Part 3 of the USB Power Supply design. A low cost 4 digit LED display solution = Dave's Decade Digit Display Is it a triumph of novel idea over practicality perhaps?
lacombar So your solution to designing a cheap, low power read display is to design an expensive low power display? I hope you don't have a job in engineering because you won't get too far by just throwing money at every job.
mikerjuk I do have a job in engineering. I found out I am much better working on the engine and internal of the product than working on the user-interface. Engineer tends to be really good at abstracting concepts, finding smart way of doing things, even nit-picking, but UI are of a completely different realm. I am lazy. I want it to be simple. The UI should only rely on quasi-instinctive behavior. This is why Apple is so successful, their product are all about conviviality and a simple [arbitrary] user interface, not a panel of 1000's knobs to adjust the alpha level of the right hand corner of a window. It's the quintessential "plug'n'play" easy to use concept. If you have to read a manual to understand how it works, you failed.
lacombar I agree with you in a broad sense, designers have an artistic flair that few engineers possess, I certainly don't. However, it's pointless designing a beautiful, easy to read interface if it can't be built to meet the requirements set. I fully admit that Dave's display isn't the easiest thing to read, but it does meet the requirements. If someone can design better display for a similar power, cost and size then they need to get posting on the Eevblog forums; I'm sure Dave would be delighted to see it.
I think this is a cool idea and hits many points of good design. Anyone who would actually use this product should appreciate the engineering process...
though impressive, i dont revere this as being to practical, if i had any need for this device i think a cost increase of a few dollars for an easily read display is almost essential, i would have to spend more time checking / working out what the voltage and mA i am about to put through whatever i have hooked up to the device,
Pretty awful for the user. I'm pretty sure that display would piss me off if I had to use it for any real amount of time. I'd rather buy something decent than save a couple bucks.
I love that solution, obviously not on a full bench supply, but for a portable usb bench supply that has to be one of the best ideas for displaying voltage at low cost/current use - just make sure they are aren't the ones that blind you when you turn them on
That's funny while watching this I had a similar idea only using the decade bar graphs but I like this idea. It's easier to customize than my idea. Beautiful work. Love it
You are missing the whole point about the extra cost, size, and power dissipation required in the isolated converter (and following circuitry). It's got nothing to do with being able to extract extra current from the USB port using various means.
Something you could do to make it more readable would be to a plastic overlay over it so that the leds would illuminate through a mask of the number they represent. I can't imagine it would keep it below the cost of 7 segment module though, but would keep the power down.
Another issue could be the lack of strong visual cues to tell you how far up or down the scale you are - you'd need to see the ends to get the position, which may be hard in darker conditions. Maybe an answer would be to have all LEDs on really really dim to give better contextual information.
Absolutely. My point exactly. Toss the display. Anyone buying this device will have a DMM that can be used to set the current. The point is "constant current" so it should be a set-it-and-forget-it device. The only thing I "might" add is a warning LED if the current drifts beyond a predetermined amount. Just as a heads up.
I like this idea a lot. This is an "old school" idea-- In the olden days (before Nixie tubes were invented), this is how digital counters represented their value for each decade. (This was done with incandescent lamps-- as there were no LEDs back then). My take? 1) It's READABLE. 2) It's CHEAP. 3) It's LOW POWER. The verdict? Great job Dave! SHIP IT!
I can't imagine a huge difference between a 7 seg display flickering 4.99-5.01 and this display flickering the same. Yes, it's probably needs real-world prototyping. It's just an idea at this stage.
There's actually a phenomenon due to the way our vision works that allows a pulsed LED to look brighter than an LED on constantly even if the average current is the same.
May be OK for setting values but for measuring I think it will totally suck ass - you'd need to damp it a lot to avoid jitter, making it less useful.. A small cluster of LEDs in seven-seg format would be better. If you go with red you can use series pairs for extra length and brightness without more current. Low-current LEDs behind a red tint filter would draw maybe 8-10mA for 6 digits - not a lot from your 500mA available. You really need to prototype it to see how well it works in practice.
