Does it hurt to put a fan on an SSD?

I have a plugin drive socket. Some HDD's get very hot inside it, so I
put a fan on it. Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the fan
was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?

On 12/28/18 2:24 AM, Mike wrote:
I have a plugin drive socket.Â* Some HDD's get very hot inside it, so I
put a fan on it.Â* Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the fan
was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?


Hi Mike,

Heat is the enemy of all electronic components. In the
failure analysises (MBTF) I did for the military years ago,
the failure curve for temperature was not linear, but
exponential (e^x). It sky rockets.

So I would stick with cooler is better.

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

-T

T wrote:
On 12/28/18 2:24 AM, Mike wrote:
I have a plugin drive socket. Some HDD's get very hot inside it, so I
put a fan on it. Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the
fan was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?


Hi Mike,

Heat is the enemy of all electronic components. In the
failure analysises (MBTF) I did for the military years ago,
the failure curve for temperature was not linear, but
exponential (e^x). It sky rockets.

So I would stick with cooler is better.

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

-T

That Kingston drive uses a Sandforce controller.

And I'm sure when they wasted money on thermal pads for
that SSD, it was "just cosmetic" :-) Thermal pads cost
a fortune, relatively speaking.

https://www.storagereview.com/kingst...ow_v200_review

There are measured numbers in the StorageReview article,
that don't quite align with how Kingston states their numbers.

https://www.kingston.com/datasheets/SV200S3_us.pdf

SV200S3/256G

Power Consumption 2.0 W (TYP) Idle === controller runs hot ?
4.8 W (TYP) Write Sandforce compressor ?

Those could well be numbers for the highest-capacity drive
in the drive family.

********

Flash memory is actually *annealed* at high temperature.
A write done at high temperature, does less damage
to the gate oxide than a write done at room temperature.

But the retention of what is written, is also affected
by temperature.

If both factors "moved in the same direction", we would
not be wasting pink bubblegum to keep the controller cool.
When they made that drive, they could have put a pink pad
just on the side with the Controller chip. However, they
elected to put pink pads on *both* sides of the PCB,
including the side which just has Flash chips on it.

Note that not every SSD drive has pads in it.
Some designs have elected to dispense with
frippery, to save a few pennies.

*******

The Arrhenius equation is a tricky beast. I was taught in
Chemistry, that "reaction rate doubles for every 7C rise
in temp". However, in Electronics, Arrhenius is "curve fitted"
to field data, to extract a and b. For example, an
electrolytic capacitor reliability doubles for each
15C drop in temperature. The curve is more gradual
than my Chemistry classes would have suggested.
Some other capacitors happen to be 10C per doubling effect.
(There isn't "two times seven" hiding in there somewhere,
the exponent can assume values not consistent with that idea.)

The best data to use is actual field data - quite often
a PhD in reliability will reach a wrong conclusion by
running with extrapolated numbers, without asking others
"whether they've seen the component type, drop like flies".
When the hardware engineers can vouch that a component
doesn't "spontaneously burst into flame", the reliability
engineer then has to go back to his desk and use some
other method to compute a more (intuitively reasonable) value.
Even without field data, people have some idea how often
real-world components fail.

I find the failure rate of my DRAM here, is pretty bad. Personally.
And without any statistically-significant observations
either. Mine seems to "drop like flies" here, which helps
me validate conclusions others might reach. It does
suggest in at least some cases, that chemical contamination
is involved (I have one lot of memory that reliability
is affected by what appears to be metallurgy - the memory
will *die* if you leave it in the DIMM slots while
the PC sits in storage - I have only three sticks of
that crap left).

*******

Summary: Don't leave your electronics in PCs in "thermal shadows"
or dead zones. You don't have to fit squirrel
cage blowers to everything in a PC. And remember
that human body temp is 37C, and 41C isn't really
all that hot.

