So now I'm wondering about different types of LCD.

At the bottom end, you have those silver ones they put in calculators. 
Usually a 7-segment one. I have a graphing calculator somewhere with a 
dot matrix LCD, but still silver, with no greyscale capability.

I'm wondering... Is there a specific *name* for this type of LCD? What 
does it cost to put one of these into your product? Like, if you're 
buying a couple of thousand of those things, what's the unit price? How 
much circuitry does it take to drive it?

At the other end of the scale, you have the stupid-DPI full-colour 
back-lit LCDs with touch sensitivity that they put into every mobile 
phone, ever. What do *those* things cost? I'm guessing you need way, 
*way* more hardware to drive it. (An entire framebuffer, for starters...)

I've heard it said that on "most" electronic devices, the buttons and 
lights are the most expensive part. As in, removing one button or one 
indicator light is a significant cost saving. So a toaster with three 
buttons is "much" more expensive to make than the same toaster with only 
two buttons. I'm not sure why this is; presumably because it's awkward 
to assemble a mechanical switch? (I.e., you have to have an extra step 
where a machine inserts all the moving parts into the right places.) I'd 
be interested if anybody has numbers.



I'm just wondering... High-colour LCDs are in even cheap phones now, so 
the LCD itself can't be all that expensive. So why don't more devices 
have these displays? For example, at the gym, the treadmill has a 
7-segment LED display, and three buttons for cycling through the menues 
and selecting the option you want.

So you've got this huge black slab of plastic, with a tiny, tiny little 
display in the middle. Why not just replace the entire thing with an LCD 
touch-screen? I mean, you damn well *know* there's a microcontroller in 
there. (Something has to drive the complex, unintuitive menu system!) 
You could show all the statistics *at once*, without having to scroll 
through them. (And in metric and imperial too.) You could show all the 
available programs at once, and a single touch would choose one. It 
would be drastically easier to use, and it would look hip and modern. 
(I.e., people would totally buy it.)

Naturally, a full-colour multi-touch LCD costs more than a 7-segment LED 
display. But then, you *know* a quality treadmill that's going into a 


expensive than that. In terms of material cost, it should be negligible. 

probably a more realistic comparison in terms of size.)

What *will* cost money, of course, is the cost of changing a product 
you're already making. Like, if you already make this thing, and people 
are already buying it, why change it?



I don't know if you knew, but a *huge* amount of lab equipment is like 
this. We had a temperature monitoring system. It could *only* 
communicate by RS-232. So we enquired about the networked version... It 
turns out this consists of an RS-232 to Ethernet adaptor, and then some 
software on your PC which makes the antiquated Windows 3.1 control 
software think that the remove RS-232 port is actually local.

Seriously, not only did the company not spend $1 adding an Ethernet port 
to the product itself, it also didn't bother to update its utterly 
ancient control software. When we upgraded to Windows XP, it stopped 
working, because it assumes that every user will have write permission 
to the C:\WINDOWS folder (which is here it stores TempConfig.ini, which 
has all its settings). It took me 20 minutes to figure out (mostly by 
*guessing*) which file it couldn't access, and manually tweak the file 
permissions. Seriously, they didn't even need to update the *software*, 
just fixing the *installer* would have been enough. But noooo...

(Did I mention the software comes on a CD-ROM, which contains four 
folders named "DISK1", "DISK2", "DISK3" and "DISK4", each of which is 
almost *exactly* 1.44MB in size. Hmm...)

Then again, maybe it's not the manufacturer we ought to be mad at. When 
a lab has a validated process for doing something, if you were to 
*change* that process, you would have to revalidate everything. This 
causes massive, massive inertia to change. This is why 100% of all lab 
equipment uses RS-232, and floppy disks and dot matrix printers are 
ubiquitous. (Also, *our* lab has nowhere near enough money; I don't know 
if that's common to all labs, or just ours. I imagine it's not 
*uncommon*...)

One time, we validated a new release of a program, and upgraded all our 
stuff. Then we found some really bad bugs in the software. I installed 
the hotfix, and we found it fixed all the problems. And then I had to go 
*uninstall* the hotfix, because it is not validated. So real bugs that 
are causing real problems and real lost productivity could not be fixed 
because nobody did the magic "validation" dance. Sad, really...