Well, Dave, I think you found a cheap solution, granted. You outdid yourself in lowering the component cost. I do think, however, that the potential user will be happy to pay $2 more for a useful display :) For the sake of educational value and figuring out mass production practicability, it might actually be interesting to consider the self-made 7-segment displays with slots in the board (which then have to be filled with some kind of resin I presume). Pro: More multiplexing control and flat.
I appreciate you sharing your thought process with us. Your reasoning is sound, but I would not buy a power supply with this kind of display. UNLESS..... this allowed a USB dongle type form factor. Even then there are tiny LCD displays that do this nicely.
It's probably mentioned before, but on a USB power supply, I would not expect any knobs or segments at all, but a PC program controlling. Even when its hooked to a wal-wart USB supply, it could have been either pre-programmed or have a life USB as well as a power only USB port to allow for PC control. this would also reduce size and component count, and probably cost
Dave, your led sollution is already a matrix that can display 3 digits (3x3). Just add a few of those low cost leds and make a 12x3 multiplexed led matrix that can display 4 digits... That way you could read it from a few meters distance...
Probably needs to be tested in practice, but I generally like the idea. I would probably still add the zero LED for the most-significant digit. That way any potential "flickering" across the left-most decimal boundary will be easier to detect.
Just curious: The above bar-graph display uses 29 LEDs. Wouldn't a 7-segment LED for 4 numbers + 1 decimal bit use the same number of LEDs? (Assuming you use 1 LED per segment, of course). It may not be the brightest display, but would be more easy to read, wouldn't it?
The way I understand the spec, that 500mA is the maximum you can guarantee to be delivered over the wire from a standard outlet. What actually can be consumed is a factor of the impedance of the load. Since this is a electronics tool it seems reasonable to leave that impedance in the hands of the tools user.
Neat solution. A couple of observations... 1-- do you really need the '0' row? Could save 3 LEDs and a little space. 2-- Most of those SMT LEDs are almost painfully bright when you're looking right at them. You're going to have to spec them carefully, or power them at the lower end of their range.
I really think it's a great idea Dave, not only it's cheap and small but it also makes you double check the voltage on it, awesome! Also I think would be great to make a bigger gab after 5 so it would be more easy to orientate and not count 1,2...all the time. Maybe two color LEDs for both volts and amps on one array?
I like it. Keeping in mind that this is a device designed for electronics people, who shouldn't have any problem interpreting displayed values. The only thing I wonder is how are you going to distinguish between mA and V? Maybe you can market it as new 4D display technology :)
OK, after all the ideas I've read here I must post this one. Use a voice recorder chip & piezo speaker. For each digit record the digits 0-9, the decimal point, miliamps and volts. Then just have it read back the digits. It would be cool as hell to have Dave reading off the readings. When you power it on Dave says "Dave's USB Power Supply." I'd pay an extra $10 or so dollars just for the fun of it.
This display idea was IMHO the classic penny wise/pound foolish trade off. I’d like both nice knobs and a readable display, and would gladly pay for that. The knobs can lay flat on the other side of the board, opposing the banana jacks. All those cost cutting measures are turning a good idea into a piece of junk…
You can put 10 RGB LEDs in a large circle, number them from 0-9, and then assign each colour to each digit, for example red for the most significant digit, blue for least etc.. or you can do 3 circles inside each other with a single colour LEDs. It will be easier to read from a distance and you get maximum of 3 LEDs on at any given time.
Before you get over excited by this, what does your board assembler charge per component placed. Last time I did this, it was between $0.10 and $0.25 per component depending on production size and number of identical parts placed. It made the time I spent ( > 1 hour ) digging through Digikey/Mouser/Newark trying to save $0.001 per 1nF capacitor totally pointless.
Interesting display solution! How about arranging the LEDs in concentric rings, like the hands on an analogue clock, or have each digit group arranged in a circle like the dials on a electricity meter?
There's an equation for any display to obtain the maximum contrast and efficiency based on refresh rate, wavelength, discrimination ratio and a few other parameters that I can't remember. I think it was linked with one of the equations in ISO-9241-3 relating to refresh rates as a function of luminous intensity.
You can limit the current lower than in specs with little loss of light output. I have used 10k resistor for 20mA LED successfully for lower power consumption on 9V. Months on single 6f22.
