Hacker News new | past | comments | ask | show | jobs | submit login
Don’t use 7-segment displays (2011) [pdf] (thimbleby.net)
109 points by parsecs 35 days ago | hide | past | favorite | 78 comments

In Oxford city centre, and probably elsewhere, they replaced the live signage at bus stops. These tell you when the next bus is due, the time and date, etc.

They used to be amber LED matrixes. High contrast, high legibility, low resolution.

Now they are colour LCD panels. High resolution, high tech, probably high cost. Now they can show the logos of the bus operators, smooth scrolling text and any arbitrary images. None of which anyone really needs except perhaps the egos of the bus companies.

And they are completely illegible. The viewing angle is limited, the text is smaller, unnecessary stuff is there because the implementors can.

Careful what you ask for.

Amber LED matrices are my favorite tech for signage. The individual LEDs are sunken into the black matrix to provide extra contrast from direct light exposure. There's nothing that comes close for public signage that needs to be placed outdoors. These are by far not cheap though.

Still, this article is about 7 segment displays. It's a quirky article, but I get the point: many people abuse 7-segment displays as an aesthetic choice where there wouldn't be no point in it.

7 and 16-segment displays can still be a superb choice in many scenarios. The contrast is still unmatched to other display techs. For bench instruments, 7-segment displays (especially the VFD ones) are much more readable than the alternatives IMHO. The second best choice IMHO is OLED, however glare in bright environments is still worse, and getting a large OLED gets expensive very quickly.

For big machinery, a large 7-segment display can be read across the room in any condition. A large readable display on a lathe is much better in terms of safety than an smaller OLED display, even considering the chance of a stuck segment.

The main airport in my country switched their digital signage to LCD screens with fancy animations.

And now they periodically show full screen ads instead of arrival / departure information.

It makes me miss the big red letter panel from the early 2000s.

This is really where OLED is the right answer, I think. On smaller devices it's more or less entirely replaced 7-segs and similar displays. High contrast, high legibility, high resolution, but high cost. But if you're considering the cost of a monochrome/limited-colour OLED vs an LCD against the rest of the cost of upgrading a bus stop signage display the differential won't be a huge factor - labour, metalwork, etc will be. I've also seen high-resolution signage-oriented LED/LCD displays used for this sort of thing successfully.

Oxford's travel signage has always been a bit crap, though (and not alone, of course, in this - lots of places in the UK and Europe (and I'm sure US) have crap signage following similar patterns).

The amber matrices were at least clearly readable, but the presentation of information left a lot to be desired. Even on legible displays, constant scrolling displays that alternate between scrolling directions for different text displays for no good reason, and other such madness make it so hard to actually get information about stops that I ended up writing my own wrappers around APIs/website scraping to make a legible display on my phone.

This sort of thing should really be considered in terms of accessibility - for able-bodied people with good eyesight and a quick mind these displays may be legible, but a good % of your users are disabled, elderly, etc - none of the moving display, low-viewing-angle, small-text displays will be usable, but partially sighted/elderly people will generally make do with big friendly text that doesn't change except when it needs to.

The cost of small (128x32) OLED displays with integrated drivers has absolutely plummeted in recent years. They are now cheaper than 7 segments unless you only have 2 digits to display. On top of that the oled display modules are typically 2 wire serial or something similarly economical, while 7 segments require more connections and external components. It's hard to imagine them coming back in small devices.

MicroLED is the future for such use case. Sony's CREDIS is already deployed on some space like Apple Store. It's really wide view angle, bright, no burn-in, simple technology.

How long do new OLED displays hold up? (Honest question, I don't know). And how is the viewing angle? Do you need a glass panel to protect it, does that have issues with glare?

The viewing angle is good.

The burn-down is a real issue with OLEDs, though :(

You're not thinking big picture, nor looking at things from a business perspective. The transit authority can sell ad space on those LCDs, thus increasing revenue as well as reducing TCO of the displays.