"If you burn yourself on something, put some cooling on it"

Paul

In article , wrote:

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

it's not nonsense.

one of the best references is google's drive study, which looked at
more than 100k drives from numerous manufacturers and found that cooler
temperatures actually have a *higher* failure rate:

http://static.googleusercontent.com/....com/en//archi
ve/disk_failures.pdf
The data in this study are collected from a large number of disk
drives, deployed in several types of systems across all of Google¹s
services. More than one hundred thousand disk drives were used for
all the results presented here. The disks are a combination of serial
and parallel ATA consumer-grade hard disk drives, ranging in speed
from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in
this study were put into production in or after 2001. The population
contains several models from many of the largest disk drive
manufacturers and from at least nine different models. The data used
for this study were collected between December 2005 and August 2006
....
We first look at the correlation between average temperature during
the observation period and failure. Figure 4 shows the distribution
of drives with average temperature in increments of one degree and
the correspond- ing annualized failure rates. The figure shows that
failures do not increase when the average temperature increases. In
fact, there is a clear trend showing that lower temperatures are
associated with higher failure rates. Only at very high temperatures
is there a slight reversal of this trend.

figure 4:
http://www.storagemojo.com/wp-conten..._temp_age_dist.
png

and as for ssds, heat is even less of an issue, since there are no
moving parts.

On 28/12/2018 16.00, nospam wrote:
and as for ssds, heat is even less of an issue, since there are no
moving parts.

I don't think heat under 100°C affects moving parts much.

There is heat expansion, change of fluidity of oil... I would assume the
designers would make the design best for the typical temperature range.

Consider car engines: they don't work well cold. The cylinders have to
expand, the oil gains fluidity.

--
Cheers, Carlos.

In article , Carlos E.R.
wrote:

and as for ssds, heat is even less of an issue, since there are no
moving parts.

I don't think heat under 100°C affects moving parts much.

that depends on the parts. there is no universal rule.

There is heat expansion, change of fluidity of oil... I would assume the
designers would make the design best for the typical temperature range.

exactly what they do.

Consider car engines: they don't work well cold. The cylinders have to
expand, the oil gains fluidity.

the cylinders and pistons don't expand very much, however, oil designed
for lower temps would be used, and for the really cold climates, there
are engine block heaters.

Carlos E.R. wrote:
On 28/12/2018 16.00, nospam wrote:
and as for ssds, heat is even less of an issue, since there are no
moving parts.

I don't think heat under 100°C affects moving parts much.

There is heat expansion, change of fluidity of oil... I would assume the
designers would make the design best for the typical temperature range.

Consider car engines: they don't work well cold. The cylinders have to
expand, the oil gains fluidity.


The Google study neglected to record humidity.

No points are awarded for bad science.

*******

FDB motors can lose mass at high temperature, due
to migration of the lubricant, past the seal. The
amount of lubricant is so small, the hard drive companies
need a milligram balance to detect the loss. They use
mass measurements, to detect how much lube has left
the motor.

The disk drive companies do reliability studies, because
at least some generations (if not all), use Nidec motors.

When all the lube leaves an FDB motor, the shaft
just seizes up. FDB is frictionless, "until the
juice leaks out". No, holding the drive horizontal
does not prevent fluid escape. It can leave via the
vapor phase.

*******

There's more than one failure mechanism at work in drives.

On older drives, plating failure causes a "spray" of
ferric oxide like material to coat the in-drive filter packs.
The plating failure could be caused by humidity. At least,
that's what the dirty appearance of the filter packs
looks like.

Modern drives have quite different platter construction.
I have no idea whether the surfaces on those degrade
in quite the same way as the older drives. The lubricant
is a wax only three molecules thick, with a second
chemical used as a free radical trap. The surface of the
platter is plated up. A "keeper" layer at the bottom of
the plating, completes the "magnetic circuit" in
the vertically recorded data bit. I've seen no
pictures of filter packs on failed drives, to comment
on whether those "shred" or "shed" or not.

*******

There are a few drives now, with humidity sensors
*inside* the drive. No, I don't have model numbers.

*******

Helium drives are sealed. The drive companies are
climbing on that bandwagon big time. By doing so, it negates
the need for a humidity sensor. A temperature sensor
is still needed, not for "fun", but in order to compensate
writes at high temperature. Disk writing is
temperature sensitive and the drive adjusts.
Some drives also have "high flight" detection,
so if the head is bumped on a write, the drive can
detect a properties change in the head during
the write, and re-do the write on the next rotation.
This is more likely to be an Enterprise feature, but
it's pretty hard to track down all these features
in an authoritative manner (don't expect datasheets,
the companies aren't run by the engineers any more).