Cool. Did you consider making an abacus display? Put a 5 on top and four ones below for each column. Saves a few parts and just as readable (with some training).
I don't get it why are people hating it so much. I think it's a pretty neat idea, and I would like to see it in use, and try to read from it. I think it would be pretty easy and intuitive to read after only a few readings (when you get used to it).
I like that design actually, it reminds me of reading a resistor-colour value table. I think you should add indicators to tell you what it's displaying. So a LED next to V, and a LED next to mA, then you could make it toggle once or twice a second. Would be interested in seeing what your driver circuitry will look like though, shift registers or maybe just a micro with lots of pins.
It's an interesting solution. I don't know what i would think of using it, everything new takes some time getting used to. I think i would have preferred the 7-seg in PCB solution. If i remember things right the eye is most sensitive to green light, so maybe a green display would be more visible than red at a lower current.
I actually checked the date to make sure that it's not April fools :-D Seriously: I guess the LCD would be a good, classical way to go. The solution doesn't rock, but I like your lateral thinking. And I like the way you interview your possible customers before something is set.
If you use your independent digit as a single micro led; you will find that if you try to read it at say two feet away; it will go blurry and not readable. To take your concept to the extreme; use one 7 segment led and have a button so you could scan through each digit writhing down each figure on paper and then read the whole answer off the paper. Only making a point not to offend.
I somewhat agree with the complaints of "I'd spend an extra 2 dollar for the digits" but then again, I do think the idea is some what novel and I think that is great. especially considering the manufaturability of this solution. When I think about using a power supply as a voltage source, its really a set it and forget it operation so readability and all of that is really not too big of a deal. Monitoring current is a little more critical but honestly if you need it that bad, use a series DMM
I quite like the idea, and you could drastically lower power consumption by only lighting a single LED in each column, rather than the whole bar. But, as others have said, it may not be as quick to read; you'd have to consider if it's worth saving a bit of cash for that.
Have you considered that the LEDs in those displays you're using are aren't very efficient, and that if you did create your own LED display solution that you could use more efficient LEDs? The LEDs in those displays put out 6.5mcd at 10mA. These green LEDs for example would cost only a little more, but put out 2.5x as much light, so you could run them on less than half the current: LTST-C191KGKT And these red ones put out 4x as much light and cost the same: LTST-C191KRKT
I think you should still go with the slotted PCB style seven seg display. Less LEDs and lowish consumption if you PWM the LEDs so they are bright when selecting values, but then dim/off when nothing is changing. The bar graph idea is novel, but in this case it's all a bit of a faff to read it quickly while using the thing somewhere inconvenient.
The other problem is how the data is displayed, for a beginner, it's not an easy way to view the data as you have to link the LEDs. There could also be the factor of what if one of the LEDs in the left most column burning out (theoretically, even though highly unlikely) the data displayed could make someone think it's currently 3.3v output but could be 13.3v output due to a fault LED. It could cost more in terms of current/power, but you could have a little scrolling LED array display.
I saw ePaper price tags at a tradeshow once. Basically 7seg cut-out with ePaper behind it. See if you can find that. It was targeted at supermarket shelf tags so they could reprice shelves via infrared from a central computer.
I feel there is going to be a problem not just in the display of things, but also in knowing which mode that you're currently viewing. How do you tell when it's powered up for example if it's showing the voltage or current (I know it would most likely show the voltage rather than the current)? If you leave and someone plays with the buttons, do you know if it's showing the current load or the current output voltage? You need some way of telling whether it's showing voltage or current.
Have to agree with (electricstuff) Mike. Clusters of LEDs in a seven segment format would not boost your LED count by much. In fact, if you used a wide "font" you could have the vertical parts be just 1 LED and 2 for the Horizontal, for a total 10 LEDs per digit. Also, is this really going to be able to push 0.01V at 0.1 mA? For some reason that's a much finer gradation than I was expecting. Anyway, this is really fun and instructive. Keep on!
I don't like that the decimal point is in different places depending on the mode (V or I). I think it would lead to people misreading the display more often.
I think lining up the LEDs in seven segment format would be better. The main problem I have with a decade display is that it can't easily be read at a glance or out of the corner of your eye.