Telstra did a similar thing here in Australia - there are some old regulations that let them install payphones pretty much wherever they want which was probably good back in the day when payphones were essential, however they started installing new "payphones" in prominent advertising locations where the whole back of the phone was a massive advertising display.


Colour LCD panels probably work out cheaper that LED matrixes.

An LED matrix has a great number of individual components, and is probably hand assembled due to low demand (compared to an LCD panel). Whereas an LCD panel are mass produced in every size and shape, with standard controllers you can pick up for pennies.

I'm not sure if/why they would be hand-assembled, as SMD pick & place machines exist for the lighting industry: https://youtu.be/Y45COU2LdXg

(Granted, these are not individually addressable, but that depends on the PCB.)

The biggest issue for LED matrix displays is not construction, but 1) density of the display and 2) size/power draw requirements.

You need to consider there's half a million of LEDs in a typical signage device. These consume and heat-up significantly compared to an OLED/LCD panel.

For a panel of the same size, OLED/LCD has higher density, which allows to pack more information (or have better quality signage in general).

It all depends on the usage scenario. In an indoor setting with fixed lighting, a larger LCD screen can be just as readable as OLED, can be very large nowdays (or tiled to achieve the same), and cost a fraction of a LED matrix.

OLED could definitely replace LCDs entirely, but it's still very expensive when it comes to larger panels, which is why it's just slowly replacing smaller displays at the moment.

For LCD panels, one thing a lot of people don't know if that there are several different types, and the cheapest is quite awful. As all the prices come down I'm seeing fewer and fewer of the TN panels, but they can still be found. Terrible for conference rooms.

There's also a cheap kind of panel family where they don't put every color on every pixel. That is, rather than RGB RGB RGB, they are RG GB RG GB RG GB. For most video is hard to tell the difference, but it trashes text. Someone put a TV that was both TN and this reduced color in a conference room.... took me a few months to get it swapped out, but enough people finally saw it and realized it was the TV and not the computer or their eyes that I was able to get it done.

Absolutely true. I wish TN didn't exist entirely, but I guess it works in some weird price/quality zone. When going rock-bottom, LCD panels can indeed get very bad.

A big problem stems from the fact that to most people "flat panels" are all the same (LCD-whatever, OLED, even plasma)..

A LED panel consumes a lot of power for a reason. It is damn bright, and remains readable in broad summer sunlight. Most LCDs are useless in this setting, except maybe transreflexive monochrome (usually black-on-olive or similar).

I almost forgot transreflexive even existed :(. It has been a long time since I've seen some, but I remember the contrast being acceptable even in direct sunlight. But always in very small form factors.

Are there new devices which make use of this display tech?

I can find a few parts offered with a quick search, all small screens.

Also Garmin makes navigators with such screens, e.g. Garmin Montana.

I had the same irritation when our desk phones were replaced by Voip around 15 years ago, I had a nice avaya 8434 phone with bright green LED display, you could see caller ID from across the room if you were away from your desk talking to a colleague.

The new voip phones had hard to read LCD displays and previous hard key features buried in menus - they made the office look more modern but sucked otherwise.

Many of the author's points are worth consideration (especially regarding use of 7-segment displays for anything other than digits), but come on:

> A bespoke display panel was used; itwould have been safer, easier to read, more versatile, andcheaper had a hi-res display been used

Obviously the author has never actually tried to design a product both ways. Displays are expensive, and I'm not just talking about the panel itself. Usually you must go up a level or two on your processor hierarchy: a 7-segment display or matrix can be driven by a 4-bit microcontroller. A VGA-resolution display will require a high-end microcontroller, usually a 32-bit Cortex-M3 or M4 these days. A true "hi-res" display, which has admittedly drifted up in quality since the author wrote this, will push you into application processor (i.MX) territory.

