 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> So, you have some equipment for monitoring temparature sensors and
> logging their readings. And you supply some software that downloads this
> data and lets you organise and examine it on a desktop PC.
>
> You also supply your customers with a blob of software which, when
> installed on a PC, somehow tricks the PC into thinking it has a new
> serial port. All data sent to this port is *actually* tunnelled to the
> little black box.
>
> So, between the little black box at one end, and the custom software
> driver at the other end, the hardware and software geniunely believe
> they're still locally connected, yet actually they can be on different
> continents.
>
> Is this a valid solution to a design problem? Or is it a cheap hack?
>
> [Did I mention that the "software" in question appears to be designed to
> work with Windows 3.0? Does that change the answer?]
>
If it is an existing system that needs to be made network 'aware' then
it is a valid solution, but one that should be taken with caution.
Recently I built something that does exactly that.
I have some GPS modules that speak serial.
I built a custom serial/ethernet module. But this one is smart - it has
a buffer and built in web page for basic GPS info.
On the computer I run a ethernet/comm port emulator so that programs
that expect only a comm port can work.
But one better, the serial/ethernet module allows for multiple
connections - I can get multiple computers and applications pulling the
same data on comm ports.
IMHO, a serial-ethernet module should be used as a last resort when
there is no viable alternative.
If it is a mission critical application, then you are just introducing
more points for failure. When one of those points die, then it takes
all that much longer to debug.
This reminds me of what would be put in at the GM-Saturn plant I worked
in several years ago. You kludge a bunch of stuff together until it
works, then repeat and put 50 of them on the plant floor. Even tho if
one of them failed it would stop the production line. I ran into some
strange setups - parallel to serial to ethernet to serial to com
port.... and in one case it took 3 days to figure out what piece failed.
Best of luck!
LAter... Tom
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
scott wrote:
>> every 15 minutes. If they go outside a preset range, an alarm goes
>> off. How much would *you* suggest such a system costs?
>
> Depends on far more than just the raw technical performance.
>
> If I'm going to use it check the temperature in my greenhouse, I would
> imagine a 50 quid jobby from Maplins would do with some cheap hacked
> together Win3.1 program. However, if I'm putting it in a manned
> spacecraft that is orbiting the moon, I would expect way higher
> performance, and an equally higher price.
Well, we'd like to check that the blood samples we're storing actually
about this fact, urgently.
I'm guessing a temparature sensor that actually *works* at temparatures
that low is going to cost a tad more than your average room thermometer,
but beyond that I don't see why it would need to cost more. It seems the
only reason for the higher price is that this is a critical device, so
the suppliers know they can charge the Earth and we will pay it. We have to.
> BTW, the stuff we make here, we always have to put on the 1st page of
> the specification that it is not to be used in anything critical, like
> plane instruments, hospital equipment, traffic signals etc, simply
> because we don't design or test to high enough standards. If we did,
> the cost would be astronomical, for stuff like PC monitors you don't
> need that level of reliability so you get them very cheap.
The mass spectrometers we have here all say "for research and
development only; not for diagnostic procedures" on them. I can't
imagine why - it's a mass spectrometer! Either it measures masses
reliably, or it doesn't. If it does, you can use it for anything you
like. If it doesn't, it's a worthless piece of equipment. So... why the
sticker?!
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Stephen wrote:
> Is the equipment and software certified? Because that adds to the
> cost.
Not by the supplier, I don't *think*. Certainly *we* will be regularly
checking it against an ISO standards-certified reference instrument
[which really *is* insanely expensive]. But I don't *think* the
suppliers actually provide any such guarantees - I'm not sure...
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Well, we'd like to check that the blood samples we're storing actually
> about this fact, urgently.
informed? Would you trust a 20 quid meter from Maplin then?
> I'm guessing a temparature sensor that actually *works* at temparatures
> that low is going to cost a tad more than your average room thermometer,
Not really, thermocouples are cheap (like a fiver) and measure down to
below -100 degrees.
> but beyond that I don't see why it would need to cost more. It seems the
> only reason for the higher price is that this is a critical device, so the
> suppliers know they can charge the Earth and we will pay it. We have to.
Or the fact that there is a much lower chance it will break or mal-function.
A simple example I can think of is that the connector where the thermal
probe plugs in mal-functions somehow so that the electronics thinks that the
temperature is -90 when really it is only -70. Or some solder joint on the
circuit board wasn't made completely correctly and messes up some other
reading in an undetectable way once the temperature and humidity get to a
certain value. There are all sorts of failure mechanisms that needs to be
checked and fixed somehow, and that costs lots of money.