Paul

Mike wrote:

I have a plugin drive socket. Some HDD's get very hot inside it, so I
put a fan on it. Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the fan
was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?

My case has 2 drive cages: an upper cage for 5.25" drives (mostly for
optical drives) and a lower cage for 3.5" drives (mostly for HDD/SDD
drives). Obviously with adapters, a smaller drive can be positioned in
either cage. While it is possible to mount the drives next to each
other (i.e., in adjacent slots), I like to leave space between the
drives for airflow around them. I do not have a front-case fan but just
let the drives get airflow from what the other fans (back-case, GPU, and
PSU) suck out of the case. I have 4 drives: 1 SSD, 2 HDD, and 1
optical. The optical and 1 HDD are in the upper cage. The HDD was
moved up there to provide a gap in the 3-drive lower cage between the
SSD and an HDD. My drives are currently at:

SSD (lower cage): 42 C (0 - 70 C, 1M hours MTBF, 150 TBW), 48/52 C max
HDD (lower cage): 43 C (0 - 60 C, 5 years CDL, 0.8% AFR) , 49/52 C max
HDD (upper cage): 41 C (0 - 60 F, 5 years CDL, 0.8% AFR) , 47/55 C max

I consider those as normal temperatures for my drives. All you need to
care about is the *operating temperature range* for the devices. Mine
are shown with lifespan shown parenthesized above: MTBF (mean time
between failure), CDL (Component Design Life), AFR (Annual Failure
Rate), TBW (TeraBytes Written). As you can see, all my drives are
running at about the same temperature. The only thing I'm doing right
now is writing this response. When running disk I/O intensive
processes, the temperature would vary. I also listed the daily max
temperature recorded for my drives over the last month and the max
temperature ever for the drives. My SSD runs pretty much at the same
temperatures as the HDDs.

While spinners usually (but not always) indicate when they are starting
to fail, SSDs usually function right up to their catastrophic failure.
Regardless of which type you use, backups are still required for both
data and hardware recovery.

No idea what you have for a "plugin drive socket". Could be it is
enclosed (no airflow). Could be it is open. Could be a hotswap bay or
something else. Apparently however you are mounting the HDDs into the
bay/receptable provides more airflow than however you are mounting the
SDD into the same bay/receptable. My guess is the HDDs are 3.5" drives,
the SSD is a 2.5" drive, and the bay/receptable is 5.25" wide. Perhaps
the SSD adapter plate is solid, so airflow around it is more
restrictive.

Could also be that you are running more processes or more disk I/O
processes when using the SSD than when you slide in the HDDs. You might
want to use PerfMon to measure the read & write activity to compare the
levels between when using the SSD and an HDD.

As Paul pointed out, maybe your Kingston SSD is a hot drive (compared to
other SSDs). I don't know under what conditions you are trying to
compare temperatures between the Kingston SSD and the HDDs.

In article , Paul
wrote:


The Google study neglected to record humidity.

another study did, and found that humidity was a much bigger factor
than heat.

No points are awarded for bad science.

it's not bad science.

heat is not the issue people claim it to be. it's a myth that has been
busted.

On Fri, 28 Dec 2018 09:55:02 -0600, VanguardLH wrote:

Mike wrote:

I have a plugin drive socket. Some HDD's get very hot inside it, so I
put a fan on it. Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the fan
was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?

My case has 2 drive cages: an upper cage for 5.25" drives (mostly for
optical drives) and a lower cage for 3.5" drives (mostly for HDD/SDD
drives). Obviously with adapters, a smaller drive can be positioned in
either cage. While it is possible to mount the drives next to each
other (i.e., in adjacent slots), I like to leave space between the
drives for airflow around them. I do not have a front-case fan but just
let the drives get airflow from what the other fans (back-case, GPU, and
PSU) suck out of the case. I have 4 drives: 1 SSD, 2 HDD, and 1
optical. The optical and 1 HDD are in the upper cage. The HDD was
moved up there to provide a gap in the 3-drive lower cage between the
SSD and an HDD. My drives are currently at:

SSD (lower cage): 42 C (0 - 70 C, 1M hours MTBF, 150 TBW), 48/52 C max
HDD (lower cage): 43 C (0 - 60 C, 5 years CDL, 0.8% AFR) , 49/52 C max
HDD (upper cage): 41 C (0 - 60 F, 5 years CDL, 0.8% AFR) , 47/55 C max

I consider those as normal temperatures for my drives. All you need to
care about is the *operating temperature range* for the devices.


