Frying An Arm Chip

[petelewin]

Genuine questions:
How can a comany say "this will take 10 years of use in these circumstances, or five years in these circumstances, before it will break" if they haven't run it for that long, which they can't with cutting edge tech. eg a car shows no sign of rust for eight years, then within 6 months it looks like it's got measels.

You put the car in an hot, humid environment, bombard it with small stones and spray it with salt water. It now takes 2 months to rust, and you apply a mixture of science, experience, statistics and guesswork to estimate how long that would be under normal conditions.

 

[Dimacus]

I sort of remember some discussions about PXA210/250 (Early XScale) based devices haveing some overheating problems, when overclocked and placed in direct sunlight (For use as a gps navigator in a car, as an example).

So I looked up the datasheet for it http://www.alldatasheet.com/datasheet-pdf/pdf/144909/INTEL/PXA250.html And I took a quick look trough it.
If I'm not entirely mistaken it's thermal resistance is almost 10 degrees C higher than the omap.

So I tought I would lookup the memory chip placed on top, to see if would add a significant amount of heat, or if it was particularly sensitive to heat.
It seems like this is a good match http://download.micron.com/pdf/datasheets/mcp/152ball_%20nand_lpdram_j4xx_omap.pdf
That is not an exact match vs. the rev 5 board (Which has the number JW298 or JW295, i can't see which, both exist tough but no documentation is publicly available for either of them).
But to no avail, it doesn't list it's temperature characteristics, with the exception of it's recommended operating temperature, which is about the same as the omap's.

[Anecdote]
I had an Ipaq 3995 that I tought died because of an overclock, so I did some searching for a solution to try and revive my ipaq.
Turns out that most devices that 'died' actually just couldn't cope with the overclock but had no way of resetting clock speed after a failure.
So the most common solution to 'dead' devices was to let the battery drain and that reset the clock.
This wasn't the case of my ipaq tough, it turned out that a part of the voltage regulation system or some such had failed.
[/Anecdote]

 

[Exophase]

Genuine questions:
How can a comany say "this will take 10 years of use in these circumstances, or five years in these circumstances, before it will break" if they haven't run it for that long, which they can't with cutting edge tech. eg a car shows no sign of rust for eight years, then within 6 months it looks like it's got measels.

Statistically, how many units do you have to test to destruction to know that the result is represenative?

Where possible they perform testing under accelerated conditions. When something is said to last for a certain amount of time it really means under "normal" operating conditions that are derived statistically. This translates into a certain number of power cycles, thermal cycles (to varying extents), clock gating, load, current draw, etc. Most of these parameters can be accelerated by performing them much more often than normal.

Where this is not an option, conclusions can be drawn by extrapolating trends. If a good amount of data is collected regarding degradation then a curve can be fit to predict how long before it passes some threshold. Although failure points often look binary to us there are really a lot of subtle attributes that can be measured.

 

[WizardStan]

By the way, TI only specifies 500MHz at OPP3. 600MHz is only rated at OPP5. TI doesn't make any official allowance for > 600MHz. So if you consider OPP5 overvolting then even the advertised clock speed officially requires overvolting. On the other hand, OMAP3530s have been shown to run fine at 600MHz/OPP3.

So with that in mind, would running it at 600MHz on OPP3 be considered overclocking then? Should OpenPandora take this into account when drafting up how overclocking affects the warranty? If there are some poor unfortunates that can't reach 600 even in OPP3, and have to go to OPP5 in order to get the full 600, this should be accounted for in their warranty, I should think.

Genuine questions:
How can a comany say "this will take 10 years of use in these circumstances, or five years in these circumstances, before it will break" if they haven't run it for that long, which they can't with cutting edge tech. eg a car shows no sign of rust for eight years, then within 6 months it looks like it's got measels.

If you're business is to make candles, and you have decades of documents which show burn rates of different sizes of candle with different mixtures, you can then engineer a new candle, burn it for a short length of time, and using all the data you've previously collected, extrapolate how much longer it would have burned with reasonable accuracy.
Now somehow make that analogy work for microchips instead of candles and you'll understand where I was going with it. And then explain it to me, because I got lost.