> The mass spectrometers we have here all say "for research and development
> only; not for diagnostic procedures" on them. I can't imagine why - it's a
> mass spectrometer! Either it measures masses reliably, or it doesn't. If
> it does, you can use it for anything you like. If it doesn't, it's a
> worthless piece of equipment. So... why the sticker?!
Because they don't guarantee it will work reliably the whole time. For them
to guarantee that, they would need to do lots of expensive testing on every
unit, probably design in lots of redundant systems, use more expensive
components that have longer lifetimes, use better assembly methods etc.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Well, we'd like to check that the blood samples we're storing actually
>> about this fact, urgently.
>
> is informed? Would you trust a 20 quid meter from Maplin then?
My point exactly - it seems to have nothing to do with how reliable the
device is, only the subjective feeling it gives you in your head knowing
that "I paid lots of money for this, therefore it must be really
reliable". Which isn't terribly scientific...
>> I'm guessing a temparature sensor that actually *works* at
>> temparatures that low is going to cost a tad more than your average
>> room thermometer,
>
> Not really, thermocouples are cheap (like a fiver) and measure down to
> below -100 degrees.
Well, maybe you have to be slightly more careful that, e.g., the
insulator on the wire doesn't become as brittle as glass when it gets
too cold or something.
somebody has to check it won't break or something when it gets that damn
cold.
>> It seems
>> the only reason for the higher price is that this is a critical
>> device, so the suppliers know they can charge the Earth and we will
>> pay it. We have to.
>
> Or the fact that there is a much lower chance it will break or
> mal-function.
Why? Do they actually *do* something different with it? Does the
hardware itself have any actual physical differences to a cheap model?
a several thousand percent markup, and started marketting it at a
different market segment.
On the other hand, I remember researching printers one time. I ended up
looking at two printers. Both Hewlett Packard. Both print at 1200dpi.
Both have built-in network support. Both do duplex printing. Etc. But
one printer was about 10x the price of the other. Why?
The answer is a little number hidden at the bottom of the page
somewhere: "duty cycle". One printer said 1,000 pages/month, the other
said 100,000 pages/month. And you know what? We ended up buying both
printers, and one is still working to this day, the other one we
eventually threw away because it just kept breaking so often.
In this case, it seems paying 10x the price does, in fact, get you a
device which is *physically different*. I don't know if they use thicker
plastic for the drive gears or what, but the more expensive printer was
far more reliable. The cheap one is basically designed to sit in your
house. [Actually, I have one at home. It works just fine. But then, I
hardly ever print anything!] Put that printer in a busy office and it
just can't cope.
>> The mass spectrometers we have here all say "for research and
>> development only; not for diagnostic procedures" on them. I can't
>> imagine why - it's a mass spectrometer! Either it measures masses
>> reliably, or it doesn't. If it does, you can use it for anything you
>> like. If it doesn't, it's a worthless piece of equipment. So... why
>> the sticker?!
>
> Because they don't guarantee it will work reliably the whole time. For
> them to guarantee that, they would need to do lots of expensive testing
> on every unit, probably design in lots of redundant systems, use more
> expensive components that have longer lifetimes, use better assembly
> methods etc.
Well, you're the engineer. But I wonder - if a mass spectrometer that
does one you *can* use for diagnostic procedures cost?! o_O
--
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 15 May 2008 15:18:18 +0200, "scott" <sco### [at] scott com> wrote:
>
>Or the fact that there is a much lower chance it will break or mal-function.
>A simple example I can think of is that the connector where the thermal
>probe plugs in mal-functions somehow so that the electronics thinks that the
>temperature is -90 when really it is only -70. Or some solder joint on the
>circuit board wasn't made completely correctly and messes up some other
>reading in an undetectable way once the temperature and humidity get to a
>certain value. There are all sorts of failure mechanisms that needs to be
>checked and fixed somehow, and that costs lots of money.
You need to ensure that the cold joint (or in this case the hot joint)
is compensated properly for changes in the ambient temperature.
For medical/critical equipment you need certification to ensure
compliance with the laws of the country. And not just UK laws it
should comply with the laws of the registered country. In your case I
imagine that is the USA.
--
Regards
Stephen
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On Thu, 15 May 2008 13:52:37 +0100, Invisible <voi### [at] devnull> wrote:
>Stephen wrote:
>
>> Is the equipment and software certified? Because that adds to the
>> cost.
>
>Not by the supplier, I don't *think*. Certainly *we* will be regularly
>checking it against an ISO standards-certified reference instrument
>[which really *is* insanely expensive]. But I don't *think* the
>suppliers actually provide any such guarantees - I'm not sure...
Buy the cheapest and that is what you get. And is your secondary
calibrator regularly checked against a primary calibrator?
--
Regards
Stephen
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> My point exactly - it seems to have nothing to do with how reliable the
> device is, only the subjective feeling it gives you in your head knowing
> that "I paid lots of money for this, therefore it must be really
> reliable". Which isn't terribly scientific...
Surely you check out the specification before buying something? It will
include details about *how* reliable it is, with real numbers, and under
what conditions it has been tested in. Something you buy from Maplins won't
have this.
> somebody has to check it won't break or something when it gets that damn
> cold.
Exactly - someone has to check - and that costs money.
> Why? Do they actually *do* something different with it?
Yes, it is designed to be more reliable, and they test it to prove this so
they can write it in the specification of the device.
> Does the hardware itself have any actual physical differences to a cheap
> model?
Probably, if you're lucky you could get a *really* reliable one from Maplin
that lasted for 1000 years without any problems. *Lucky* being the key word
here, try explaining to someone's family that they died because some monitor
failed, you can't say "we were unlucky".
> several thousand percent markup, and started marketting it at a different
> market segment.
And put in place processes for guaranteeing the reliability of the product,
at least testing each one and working with their suppliers to ensure that
correct procedures are followed to avoid any unreliabilities. You see those
news stories about paint from China causing problems? Without controlling
your suppliers properly stuff like that happens - and if you are supplying
something you claim to be suitable for critical processes, you are in deep
trouble if you allow your suppliers or your own processes to screw up.
> The answer is a little number hidden at the bottom of the page somewhere:
> "duty cycle". One printer said 1,000 pages/month, the other said 100,000
> pages/month. And you know what? We ended up buying both printers, and one
> is still working to this day, the other one we eventually threw away
> because it just kept breaking so often.
Exactly.
> In this case, it seems paying 10x the price does, in fact, get you a
> device which is *physically different*. I don't know if they use thicker
> plastic for the drive gears or what, but the more expensive printer was
> far more reliable. The cheap one is basically designed to sit in your
> house. [Actually, I have one at home. It works just fine. But then, I
> hardly ever print anything!] Put that printer in a busy office and it just
> can't cope.
The same way that buying a 20 quid power drill from B&Q is fine if you only
use it once a month to drill a few holes. But if you are a builder and you
use it 20 times a day, it's going to break after a week. Which is why you
invest in a way more expensive model. OK so if you bought the cheap one it
*might* last for several months, but it's not worth the risk if you really
need your drill to work every day.
> Well, you're the engineer. But I wonder - if a mass spectrometer that you
> one you *can* use for diagnostic procedures cost?! o_O
Medical equipment is extremely expensive, precisely because there is so much
at risk when things go wrong. The companies that make this stuff need to do
a huge amount of testing and robust design, way above what most other
industries require. It all costs money.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
And lo on Thu, 15 May 2008 14:32:47 +0100, Invisible <voi### [at] devnull> did
spake, saying:
Heh I love it when you get ranty about the price of mission-critical stuff.
Here's a story I was told I forget the exact measurements and prices but
still -
A women approaches a company that make bespoke wooden doors for a new
front door. She can't buy an off-the-rack for £100 because they're two
inches too narrow, they quote her a price of £250. She blows her top
shouting "You're charging me £150 for those two extra inches" and storms
off never to return.
The moral, of course, is that she wasn't be charged an extra £150 she was
being charged £250 to have workmen create for her a uniquely sized door
from scratch.
--
Phil Cook
--
I once tried to be apathetic, but I just couldn't be bothered
http://flipc.blogspot.com
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Invisible wrote:
> but beyond that I don't see why it would need to cost more.
In the US, I'd say it's because if the device fails and all your blood
samples warm up and you lose $1M of revenue, you could reasonably expect
to sue the company that made the device to get your money back.
It's called "incidental damages" here.
If that's what you're paying for, then that's a big part of the price
difference. If the company doesn't warrant that the device will actually
work, then I couldn't say. Around here, a $20 thermometer is going to
have a warranty that says "If it breaks, you might be able to get your
$20 back, but that's about it."
> If it doesn't, it's a worthless piece of equipment. So... why the
> sticker?!
Liability control. If you can't tell when it's broken, and you overdose
someone because your mass spectrometer misread how much drug was in the
solution, it's much more expensive than if you (say) ruin an
experimental car engine by putting the wrong mix of fuel in it.
--
Darren New / San Diego, CA, USA (PST)
"That's pretty. Where's that?"
"It's the Age of Channelwood."
"We should go there on vacation some time."
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