Yes, I agree.

I have the same four kinds of drives you do. But I haven't looked
inside the case for a while and don't remember what the cage
configuration is.

My SSD is at 31.0 C, and the two hard drives at 32.0 and 33.0

On 12/28/2018 7:00 AM, nospam wrote:
In article , wrote:

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

it's not nonsense.

one of the best references is google's drive study, which looked at
more than 100k drives from numerous manufacturers and found that cooler
temperatures actually have a *higher* failure rate:

http://static.googleusercontent.com/....com/en//archi
ve/disk_failures.pdf
The data in this study are collected from a large number of disk
drives, deployed in several types of systems across all of Google¹s
services. More than one hundred thousand disk drives were used for
all the results presented here. The disks are a combination of serial
and parallel ATA consumer-grade hard disk drives, ranging in speed
from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in
this study were put into production in or after 2001. The population
contains several models from many of the largest disk drive
manufacturers and from at least nine different models. The data used
for this study were collected between December 2005 and August 2006
...
We first look at the correlation between average temperature during
the observation period and failure. Figure 4 shows the distribution
of drives with average temperature in increments of one degree and
the correspond- ing annualized failure rates. The figure shows that
failures do not increase when the average temperature increases. In
fact, there is a clear trend showing that lower temperatures are
associated with higher failure rates. Only at very high temperatures
is there a slight reversal of this trend.

figure 4:
http://www.storagemojo.com/wp-conten..._temp_age_dist.
png

and as for ssds, heat is even less of an issue, since there are no
moving parts.

This thread is spinning out of control.
Detailed data on spinning drives is not relevant.

I read that, in a TLC drive, the difference between
stored value might be as little as 15 electrons.
That ain't much margin.

My engineering training taught me that...
Within normal temperature ranges encountered in
inhabited spaces...
For most electronic things...
Failure rate is an exponential function of temperature.

I've taken steps to reduce the SSD temperature with a fan.
It's idling at 41C. I don't have any high write rate data
because, at this stage, I'm doing everything I can to REDUCE
writes.

I've read that SSD writes are less stressful
at slightly elevated temperatures.

I'm experimenting on the Kingston SSD, but the eventual
drive will be a Samsung 860 EVO 500GB.

The question is:
Would service life of the SSD INCREASE by a significant amount
if I let the temperature rise to 50C? 60C?

On 28-12-2018 23:06, mike wrote:
On 12/28/2018 7:00 AM, nospam wrote:
In article , wrote:

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

it's not nonsense.

one of the best references is google's drive study, which looked at
more than 100k drives from numerous manufacturers and found that cooler
temperatures actually have a *higher* failure rate:

http://static.googleusercontent.com/....com/en//archi
ve/disk_failures.pdf
The data in this study are collected from a large number of disk
drives, deployed in several types of systems across all of Google¹s
services. More than one hundred thousand disk drives were used for
all the results presented here. The disks are a combination of serial
and parallel ATA consumer-grade hard disk drives, ranging in speed
from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in
this study were put into production in or after 2001. The population
contains several models from many of the largest disk drive
manufacturers and from at least nine different models. The data used
for this study were collected between December 2005 and August 2006
...
We first look at the correlation between average temperature during
the observation period and failure. Figure 4 shows the distribution
of drives with average temperature in increments of one degree and
the correspond- ing annualized failure rates. The figure shows that
failures do not increase when the average temperature increases. In
fact, there is a clear trend showing that lower temperatures are
associated with higher failure rates. Only at very high temperatures
is there a slight reversal of this trend.

figure 4:
http://www.storagemojo.com/wp-conten..._temp_age_dist.
png

and as for ssds, heat is even less of an issue, since there are no
moving parts.

This thread is spinning out of control.
Detailed data on spinning drives is not relevant.

I read that, in a TLC drive, the difference between
stored value might be as little as 15 electrons.
That ain't much margin.