This I agree with. Although to me, if you lose the knobs and familiar 7 seg display, you're taking away the consistency between almost every power supply out there (except programmable types). Push buttons and funny bargraphs aren't ideal.
What about building a 7seg display right onto the pcb out of surface mount led's? Then code it to illuminate each led for a split second to draw the numbers and that way only one, maybe two led's are lit at a time. This would cut cost down and reduce the current draw as well. your looking at a total of 21-24 leds.
Also, it could be really hard to read in a low light environment. You get a bit blinded by the LEDs and cant read the numbers next to them. If you used LEDs that put out light sideways they could all have numbers next to them to light up.
I kinda like the binary idea. 5 leds for the whole number up to 31 above each led you screen print 16,8,4,2,1. Then for the fractional part you have 10 leds with the numbers 512,256,128,64,32,16,8,4,2,1 which would read up to .999. 29.999V would be shown as... 11101 1111100111 5.100V 00101 0001100100
Dave, you could just as well form the 7segments out of single LEDs and multiplex them for digits AND segments. It's a bit more work for the µC (its just for measuring and display anyway, so no biggie), but saves pins (compared to the LCD solution), price (compared to the premade 7 segment display AND is easily readable in not-that-perfect light. If your bar display shows 22.22, how the heck are you going to see its not 11.11? Same for 4.44, 5.55 and so on.
Dave, that few cents that you managed to save might shave off some ease of usability as well. Why not stick to the idea of making your own 7-segs using leds BUT instead of just multiplexing the digits, multiplex the segments too (provided of course that the micro manages to blink all of them in ~15ms). That way you'll have current consumption of just one led AND the practical usability of a 7-segment display. Best of both worlds.
Why don't you have it emit a perforated strip of paper with the voltage readings and also get the settings from a similar perforated strip of paper? It would totally give that WWII feeling to the user that too many products lack today.
That's a not-too-bad idea, but it will take some time to learn how to use it. You can actually use any microcontroller you want to control the LCD modules. What the LCD capable controllers do have, is a bunch of shift registers, with one or more common pins, as required by the LCD. What happens, is that the shift register outputs and the common pin are inverted every couple ms. I've tried that before with a status LCD from an old laptop and an ATmega88 in software. It worked out just fine
It will work with "only" 500mA, but if you require more current (which seems unlikely most of the time you would be using a little thing like this) you can get a 1.5A USB wall wort and call it a day.
Cheap but could be greatly improved quite easily. Create a negative image of some numbers, in the same positions as the LEDs, on a sheet of acetate. Place the acetate under the transparent enclosure lid (LEDs would need to be close up to avoid light leaking into the other numbers). Numbers could then be read directly without having to jump to the side to see. A further improvement would be a diffusing matt film between LEDs and acetate.
I think those LCD's have only 1 common pin for all the segments rather than a common pin for each digit. Maybe they do that to save the pins on the module? dunno
I agree with others that a USB micro and using the PC for instrumentation with no physical display would be neat and really useable. It is connected to a USB port after all! How many people are going to accidentally fry something misreading the LED's?
Except I think the brain will intuitively recognise digits faster than a bunch of arbitary dots in an unfamiliar format, so the damping time could probably be set faster with 7-seg.
To me it comes across as a school project, I would like the msb to have a LED to remove any doubt if an LED has failed (however unlikely) if the 2 msb led failed you would assume under 10v Cheap is good but everyone has an extra $1 for a significantly better solution even if they have to go without their can of V that day?
I definitely think it is a novel idea. i would like to see it prototyped though. I, like some others on here, think once the distance from the surface is factored in it might be less readable. definitely go for it though. Cheers!
your led design is very good. but how can we measure the current or voltage and send this datas to the led's and show what is current current or voltage?