That may not sound like much, but typical costs for those levels including supporting circuitry are in the neighborhood of $1, $10, and $100. This is because the tiny microcontroller needs nothing; the fancy apps processor needs DDR SDRAM, NAND flash, a PMIC, and is probably a BGA you have to route on an 8-layer board, with GHz MIPI signalling flying around, and don't even get me started on the software complexity.... You can do all of these things, and if it genuinely makes your product better, you should. But do not pay the cost if your product does not benefit.

A coworker of mine likes to tell a story about cutting the sales price of a medical device from $5,000 to $500. The main source of that savings would have been removing the display. (Yes, really. Medical displays are expensive.) This was rejected by product management, on the reason that "The people who buy this stuff are not the people who use it. The people who buy it like shiny-looking things. It is much easier to sell a product that looks complicated and expensive to a procurement officer than one that looks simple and possibly less advanced, even if they are the same inside. That display does do something -- it sells units."

In operation, the display had a bunch of whiz-bang crap that could have been deleted with no consequence, and two touch buttons for "arm" and "stop". Of course, you had to use a physical button to actually activate it; IEC 60601 wouldn't allow using just a touchscreen to start or stop a medical procedure! So the display really didn't do anything functional....

So, no, high-res displays are not always the answer. Think carefully about the cost!

To be fair, it's not always either-or: an OLED display of 128x32 or 128x64 resolution costs peanuts and requires 4 lines for routing (I2C), and I've used them on 8bit uCs.

Like https://s.click.aliexpress.com/e/_9flw1u

And there are higher resolution and higher color fidelity displays which are also cheap and driven by I2C.

$1.30 isn't peanuts.

1 penny is peanuts: https://www.alibaba.com/product-detail/6-digit-8-characters-... https://www.alibaba.com/product-detail/TN-segment-custom-LCD...

People use seven segment LCDs because they're cheap. No more, no less.

On really high volume consumer electronics, multiply any cost savings by tens of millions. Now you're talking real money. The HW engineers have knife fights in the hallways over resistors that cost less than a penny.

I'm kidding about the knives. But cost reduction meetings can be pretty brutal, especially when the HW team is looking at the SW and firmware folks across the table and saying, "Sure, this is painful, but you can program around it. It saves us ten cents." Ten cents is a fortune.

In pre-ARM days I once saw a HW team seriously propose half populating memory. No, not that way. The other direction, where they'd just take away the upper half of each byte of RAM and thus save a chip. "Only use the low four bits of each byte, you can just program around that, right?" The only reason that cost reduction didn't fly was there was no way to make the stack work properly for interrupts.

First rule of consumer electronics: "No one cares how much you suffered getting it out the door."

I think the author's argument stands: except for novelty products, there are very few applications where the misinterpretation risk is worth less than $1. Certainly nothing that has to do with safety, medical equipment, professional and industrial tooling, food processing, automotive (pressures, temperatures etc.). An one segment error leading to a 30 PSI reading when the pump is set at 90 could mean instant death when the tire blows up.

> there are very few applications where the misinterpretation risk is worth less than $1

Any product required to run on a battery 5+ years can only be met by seven segment LCDs. That's includes safety equipment. Now what?

In addition, that seven segment display has well characterized failure modes and you can make very solid statistical predictions. When was the last time you actually remember a burnt out LCD segment (I can't remember one) vs. the last time you had nasty lines running through a dot matrix LCD (happens to flat panels all the time)? On a matrix, display RAM failure is a display failure. Seven segment LCDs don't need RAM backing.

Are you willing to have your medical device offline because you lost an LCD matrix row and now you are required by process to wait for a replacement? How much time and money did that cost you?

These are the exceptions. LCDs are normally about cheap. And if the device isn't cheap enough, you won't have that safety/medical device at all because nobody will buy it.

Engineering is tradeoffs, not absolutes.

The main takeaway here is exactly this huge risk of gradual deterioration leading to invalid output that appears valid. That a thermometer fails in the hospital or runs out of battery is not a major problem - another one is close by and procedures are in place to deal with it. Meanwhile, a wrong temperature read can lead to life critical decisions by the medical staff (ex. 39 vs 33).