My engineering training taught me that...
Within normal temperature ranges encountered in
inhabited spaces...
For most electronic things...
Failure rate is an exponential function of temperature.

I've taken steps to reduce the SSD temperature with a fan.
It's idling at 41C. I don't have any high write rate data
because, at this stage, I'm doing everything I can to REDUCE
writes.

I've read that SSD writes are less stressful
at slightly elevated temperatures.

I'm experimenting on the Kingston SSD, but the eventual
drive will be a Samsung 860 EVO 500GB.

The question is:
Would service life of the SSD INCREASE by a significant amount
if I let the temperature rise to 50C? 60C?

No.

In article , mike
wrote:

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

it's not nonsense.

one of the best references is google's drive study, which looked at
more than 100k drives from numerous manufacturers and found that cooler
temperatures actually have a *higher* failure rate:

http://static.googleusercontent.com/....com/en//archi
ve/disk_failures.pdf
The data in this study are collected from a large number of disk
drives, deployed in several types of systems across all of Google¹s
services. More than one hundred thousand disk drives were used for
all the results presented here. The disks are a combination of serial
and parallel ATA consumer-grade hard disk drives, ranging in speed
from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in
this study were put into production in or after 2001. The population
contains several models from many of the largest disk drive
manufacturers and from at least nine different models. The data used
for this study were collected between December 2005 and August 2006
...
We first look at the correlation between average temperature during
the observation period and failure. Figure 4 shows the distribution
of drives with average temperature in increments of one degree and
the correspond- ing annualized failure rates. The figure shows that
failures do not increase when the average temperature increases. In
fact, there is a clear trend showing that lower temperatures are
associated with higher failure rates. Only at very high temperatures
is there a slight reversal of this trend.

figure 4:
http://www.storagemojo.com/wp-conten..._temp_age_dist.
png

and as for ssds, heat is even less of an issue, since there are no
moving parts.

This thread is spinning out of control.

bad pun.

Detailed data on spinning drives is not relevant.

it is when one claims heat causes mechanical drives to fail.

I read that, in a TLC drive, the difference between
stored value might be as little as 15 electrons.
That ain't much margin.

yet they're are very reliable, and there is now qlc.

On 12/28/18 6:16 AM, Paul wrote:
T wrote:
On 12/28/18 2:24 AM, Mike wrote:
I have a plugin drive socket.Â* Some HDD's get very hot inside it, so I
put a fan on it.Â* Brought temperatures down dramatically.

I have a SSD in a 2.5-3.5" converter box that plugs in.
Whoever designed the box didn't pay any attention to air flow.
I drilled it full of holes, but there's really no internal airflow
path from bottom to top.
This particular Kingston SV200s3256G runs at 41C with little
activity.
That's about 10C greater than the spinners in the box.

Then I read this article:

https://www.anandtech.com/show/9248/...data-retention

That suggests that hotter temperatures are less stressful for SSD
writes and that data retention is better. Now, I'm wondering if the
fan was a bad idea.

I bought a Samsung EVO 860 500GB drive, but thought I'd get the kinks
worked out on the old Kingston.

My rule is, "cooler is better", but seems to fail here.
So, What's the optimal operating temperature for an SSD?


Hi Mike,

Heat is the enemy of all electronic components.Â* In the
failure analysises (MBTF) I did for the military years ago,
the failure curve for temperature was not linear, but
exponential (e^x).Â* It sky rockets.

So I would stick with cooler is better.

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

-T

That Kingston drive uses a Sandforce controller.

And I'm sure when they wasted money on thermal pads for
that SSD, it was "just cosmetic" :-) Thermal pads cost
a fortune, relatively speaking.

https://www.storagereview.com/kingst...ow_v200_review

There are measured numbers in the StorageReview article,
that don't quite align with how Kingston states their numbers.

https://www.kingston.com/datasheets/SV200S3_us.pdf

SV200S3/256G

Power Consumption 2.0 W (TYP) Idle === controller runs hot ?
Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â* 4.8 W (TYP) WriteÂ*Â*Â*Â* Sandforce compressor ?

Those could well be numbers for the highest-capacity drive
in the drive family.

********

Flash memory is actually *annealed* at high temperature.
A write done at high temperature, does less damage
to the gate oxide than a write done at room temperature.