I'm sorry, what was I talking about? Oh, yeah, LCDs. How come (for 
example) my washing machine doesn't have one? I've only ever seen them 
on the most expensive ultra-luxury models. Surely this stuff isn't 
actually that expensive to manufacture anymore?

Post a reply to this message

orchid, do you ever submit your lengthy rants to

http://thedailywtf.com

? :)

Post a reply to this message

Am 11.10.2015 um 13:20 schrieb Orchid Win7 v1:
> So now I'm wondering about different types of LCD.
> 
> At the bottom end, you have those silver ones they put in calculators.
> Usually a 7-segment one. I have a graphing calculator somewhere with a
> dot matrix LCD, but still silver, with no greyscale capability.
> 
> I'm wondering... Is there a specific *name* for this type of LCD?

"7-segment LCD" maybe?

> What
> does it cost to put one of these into your product? Like, if you're
> buying a couple of thousand of those things, what's the unit price?

You might ask a quote from a company that manufactures those thingies.
Or a distributor. You know, like Farnell:

http://uk.farnell.com/displays-lcd_7-segment

> How much circuitry does it take to drive it?

Zip, as long as you choose a static panel (one single common contact,
plus one contact per segment) as opposed to a multiplexed one, or go for
a complete module. A microcontroller with enough output pins and the
correct output voltage, that's all.

Directly driving a multiplexed panel (segments connected in a
matrix-like fashion) is a much more complex issue (especially so in the
case of non-TFT panels), as you'll need the capability to drive each
line to more than two different voltage levels. Some microcontrollers
provide built-in multiplexed LCD support though, and may only need a
dozen pieces of "bird seed" (resistors and capacitors) to generate and
stabilize a bunch of different voltages.

> At the other end of the scale, you have the stupid-DPI full-colour
> back-lit LCDs with touch sensitivity that they put into every mobile
> phone, ever. What do *those* things cost? I'm guessing you need way,
> *way* more hardware to drive it. (An entire framebuffer, for starters...)

No, not really - you only need more output pins to drive the thing. And
as they're usually TFT, they may actually be easier to drive than
7-segment muxed displays. Depending on your display content you don't
even need a framebuffer: If text is all you want to display, you could
create the actual data on the fly from a text buffer as you refresh the
panel.

> I've heard it said that on "most" electronic devices, the buttons and
> lights are the most expensive part. As in, removing one button or one
> indicator light is a significant cost saving. So a toaster with three
> buttons is "much" more expensive to make than the same toaster with only
> two buttons. I'm not sure why this is; presumably because it's awkward
> to assemble a mechanical switch? (I.e., you have to have an extra step
> where a machine inserts all the moving parts into the right places.)

It's often worse than that: Depending how the switch is mounted, manual
labour may be required.


> I'm just wondering... High-colour LCDs are in even cheap phones now, so
> the LCD itself can't be all that expensive. So why don't more devices
> have these displays? For example, at the gym, the treadmill has a
> 7-segment LED display, and three buttons for cycling through the menues
> and selecting the option you want.

Probably because (1) display quality is /the/ top selling point in the
mobile phones market, even in the low-cost niche, while it is less
relevant in treadmills; (2) in the mobile phone market per-unit margins
are minimal, and profits are made by volume, whereas in the treadmill
business volume is low, and profits are made on a per-unit basis, so
mobile phone manufacturers can get significantly higher discounts on
displays than treadmill manufacturers; and (3) mobile phones have
mind-bogglingly fast development cycles, and adapting an existing
hardware design to use a different display is a routine process; also,
serving a volume market, development teams can be large, possibly even
with a dedicated expert for display technology; in the treadmill
business, development cycles are much longer, adapting the hardware
design to a different display is less pressing and therefore not done
with every cycle, and the development budget may even be too small to
have any dedicated embedded computing hardware expert on the team at all.


> What *will* cost money, of course, is the cost of changing a product
> you're already making. Like, if you already make this thing, and people
> are already buying it, why change it?

Also, what *will* cost money is changing an existing embedded computer
design to drive an entirely different type of display, as opposed to
just adding one more switch or even just changing the software.


> I'm sorry, what was I talking about? Oh, yeah, LCDs. How come (for
> example) my washing machine doesn't have one? I've only ever seen them
> on the most expensive ultra-luxury models. Surely this stuff isn't
> actually that expensive to manufacture anymore?

The question is, would you pay for it?
Would your decision to purchase model X of brand A instead of model Y of
brand B really depend on the presence of an LCD, or would it instead be
driven by a gazillion of other factors? And would you really pay, say,
30 GBP (*) for a multi-color 3.5" 320x240 dot matrix LCD that doesn't
show any more information than the otherwise identical model with just a
rotary knob, a set of switches and a 7-Segment LED display?

(* Wildly guessing here, based on Farnell's offer of the MIDAS
MCT035L6TW320240LML for about 15 GBP apiece at a quantity of 600+, plus
an overhead of 15 GBP for protecting the module against humidity and
vibrations, using a more powerful microprocessor with more I/O pins,
physically connecting the display to the microcontroller, and last not
least developing the modification in the first place.)

Post a reply to this message

> At the bottom end, you have those silver ones they put in calculators.
> Usually a 7-segment one. I have a graphing calculator somewhere with a
> dot matrix LCD, but still silver, with no greyscale capability.
>
> I'm wondering... Is there a specific *name* for this type of LCD?

Passive matrix. As opposed to active matrix.

> What
> does it cost to put one of these into your product? Like, if you're
> buying a couple of thousand of those things, what's the unit price? How
> much circuitry does it take to drive it?

$1 would be my guess. You can probably find a bulk lot of 1000 on 
aliexpress or something. Regarding the electronics it's quite simple, 
just bang a voltage onto each segment you want "on", reverse the 
voltage, and then repeat very rapidly. The most important thing to 
remember with any LCD type is that the *time-average* voltage across any 
segment must be zero, or you will damage the LC material. If you've got 
a microcontroller or CPU already it should be easy to do in software.

> At the other end of the scale, you have the stupid-DPI full-colour
> back-lit LCDs with touch sensitivity that they put into every mobile
> phone, ever. What do *those* things cost? I'm guessing you need way,
> *way* more hardware to drive it. (An entire framebuffer, for starters...)

Price is roughly proportional to screen area (the factory doesn't care 
much how many screens it gets out of each "mother sheet"). For a 4" 
smartphone screen with touchpanel I'd say about $10. A laptop display 
could be $50. Really the price depends heavily on the volume, who the 
customer is, whether it is a custom design/off the shelf, what 
additional electronics/touch is included etc.

Yes you'll need some electronics to continually pump in the pixels. Most 
screens like that will use an LVDS interface, but you can get HDMI/DVI 
-> LVDS converters if your "system" already has those.

> I've heard it said that on "most" electronic devices, the buttons and
> lights are the most expensive part. As in, removing one button or one
> indicator light is a significant cost saving. So a toaster with three
> buttons is "much" more expensive to make than the same toaster with only
> two buttons. I'm not sure why this is; presumably because it's awkward
> to assemble a mechanical switch? (I.e., you have to have an extra step
> where a machine inserts all the moving parts into the right places.) I'd
> be interested if anybody has numbers.

In the world of very high volume mass production you can almost ignore 
tooling costs, so the cost is mostly made up of two parts: materials and 
assembly. A toaster might be a bit of bent steel with two plastic sides, 
you're talking $1 or so for those parts plus heating eleemnts, wiring 
and some screws, so maybe $10 total? A switch on the other hand is like 
a complicated mini-assembly in itself, so although the material cost is 
very low, there is a high assembly cost, so a switch could easily cost 
$1 too. So you've got a total cost of $13 or $14 depending on how many 
switches you use. If you're expecting to sell a million toasters, using 
one less switch will make you $1m more profit :-)

> I'm just wondering... High-colour LCDs are in even cheap phones now, so
> the LCD itself can't be all that expensive. So why don't more devices
> have these displays? For example, at the gym, the treadmill has a
> 7-segment LED display, and three buttons for cycling through the menues
> and selecting the option you want.

How many treadmills are they expecting to sell? If it takes an 
electronics engineer 12 months to design, prototype, test and debug the 
system, will it be worth it for them? Or would it be more profitable to 
just reuse the cheapo $5 LCD and board from the previous treadmill?

> I'm sorry, what was I talking about? Oh, yeah, LCDs. How come (for
> example) my washing machine doesn't have one? I've only ever seen them
> on the most expensive ultra-luxury models. Surely this stuff isn't
> actually that expensive to manufacture anymore?

How much profit do you think the manufacturer makes on a cheap washing 
machine? Even if there were zero development costs, is it even worth 
them spending an extra $10 on a high res screen with touch panel?

Post a reply to this message

On 10/12/2015 11:47 AM, scott wrote:

> So you've got a total cost of $13 or $14 depending on how many
> switches you use. If you're expecting to sell a million toasters, using
> one less switch will make you $1m more profit :-)

Another way of looking at it is that if you add another switch you can 
charge more. So if you increase the price of a toaster by $10 and add a 
switch that cuts out one heating element to toast bagels on one side 
only. For a million units you make an extra $9 million profit.

-- 

Regards
     Stephen

Post a reply to this message

>> So you've got a total cost of $13 or $14 depending on how many
>> switches you use. If you're expecting to sell a million toasters, using
>> one less switch will make you $1m more profit :-)
>
> Another way of looking at it is that if you add another switch you can
> charge more. So if you increase the price of a toaster by $10 and add a
> switch that cuts out one heating element to toast bagels on one side
> only. For a million units you make an extra $9 million profit.

Yes, and I guess that's why most manufacturers will have several 
different models so that no matter how much you are willing to spend 
they will still maximise their profit :-)

BTW I was just in Tescos and the cheapest toaster was £4.50. So maybe my 
estimates of part costs easrlier were a little high.

Post a reply to this message

On 10/12/2015 1:02 PM, scott wrote:
> Yes, and I guess that's why most manufacturers will have several
> different models so that no matter how much you are willing to spend
> they will still maximise their profit :-)
>

I don't know why you would think that. ;-)

> BTW I was just in Tescos and the cheapest toaster was £4.50. So maybe my
> estimates of part costs easrlier were a little high.

I thought that the figures were for simplicity.

-- 

Regards
     Stephen

Post a reply to this message

On 11/10/2015 08:53 PM, clipka wrote:
> Am 11.10.2015 um 13:20 schrieb Orchid Win7 v1:
>> What
>> does it cost to put one of these into your product? Like, if you're
>> buying a couple of thousand of those things, what's the unit price?
>
> You might ask a quote from a company that manufactures those thingies.
> Or a distributor. You know, like Farnell:
>
> http://uk.farnell.com/displays-lcd_7-segment

I've never heard of Farnell before. And that's kinda the thing; I don't 
work in this industry, so I have no idea who you'd go to.

I *am* kinda surprised you can get prices from a website. I thought 
these things are usually like "call us and we'll think about telling you 
a price [which will only be valid for 21 days]"...

>> I've heard it said that on "most" electronic devices, the buttons and
>> lights are the most expensive part.
>
> It's often worse than that: Depending how the switch is mounted, manual
> labour may be required.

Ouch! Man, that's gotta get expensive fast...

>> I'm just wondering... High-colour LCDs are in even cheap phones now, so
>> the LCD itself can't be all that expensive. So why don't more devices
>> have these displays?
>
> Probably because (1) display quality is /the/ top selling point in the
> mobile phones market, even in the low-cost niche, while it is less
> relevant in treadmills; (2) in the mobile phone market per-unit margins
> are minimal, and profits are made by volume, whereas in the treadmill
> business volume is low, and profits are made on a per-unit basis, so
> mobile phone manufacturers can get significantly higher discounts on
> displays than treadmill manufacturers; and (3) mobile phones have
> mind-bogglingly fast development cycles, and adapting an existing
> hardware design to use a different display is a routine process; also,
> serving a volume market, development teams can be large, possibly even
> with a dedicated expert for display technology; in the treadmill
> business, development cycles are much longer, adapting the hardware
> design to a different display is less pressing and therefore not done
> with every cycle, and the development budget may even be too small to
> have any dedicated embedded computing hardware expert on the team at all.

That... is quite an interesting analysis, actually.

So essentially you're saying it has nothing to do with the manufacturing 
costs at all, and it's all about what it will cost to design the thing 
versus what they can sell it for. (?)

> Also, what *will* cost money is changing an existing embedded computer
> design to drive an entirely different type of display, as opposed to
> just adding one more switch or even just changing the software.

I have no idea how this stuff works. Like, I have no idea what kind of 
circuitry it takes to hook a microcontroller up to a display, and how 
much different display types differ.

> The question is, would you pay for it?



some washing machines. (I.e., I could buy 2 for that price.)

> Would your decision to purchase model X of brand A instead of model Y of
> brand B really depend on the presence of an LCD, or would it instead be
> driven by a gazillion of other factors?