While I agree with your engineering tradeoffs perspective, failing safe is also a major recurring theme in engineering. When errors or loss of functionality happen they should be obvious and stop the system (machine + human operators) from spiraling out of control. A compiler won't silently write integers to a byte* pointer, because that might kill your patient where the parameter goes over 255. A failed display is preferable to a working display that outputs wrong data.

By the way, 14 and 16 segment displays are widely available in passive LCD formats and draw the same power, while the upgrade to dot matrix LCD is not that prohibitive, I had it on my MP3 player 15 years ago.

People make far more errors like not checking at all--and you can't even get them to use checklists which are close to free. LCD segment failure causing a misinterpretation is so far down the list of things to go wrong that it might as well be zero.

14/16 segment displays require more microcontroller pins (twice as many, in fact). That goes back to cost.

Your MP3 has a very expensive, high-capacity rechargeable battery which will not hold a charge 5 years even if you connect it to nothing.

Yeah, $1.30 isn't peanuts... but that price is for low volumes (single modules) on a retail site. High volume production probably slashes that down to about half, and then we're in the price range of 7 segment LCD panels like linked above, especially when counting in a controller (and we'll likely need a controller if driving it from an 8 bit uC).

Of course, when absolute $ savings are more important than resolution / legibility, tens of cents are worth looking into.

Power requirements are another topic, and monochrome LCDs are unmatched there.

Fonts for that can still take up a considerable amount of storage. And 7-segmented displays can typically update faster, especially the LED variants.

That said, I mostly agree that a small OLED or similar is usually a lot better to read.

I think in the sense they are using it, even b/w 128x64 would be "high-res": enough to draw nicely readable digits. (design trade-offs of course also apply to that, but not as starkly)

The STM32G0 line might surprise you. It’s a Cortex M0+ and allows using SPI flash for graphics assets:


What a quirky paper. I worked at a major winery that had dozens of tanks, up to 100,000 liters each. A single failure in a tank represented many millions of dollars. Our extremely expensive micro-oxygen pump used a standard 7-segment display combined with little red LEDs representing On/Off next to a list of tank numbers. The 7-segment display was used to read and to set the units (μg, mg , etc), amount, and rate of oxygen delivery. I never made the mistake of misconfiguring, but one of the senior Winemakers did resulting in some very angry customers, a 7 figure loss, and many contracts.

I'm left with two opinions: 7-segment displays are not usually fail-safe, and have the potential to be fail-deadly (figure 1). In my entire life I've never had occasion to use a tool with a 7-segment readout, handheld or otherwise, that actually failed or that I misread due to ambiguity.

So, I think a better conclusion would be don't use 7-segment displays in important places.

Alternatively, if the failure would cost you millions, stick in 2 of those displays which will cost you $1 each, so if they fail you're likely to notice.

For a seven figure potential loss, this probably justifies a raspberry pi or two, on different UPSes, hooked up to backup sensors that can email a few different supervisors when the parameters exit the acceptable envelope.

You didn't even read the introduction.

> In this paper, we raise design issues and present recommendations; for clarity we say “never” for design choices that are inappropriate for dependable or safety-critical applications. For novelty and other non-dependable applications, obviously examination of cost and design trade-offs may lead to other decisions, and such decisions should be backed by competent empirical evaluation.

> In this paper, we raise design issues and present recommendations; for clarity we say “never” for design choices that are inappropriate for dependable or safety-critical applications.

How does your conclusion differ from their initial recommendation?

For hobbyist use (where quantity is usually 1 and shipping and work matter more than a $1 part cost), I've found that going overkill is usually worth it.

A ESP32 board with a built in display costs around $10, and is pretty universal and easy to use, and scale makes them cheap. Need more? Stick a phone to it.