But the retention of what is written, is also affected
by temperature.

If both factors "moved in the same direction", we would
not be wasting pink bubblegum to keep the controller cool.
When they made that drive, they could have put a pink pad
just on the side with the Controller chip. However, they
elected to put pink pads on *both* sides of the PCB,
including the side which just has Flash chips on it.

Note that not every SSD drive has pads in it.
Some designs have elected to dispense with
frippery, to save a few pennies.

*******

The Arrhenius equation is a tricky beast. I was taught in
Chemistry, that "reaction rate doubles for every 7C rise
in temp". However, in Electronics, Arrhenius is "curve fitted"
to field data, to extract a and b. For example, an
electrolytic capacitor reliability doubles for each
15C drop in temperature. The curve is more gradual
than my Chemistry classes would have suggested.
Some other capacitors happen to be 10C per doubling effect.
(There isn't "two times seven" hiding in there somewhere,
the exponent can assume values not consistent with that idea.)

The best data to use is actual field data - quite often
a PhD in reliability will reach a wrong conclusion by
running with extrapolated numbers, without asking others
"whether they've seen the component type, drop like flies".
When the hardware engineers can vouch that a component
doesn't "spontaneously burst into flame", the reliability
engineer then has to go back to his desk and use some
other method to compute a more (intuitively reasonable) value.
Even without field data, people have some idea how often
real-world components fail.

I find the failure rate of my DRAM here, is pretty bad. Personally.
And without any statistically-significant observations
either. Mine seems to "drop like flies" here, which helps
me validate conclusions others might reach. It does
suggest in at least some cases, that chemical contamination
is involved (I have one lot of memory that reliability
is affected by what appears to be metallurgy - the memory
will *die* if you leave it in the DIMM slots while
the PC sits in storage - I have only three sticks of
that crap left).

*******

Summary: Don't leave your electronics in PCs in "thermal shadows"
Â*Â*Â*Â*Â*Â*Â*Â* or dead zones. You don't have to fit squirrel
Â*Â*Â*Â*Â*Â*Â*Â* cage blowers to everything in a PC. And remember
Â*Â*Â*Â*Â*Â*Â*Â* that human body temp is 37C, and 41C isn't really
Â*Â*Â*Â*Â*Â*Â*Â* all that hot.

Â*Â*Â*Â*Â*Â*Â*Â* "If you burn yourself on something, put some cooling on it"

Â*Â* Paul


Hi Paul,

You are only taking into account he memory cells. You are leaving
off the effect of temperature on all the surrounding components

-T

On Fri, 28 Dec 2018 15:00:02 -0000, nospam wrote:

In article , wrote:

By the way, the same heat is better nonsense kicks around
every so often for mechanical hard drives as well.

it's not nonsense.

one of the best references is google's drive study, which looked at
more than 100k drives from numerous manufacturers and found that cooler
temperatures actually have a *higher* failure rate:

http://static.googleusercontent.com/....com/en//archi
ve/disk_failures.pdf
The data in this study are collected from a large number of disk
drives, deployed in several types of systems across all of Google¹s
services. More than one hundred thousand disk drives were used for
all the results presented here. The disks are a combination of serial
and parallel ATA consumer-grade hard disk drives, ranging in speed
from 5400 to 7200 rpm, and in size from 80 to 400 GB. All units in
this study were put into production in or after 2001. The population
contains several models from many of the largest disk drive
manufacturers and from at least nine different models. The data used
for this study were collected between December 2005 and August 2006
...
We first look at the correlation between average temperature during
the observation period and failure. Figure 4 shows the distribution
of drives with average temperature in increments of one degree and
the correspond- ing annualized failure rates. The figure shows that
failures do not increase when the average temperature increases. In
fact, there is a clear trend showing that lower temperatures are
associated with higher failure rates. Only at very high temperatures
is there a slight reversal of this trend.

figure 4:
http://www.storagemojo.com/wp-conten..._temp_age_dist.
png

That would be http://www.storagemojo.com/wp-conten...p_age_dist.png

Otherwise I get "This page isn’t working, storagemojo.com redirected you too many times."

Don't wrap links please.

and as for ssds, heat is even less of an issue, since there are no
moving parts.