1) If you're going to use discrete LEDs, why not lay them out in a 7-segment fashion? 2) Why limit yourself to a micro with a built-in driver when writing your own is just a few hundred LOC? 3) Aren't there 3-digit 7-segment displays wired up in a matrix, that don't need as many pins? If not then did you consider 3x individual 1-digit 7-segments? Three of them could be wired up in a matrix
Sure you are right but I think setting with dedicated buttons will be quicker and better. Also you won't gat problems with new/old OS support. I like controlling devices from PC but only when it is necessary (some scheduled tasks, automatization...). Lab power supply is better with dedicated voltage display and config buttons/knobs...
damn 500 char limit. back to monitoring current, from a distance, I would think looking at a bar graph would be more sematic than some digits bouncing around so we might just need to see it to believe it. Now being able to log current readings via a PC app and a graph, that would be awesome but talk about cost and your goal to isolate this thing... that would be tough. Let's hear more about the supply, whats the topologies that will offer up isolation, step up and step down functions cheap?
I had the same idea as takati68... stick with what you have and turn them off after a time. Blink the decimal point every few seconds to show that it's on. With a push button to wake it up. I really like the idea, but I like the 7 segs better.
That is not a very compelling reason because 500mA is the MINIMUM specification, not the maximum or required spec. I have at least a dozen power supplies with a USB end and > than 500mA.
I think it REALLY needs separate labeling just next to each LED row just not to swipe your finger over it all the time looking is it say 7 or 8. Apart from that it is a kind of OK. As others mentioned, 3-color LEDs could be cool for displaying current and voltage at once, that would be a feature that a normal display doesn't have. But... I'd like an LCD a little more I suppose, just for comfort of use.
uknow u could paint the transparent cover, leaving number shapes over the leds clear. it would slightly improve readability not having to reference the numbers on the side...
If you *really* wanna go cheap, you could do it all in software and let the µController emulate a USB keyboard. Just add a switch labeled 'output on/off', and as soon as the switch is turned on, it prints the values in Notepad or whatever, so you could even use it as an easy data logging solution. (google for 'usb business card' for someone who implemented this)
You can try multiplexing single led in display instead of multiplexing single digit. I will be more readable than table of small leds and also draw a current of one LED. Of course you can make display from LEDs or use off the shelf one.
Design rule #1: never put an engineer in charge of the UI design.
Thanks to prove that rule.
If you have a better solution that is very cheap, low power and easy to read, why not suggest it to Dave?
mikerjuk Put a few more $$ in a proper human readable display. To my knowledge, Dave did not proceed further about this idea.
lacombar
So your solution to designing a cheap, low power read display is to design an expensive low power display? I hope you don't have a job in engineering because you won't get too far by just throwing money at every job.
mikerjuk I do have a job in engineering. I found out I am much better working on the engine and internal of the product than working on the user-interface. Engineer tends to be really good at abstracting concepts, finding smart way of doing things, even nit-picking, but UI are of a completely different realm. I am lazy. I want it to be simple. The UI should only rely on quasi-instinctive behavior. This is why Apple is so successful, their product are all about conviviality and a simple [arbitrary] user interface, not a panel of 1000's knobs to adjust the alpha level of the right hand corner of a window. It's the quintessential "plug'n'play" easy to use concept.
If you have to read a manual to understand how it works, you failed.
lacombar
I agree with you in a broad sense, designers have an artistic flair that few engineers possess, I certainly don't. However, it's pointless designing a beautiful, easy to read interface if it can't be built to meet the requirements set. I fully admit that Dave's display isn't the easiest thing to read, but it does meet the requirements. If someone can design better display for a similar power, cost and size then they need to get posting on the Eevblog forums; I'm sure Dave would be delighted to see it.
I think this is a cool idea and hits many points of good design. Anyone who would actually use this product should appreciate the engineering process...
though impressive, i dont revere this as being to practical, if i had any need for this device i think a cost increase of a few dollars for an easily read display is almost essential, i would have to spend more time checking / working out what the voltage and mA i am about to put through whatever i have hooked up to the device,
Pretty awful for the user. I'm pretty sure that display would piss me off if I had to use it for any real amount of time. I'd rather buy something decent than save a couple bucks.
i should learn for my exam "digital signal processing" but i prefer watching this. eevblog, i like :>
I really like the solution, elegant and funky and shouldn't be too hard to see.