I'm surprised there isn't some common "overkill" industrial board similar to a very cheap smartphone, minus battery/charging, camera, and GPS, plus connectors for various protocols/pins for interfacing with other devices. Cheap smartphones cost less than $30, so such a board shouldn't the very expensive, and it could save a lot of development time and costs for bespoke hardware design when building devices that need a touchscreen and WiFi (think a fancy washing machine or fridge).

You might be interested in this: https://m5stack.com/ (not affiliated)

That’s actually a really slick design!

I have severe astigmatism, meaning I see double without glasses. Seven segment displays are hopeless without my glasses on.

This is why I prefer analog clock faces, and an analog watch. I can read those just fine without my glasses on. I never liked digital volt-ohm meters, either, preferring an old fashioned analog one.

If you want to use 1970's technology for that retro look, use nixie tubes.

You should wear glasses constantly both to be safer towards yourself and towards others and to also reduce the cognitive load you experience.

Interesting, had never heard of nixie tubes before, definitely neat.


Do you dislike digital speedometers?

Yes. Although I always wear glasses when driving, with an analog speedo I can see it and know the speed without needing to directly look and focus on it.

Boeing recognized this on their flight deck gauges. A gauge would use both a needle (digitally rendered) and a numeric value, reaping the benefits of both. They went even further - the gauge display would be green, amber, or red depending on good, caution, bad. This makes it real easy to check the instruments quickly - all green => everything is good.

Digital speedometers are horrible. A constantly-changing digit is so much less pleasant to look at than a smoothly-moving needle.

A changing digit display also gives no feel for how fast it is changing, which can be just as important as the value. (The same goes for my volt-ohm meter.)

And indeed in e.g. avionics one often sees pixel displays, even in older hardware, where it's AFAIK often small LED matrices (assuming its new enough to have displays, and not using gauges). Random example via image search: https://www.comtronic-schoenau.de/en/solutions/afficheurs-le...

Sort-of related maybe, a special font designed for aircraft cockpit screens: https://news.ycombinator.com/item?id=26373751

Yeah, the HDSP-2132 is a common display component in military hardware. I'm sure that readability figures into it (especially for NVGs), but so does hardiness. My hopes of an 80x24 dumb terminal were dashed when I finally learned the unit replacement price... I don't need a rad hardened serial display that bad.

I've been looking into those Avago/Broadcom dot matrix displays recently. I don't think it'd be too hard to recreate something similar with tiny SMD LEDs on a PCB and a slightly intelligent driver board, for much cheaper. Would there be a market for something like that?

Only if you can get them listed at Digi-Key/Mouser/etc. Even if it's under a "maker" brand like Adafruit, Sparkfun, etc. Anyone who can replacement TIL311 and friends displays in there, and advertise them, will sell quite a few.

They're not hard to make, but they're the sort of thing that no one really wants to make in-house, since they're a means to an end rather than an end in themselves. Great thing to sell, though!

It has been done a few times, but I doubt there would be any kind of market for it. https://hackaday.io/project/85515-redotmatrix

Electroluminescent glass panels on the other hand... https://www.youtube.com/watch?v=Z2o_Sp2-aBo

That redot project seems like perfect candidate for a kickstarter

In grad school I taught a electronics lab class where the students would burn through a bunch of HDSP-0772 every semester. I thought about coming up with a replacement, but at the time you could still get them for a somewhat reasonable price (like $10 or $20?).

At today's price of $50 it might be cost effective to do as you say, but I imagine there's not a huge market.

That would be roughly a 0.75mm pitch.

Not rad hardened, and not quite 0.75mm, but you can buy a 1.25mm pitch LED matrix panel for a relatively low cost. It's an RGB panel instead of monochrome for the HDSP-2132.


Unless I'm misreading the page, it is claiming a resolution of 104x78. So that means 8,112 LEDs.

The HDSP-2132 is made of 8 character elements, each element has a resolution of 5x7. So a 80x24 terminal would have a resolution of 400x168... 67,200 LEDs.

So about a magnitude off, even before ignoring the built in character spacing :)

I wasn't suggesting one of them would settle the matter. I was suggesting it would be cheaper to use them instead. They are made to be grouped together, and daisy chained to a single controller.