Honestly? The main concern is the physical size of the hole it needs to 
fit into. And what colour it is. That usually limits you to only one 
specific model possible, IME...

Post a reply to this message

>> I'm wondering... Is there a specific *name* for this type of LCD?
>
> Passive matrix. As opposed to active matrix.

OK.

What about the blue/purple ones? You know, like they had on ancient 
laptop screens?

>> What does it cost to put one of these into your product?
>
> $1 would be my guess. You can probably find a bulk lot of 1000 on
> aliexpress or something.

Wow. I didn't expect it to be quite *that* cheap. Even a simple resistor 
costs more than that! (Or maybe it doesn't if you buy a thousand of 
them, IDK.)

> The most important thing to
> remember with any LCD type is that the *time-average* voltage across any
> segment must be zero, or you will damage the LC material.

Oh, that's interesting.

> If you've got
> a microcontroller or CPU already it should be easy to do in software.

Yeah, that's what I figured. And most electric devices with more than 
one button on them probably have a CPU already. (Probably...)

>> At the other end of the scale, you have the stupid-DPI full-colour
>> back-lit LCDs with touch sensitivity that they put into every mobile
>> phone, ever. What do *those* things cost? I'm guessing you need way,
>> *way* more hardware to drive it. (An entire framebuffer, for starters...)
>
> Price is roughly proportional to screen area (the factory doesn't care
> much how many screens it gets out of each "mother sheet").

Linearly proportional? Or is it "more expensive" to buy really big screens?

> For a 4"
> smartphone screen with touchpanel I'd say about $10. A laptop display
> could be $50.

OK. Again, that's not nearly as much as I thought. Makes me wonder why 

makes it a high-colour touch-screen device...

> Really the price depends heavily on the volume, who the
> customer is, whether it is a custom design/off the shelf, what
> additional electronics/touch is included etc.

I imagine touch would be really expensive - but IDK...

> In the world of very high volume mass production you can almost ignore
> tooling costs, so the cost is mostly made up of two parts: materials and
> assembly. A toaster might be a bit of bent steel with two plastic sides,
> you're talking $1 or so for those parts plus heating elements, wiring
> and some screws, so maybe $10 total?

I have a really cheap toaster somewhere that looks like that might 
literally be all there is to it.

I also have a more expensive toaster with a defrost mode, reheat mode, 
and automatic timing adjustment based on the bread thickness. Because 
when you press the lever, a pair of grills spring out and hold the slice 
in place, and (I presume) tell the microcontroller how thick the bread 
is. Also, the case is polished chrome with a two-tone paint job.

I don't know, but I'm guessing it must be reasonably expensive to 

something?)

> A switch on the other hand is like
> a complicated mini-assembly in itself, so although the material cost is
> very low, there is a high assembly cost, so a switch could easily cost
> $1 too. So you've got a total cost of $13 or $14 depending on how many
> switches you use. If you're expecting to sell a million toasters, using
> one less switch will make you $1m more profit :-)

Sure, but you don't *make* the switch yourself, surely? You buy an 
off-the-shelf microswitch, and then you just got to mount the thing 
securely and give it a nice-looking button that matches the casing. (?)

I agree, depending on production volumes, the profit difference might be 
large. ;-)

> How much profit do you think the manufacturer makes on a cheap washing
> machine?

That's actually an interesting question.


machine is one of those rare objects that contains *metal*. Obviously 

it. IDK, but I doubt it. The machine is also probably quite awkward to 
physically assemble. And they're large and heavy, so storing and 
shipping probably isn't cheap.


is pure profit. I guess there's enough people in the market that it 
can't be *that* profitable, but... IDK.

> Even if there were zero development costs, is it even worth
> them spending an extra $10 on a high res screen with touch panel?

I guess it's only really "worth it" if it makes more people buy them. I 
don't know for real how much extra money somebody would actually pay for 
a screen.

Then again, my washing machine [which I didn't pay for] has about a 
dozen lights on it, and 6 buttons. So the buttons can't be *that* 
expensive! (Well, for a _washing machine_ they probably aren't...)

Post a reply to this message

On 11/10/2015 08:43 PM, nemesis wrote:
> orchid, do you ever submit your lengthy rants to
>
> http://thedailywtf.com
>
> ? :)

Why? Do you think I should? :-P

Post a reply to this message