 

[Exophase]

So with that in mind, would running it at 600MHz on OPP3 be considered overclocking then? Should OpenPandora take this into account when drafting up how overclocking affects the warranty? If there are some poor unfortunates that can't reach 600 even in OPP3, and have to go to OPP5 in order to get the full 600, this should be accounted for in their warranty, I should think.

Yes, technically 600MHz at the default OPP3 is overclocking.

OPP5 is "overdrive", but it's not operating outside of TI specifications because TI does specify its performance. TI guarantees 23K power on hours at OPP5 with 100K at other OPP levels, 50K exclusively at OPP5, or 44K across the board. This is in line with the 0 to 90C operating range.

Given that all of these power on hours are well over a year of constant power I don't see why OPP5 should be considered a major risk, if clock speed is kept within 600MHz. We don't know of the introduced risks when clock speed is increased, but I'm still pretty skeptical that you'll damage the chip.

 

[TaG]

Thanks guys for the explanations - it's pretty much what I expected. To me that says OP can't possibly do any testing that is meaningful beyond relatively simple 'do this and it gets way too hot' because they don't have that type of data or experience, nor the time or money to perform/commission tests on multiple units.

 

[Dead1nside]

So with that in mind, would running it at 600MHz on OPP3 be considered overclocking then? Should OpenPandora take this into account when drafting up how overclocking affects the warranty? If there are some poor unfortunates that can't reach 600 even in OPP3, and have to go to OPP5 in order to get the full 600, this should be accounted for in their warranty, I should think.

Yes, technically 600MHz at the default OPP3 is overclocking.

OPP5 is "overdrive", but it's not operating outside of TI specifications because TI does specify its performance. TI guarantees 23K power on hours at OPP5 with 100K at other OPP levels, 50K exclusively at OPP5, or 44K across the board. This is in line with the 0 to 90C operating range.

Given that all of these power on hours are well over a year of constant power I don't see why OPP5 should be considered a major risk, if clock speed is kept within 600MHz. We don't know of the introduced risks when clock speed is increased, but I'm still pretty skeptical that you'll damage the chip.

I don't suppose you could explain what these different OPP levels do, could you please? Does it just increase the voltage, as I've seen in the datasheet. Why isn't there much talk of OPP4?

Thank you.

 

[petelewin]

Doesn't it all depend on TI, as supplier of the OMAP parts? If they will guarantee the OMAP parts to OpenPandora when they've been overclocked, OpenPandora should be happy to pass that guarantee on. If TI will not, then OpenPandora should not - regardless of any suspicions we may have about the silicon suppliers motives.

Thats not the way it works. You buy chips. After that your on your own. Warranty will be provided by Craig and the others, meaning it will go out of there pockets.

B!

Really? Surely TI have to guarantee their chips. I'm sure they limit liability to the price of the silicon, which doesn't help much when you have to scrap an entire motherboard. And for a small number of failures it wouldn't be worth the effort of claiming, but that still leaves the people who know the silicon best saying what they are prepared to guarantee.

 

[darkblu]

This is all your preference, and that's fine, but it's not relevant to the question at hand. While you may view a crash as unacceptable, it's not something that should be capable of bricking a Pandora and therefore isn't something that should bring warranty into question. Otherwise the warranty issue has just got a whole lot worse.

well, i thought the meta-rant nature of my post was going without saying. that said, i don't see how 'random crashes' and 'warranty' are so much unrelated, to the point of warranty being exclusively a matter of bricking alone. imagine the following scenario:

while the majority of pandora owners happily run their unit at 600MHz/OPP3, happless few pandora owners get random crashes a few times a day at the cited clock/OPP. are those pandora owners entitled to an RMA? or will you go ahead and tell them the chip maker rates OPP3 at 500MHz and guarantees nothing for 600/OPP3, ponentially leaving the guys with the impression they got lemons (even though technically they did not)? perhaps you can advise them to go with OPP5 (more on that below)? i guess we can both agree that by being liberal with overclocking, OP put their product above the level of TI support.

By the way, TI only specifies 500MHz at OPP3. 600MHz is only rated at OPP5. TI doesn't make any official allowance for > 600MHz. So if you consider OPP5 overvolting then even the advertised clock speed officially requires overvolting. On the other hand, OMAP3530s have been shown to run fine at 600MHz/OPP3.

my bad, somebody mentioned 600MHz was 'considered stable at OPP3', and i did not bother to re-check with the specs. OPP5 is clearly a state of maximal allowable overdrive, which means even though the chip maker considers the device to be stable at that OPP (mil-grade devices excluded), it practically halves the chip's projected lifespan when used at 'normal' load levels ('power-on hours' really does not say much here). now crank that usage level to extreme - make it reqularly crunch while being roasted on your car's dashboard daily, and that POH figure drops to god knows what. surely, the device will likely last its warranty time, but how many user would knowingly spend it in this manner?

 

[Grench]

So, is there a temperature sensor in the RAM stack on top of the CPU or not?  Is there one anywhere inside the case?  

the OMAP3530 temperature sensor is accessible using the Band Gap Voltage and Temperature Sensor module, which also provides capabilities for current and voltage monitoring

The temperature sensor feature is used to convert the temperature of the device into a decimal value
coded on 7 bits. An internal Analogic to Digital Converter (ADC, realized with a resistor array) is used for
conversion. The recommended operating temperatures have to be in the -40 to 125°C for standard
conversion, and can go up to 160°C when using the thermal shutdown comparator output (TSHUT signal).
This signal indicates a too high temperature of the device. This signal is internally connected to a GPIO
pin.

see page 707 of this PDF

Do we have the "Temperature Sensor Module" included in the OpenPandora console? Or was this module left out of the design? Can we see this in the OS?

Keeping the operating temp between -40 to 125°C is probably a good idea.

In a 30°C environment with 80% relative humidity at sea level with The Doors playing at standard listening levels on a Tuesday afternoon...

What temp does an OpenPandora CPU with a depleted (charging) battery max out at when run at 600Mhz (spec)?
What temp does an OpenPandora CPU with a depleted (charging) battery max out at when run at 720Mhz (spec)?
What temp does an OpenPandora CPU with a depleted (charging) battery max out at when run at 800Mhz (spec)?
What temp does an OpenPandora CPU with a depleted (charging) battery max out at when run at 800Mhz (spec)?

The 160°C thermal shutdown sounds like a must-have... Is our installation able to trigger shutdown based on this threshold?

So with that in mind, would running it at 600MHz on OPP3 be considered overclocking then? Should OpenPandora take this into account when drafting up how overclocking affects the warranty? If there are some poor unfortunates that can't reach 600 even in OPP3, and have to go to OPP5 in order to get the full 600, this should be accounted for in their warranty, I should think.

Yes, technically 600MHz at the default OPP3 is overclocking.

OPP5 is "overdrive", but it's not operating outside of TI specifications because TI does specify its performance. TI guarantees 23K power on hours at OPP5 with 100K at other OPP levels, 50K exclusively at OPP5, or 44K across the board. This is in line with the 0 to 90C operating range.

Given that all of these power on hours are well over a year of constant power I don't see why OPP5 should be considered a major risk, if clock speed is kept within 600MHz. We don't know of the introduced risks when clock speed is increased, but I'm still pretty skeptical that you'll damage the chip.

So, if I read this right, our consoles will be overclocked out of the gate? Interesting.

I'm assuming at OPP5 that thee is a higher voltage sent to the processor - wouldn't this decrease battery life compared to running at OPP3?

 

[Monk]

Now another question: what's the difference between "our" chips and the 720MHz ones TI is talking about ?

120MHz

 

[Sphinxter]

Now another question: what's the difference between "our" chips and the 720MHz ones TI is talking about ?

120MHz

Good answer!

 

[mindlord]

In our survey 35 people agreed with you. The number one answer is still on the board. Mr. Smith - what's the difference between "our" chips and the 720MHz ones TI is talking about ?

 

[WizardStan]

In our survey 35 people agreed with you. The number one answer is still on the board. Mr. Smith - what's the difference between "our" chips and the 720MHz ones TI is talking about ?

Our chips are from the set of chips that are rated for 600Mhz, the 720MHz you're talking about are from a set that is rated for 720Mhz.

 

[mindlord]

Our chips are from the set of chips that are rated for 600Mhz, the 720MHz you're talking about are from a set that is rated for 720Mhz.

Survey says! *DING* 18. Alright, Mr. Monk, will you and your family play, or pass control to the smith family?

 

[petelewin]

Now another question: what's the difference between "our" chips and the 720MHz ones TI is talking about ?

120MHz

And several dollars.

 

[WizardStan]

Our chips are from the set of chips that are rated for 600Mhz, the 720MHz you're talking about are from a set that is rated for 720Mhz.

Survey says! *DING* 18. Alright, Mr. Monk, will you and your family play, or pass control to the smith family?

Your question doesn't make any sense though.
"What's the difference between a P66 and a P100?" "One is clocked at 66 mhz, the other at 100mhz"
"What's the difference between a car traveling at 60 km/h and the same make and model traveling at 100 km/h?" "one is going at 60 km/h, the other at 100 km/h"
"What's the difference between 1kg bag of sand and a 2kg bag of sand?" "one bag weighs 1kg, the other weighs 2kg"

 

[mindlord]

Your question doesn't make any sense though.
"What's the difference between a P66 and a P100?" "One is clocked at 66 mhz, the other at 100mhz"
"What's the difference between a car traveling at 60 km/h and the same make and model traveling at 100 km/h?" "one is going at 60 km/h, the other at 100 km/h"
"What's the difference between 1kg bag of sand and a 2kg bag of sand?" "one bag weighs 1kg, the other weighs 2kg"

I didn't originally ask the question, Magic Sam did. I'm just bored, and playing Family Feud.

But truthfully the question does dig a bit deeper than just simple difference in MHZ. Since the reality may be that there is not quintessential difference between "our" 600mhz chips and TI's 720mhz chips. Since it may just be an arbitrary label applied to squeeze extra revenue out of the same chip fabrication process. Or it could mean the chips are actually tested to run at 720, and the 600's will only clock up to, but not quite 720. The fact is, we don't know. There is no way to know, and TI won't tell anyone because it would defeat the purpose of the labeling to tell anyone.

 

[yodaofborg]

You show me a TI ARM SoC that has died to over-clocking, and I will find out why it really died, or show you a newb that has over-volted.

Nuff said.

 

[darkblu]

But truthfully the question does dig a bit deeper than just simple difference in MHZ. Since the reality may be that there is not quintessential difference between "our" 600mhz chips and TI's 720mhz chips. Since it may just be an arbitrary label applied to squeeze extra revenue out of the same chip fabrication process. Or it could mean the chips are actually tested to run at 720, and the 600's will only clock up to, but not quite 720. The fact is, we don't know. There is no way to know, and TI won't tell anyone because it would defeat the purpose of the labeling to tell anyone.

to my knowledge, it's the same chip fabrication process. all the difference comes from the speed binning - a post-factum process where each die gets tested before being place in the respective clock bin. a customer-vendor agreement (as the one between OP and TI) is based on the rating provided by the chip vendor. so you may buy 10K pieces rated @ 600MHz, and of those 9K pieces may happen to be capable of 1GHz, but the remaining 1K - of plain 600MHz (+ a vendor-standard safety margin). so unless the customer has the time, resources and quipment to re-bin the acquired lot, i.e. replicate the binning process as done by the vendor, those 10K chips are 600MHz-good each - they were priced as 600MHz ones, and you can bitch to TI if they don't perform at that rate.