I believe each module would be 160 x 120 dots (19,200 dots total) though, based on module size and LED pitch. I'm unclear on where you got 104x78.

> I'm unclear on where you got 104x78.


    Module  Res     Size
    P1.923  104*78  200*150
I went ahead and counted... 128x96 (12,288 LEDs)


It's just bad pictures and errant info.

Here's a better link: https://www.aliexpress.com/item/32965783207.html

It's 160x120, as I mentioned for a 200x150mm P1.25 led module. You just divide the width/height by the pitch to see the number of dots.

As far as I can tell, 1.25mm is the smallest pitch available for these modules.

lol, I marked every tenth led manually and counted twice - the number is correct, the marketplace is wrong.



You counted dots on a picture of something other than a "200mmx150mm P1.25 led module". Pitch is the distance between the center of each LED.

1.25mm x 160 = 200mm

1.25mm x 120 = 150mm

Is over $100 that low cost? Surely there could be better prices.

19,200 RGB LEDs plus all the mux chips adds up, but these signage modules are significantly cheaper than anything else with discrete separate LEDs.

Years ago I read about a scam where people were tampering 7-segment displays in gas pump stations (e.g. by altering one segment a "9" will be displayed as a "5", an "8" as a "6" and so on), so you got less gas than what you paid for.

This worked at least in two ways:

- By showing the wrong price/quantity (1.980€/l would become 1.560€/l)

- By showing the wrong quantity dispensed (39.8 liters would become 35.6)

I guess the lesson here is to never trust a 7-segment display until you can confirm all segments are working.

My microwave for some reason uses left-aligned "1" digits and I've always hated it, but after reading this, I choose to believe that it was done intentionally in order to reduce confusion from stuck-on elements.

(But I doubt it.)


It's a Thermador.

I find it very strange that this paper ignores the existence of the 16-segment display.

I had a hell of a hard time sourcing them a few years back at anything remotely resembling decent prices.

Their cost exceeded that of using wristwatch lcd displays (the kind with fixed patterns) and the results there were far superior. The complexity of a 16-segment display is probably slightly lower, but the runtime power savings made the LCD option more lucrative.

Or any of the other non 7 or 8 segment displays.

Make clearer, monospaced, numeric fonts. Please do NOT use variably placed and sized numbers. IMO, that makes it extremely difficult to read things; and far worse to compare them. I do not need or want numerical calligraphy.

Edit: Within a fixed-width space, however, variable _weight_ numbers might work well. A very heavy 9, an extremely light 0.

Yeah, 14-segment displays aren't much more expensive but you get alphanumeric. :)


This reminds me of the time I looked over at my friend's stereo in his room at uni. "Why's it saying 'nod 15c'?" I enquired, utterly genuinely?

Anecdotal counterpoint - 7 Segs can provide a hard to beat minimalist esthetic to a design, to me it is one of the major appeals o the KORG microKORG synthesizer. (not really arguing with the point of the paper, but design also can matters)

Every digital clock in the store is 7-segment.

Darn. And I just used one in my project, too.

Site uses a self-signed cert and HTTPSEverywhere gets mad.

A bit of a clickbait title but good list of guidelines

7-seg displays are usually used because of cost and constraints, not because they're "good". But for something like a microwave oven they're just fine!

To save you a click the reasons are:

* Some digits can be easily mistaken (e.g. 8 and 0). * Sometimes the contrast between lit and unlit segments is too low. * If you turn the display upside down it might still look valid (pretty sure this was a Jonathon Creek plot). * You can't tell if a segment is off or broken (e.g. an 8 with a broken middle would be indistinguishable from 0).

Honestly not that convincing. I mean they're not exactly wrong, but they're not really fatal either, and anyway nobody uses 7-segment displays because they think they're the best displays possible for the job.

However for anyone using a 7-segment display this should be required reading! It has lots of good design advice.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact