in an environment where the SX itself is at least 50C or more, and it has to get power and run with an ambient of ~70C.
I have about 24 of these that may be out at one time on a test trip (2 ~ 3 / car) and almost everyone seems to have experienced this problem at one point. Also, of the 3 that may be in a car, 1 of the 3 might fail.
(maybe I need to bake one at home in the oven to try and duplicate/diagnose.......)
The circuit at room temp sucks about 85 mA.
The resonator I am using is the little "blue" guy parallax sells.
Watchdog is on, brownout is off. Unfortunately, I have never been on a test trip during a failure to witness the event and diagnose it.
Again, mainly curious on other users experience for reliable or unreliable start ups in extreem heat/cold.
you mention that the SXes get power when the car engine starts.
Starting the engine usually means that the battery voltage drops remarkably due to the current "eaten up" by the starter. In my BMW, it sometimes happens that my radio's nav system resets when I start the engine. Obviously because the battery voltage gets too low.
With the SX, I made the experience that it sometines fails to correctly reset when the supply voltage raises too slowly, or that the SX "mis-behaves" when the supply voltage is not stable enough.
Mabe, the problems you describe, are not caused by high or low temperatures, but by an unstable supply voltage.
Greetings,
G
As it turned out, I had misunderstood something in the SX data sheet. I thought the internal feedback resistor being enabled was all that was necessary for correct operation of the oscillator. It would work fine at normal (room temp) temperatures but would fail at high temperatures. I had to add an external 10k resistor in parallel with the resonator to guarantee that it would start up and continue to oscillate at high temperatures.
I had some chips that wouldn't oscillate at all with the 10k resistor (which was recommende by parallax), so I did some testing with a range from 10k to 100k, and I finally settled on 22k as being an adequate compromise. Seems to do the job. I believe the issue is leakage current at high temperatures; the resistor swamps that out and biases the inverter as an analog amplifier so it can work as an oscillator.
Also don't forget to set the proper oscillator drive in the device line. I use oschs3 setting.
I have run into one other problem with the resonators. The last time I bought 1000 pieces, and of that thousand, most (99.5%) had a J suffix after the 50.00 CM marking. A few had a Q suffix, and these seemed to have problems at higher temperature even with the resistors. At least I *think* that is what happened. Or they could be a different lot number that I had from a previous purchase. But, the appearance is slightly different from the J marked units, which makes me wonder if the Q marked ones might not have a different loading cap value, or no caps at all.
Mike Chadwick