I love that solution, obviously not on a full bench supply, but for a portable usb bench supply that has to be one of the best ideas for displaying voltage at low cost/current use - just make sure they are aren't the ones that blind you when you turn them on
That's funny while watching this I had a similar idea only using the decade bar graphs but I like this idea. It's easier to customize than my idea. Beautiful work. Love it
I like it Dave! I like that there are labels on the silkscreen, it makes it more readable. :)
You are missing the whole point about the extra cost, size, and power dissipation required in the isolated converter (and following circuitry). It's got nothing to do with being able to extract extra current from the USB port using various means.
Something you could do to make it more readable would be to a plastic overlay over it so that the leds would illuminate through a mask of the number they represent. I can't imagine it would keep it below the cost of 7 segment module though, but would keep the power down.
Another issue could be the lack of strong visual cues to tell you how far up or down the scale you are - you'd need to see the ends to get the position, which may be hard in darker conditions. Maybe an answer would be to have all LEDs on really really dim to give better contextual information.
Absolutely. My point exactly. Toss the display. Anyone buying this device will have a DMM that can be used to set the current. The point is "constant current" so it should be a set-it-and-forget-it device. The only thing I "might" add is a warning LED if the current drifts beyond a predetermined amount. Just as a heads up.
I like this idea a lot. This is an "old school" idea-- In the olden days (before Nixie tubes were invented), this is how digital counters represented their value for each decade. (This was done with incandescent lamps-- as there were no LEDs back then). My take?
1) It's READABLE.
2) It's CHEAP.
3) It's LOW POWER.
The verdict? Great job Dave! SHIP IT!
I'll also add that you normally harp on about UI and whatnot, so this must be a windup. ;)
I can't imagine a huge difference between a 7 seg display flickering 4.99-5.01 and this display flickering the same. Yes, it's probably needs real-world prototyping. It's just an idea at this stage.
I like the new design. It makes people think and it costs less both in power and money.
There's actually a phenomenon due to the way our vision works that allows a pulsed LED to look brighter than an LED on constantly even if the average current is the same.
May be OK for setting values but for measuring I think it will totally suck ass - you'd need to damp it a lot to avoid jitter, making it less useful.. A small cluster of LEDs in seven-seg format would be better. If you go with red you can use series pairs for extra length and brightness without more current. Low-current LEDs behind a red tint filter would draw maybe 8-10mA for 6 digits - not a lot from your 500mA available.
You really need to prototype it to see how well it works in practice.
Well, Dave, I think you found a cheap solution, granted. You outdid yourself in lowering the component cost.
I do think, however, that the potential user will be happy to pay $2 more for a useful display :)
For the sake of educational value and figuring out mass production practicability, it might actually be interesting to consider the self-made 7-segment displays with slots in the board (which then have to be filled with some kind of resin I presume). Pro: More multiplexing control and flat.
I appreciate you sharing your thought process with us. Your reasoning is sound, but I would not buy a power supply with this kind of display. UNLESS..... this allowed a USB dongle type form factor. Even then there are tiny LCD displays that do this nicely.
"one batch wonder!" lol
btw: I like the decade display! It's like my tokyoflash watches ;-)
It's probably mentioned before, but on a USB power supply, I would not expect any knobs or segments at all, but a PC program controlling. Even when its hooked to a wal-wart USB supply, it could have been either pre-programmed or have a life USB as well as a power only USB port to allow for PC control.
this would also reduce size and component count, and probably cost
Dave, your led sollution is already a matrix that can display 3 digits (3x3). Just add a few of those low cost leds and make a 12x3 multiplexed led matrix that can display 4 digits... That way you could read it from a few meters distance...
Probably needs to be tested in practice, but I generally like the idea. I would probably still add the zero LED for the most-significant digit. That way any potential "flickering" across the left-most decimal boundary will be easier to detect.
It is still easy to read. I like the solution. Usable and nice.
Just curious: The above bar-graph display uses 29 LEDs. Wouldn't a 7-segment LED for 4 numbers + 1 decimal bit use the same number of LEDs? (Assuming you use 1 LED per segment, of course). It may not be the brightest display, but would be more easy to read, wouldn't it?
The way I understand the spec, that 500mA is the maximum you can guarantee to be delivered over the wire from a standard outlet. What actually can be consumed is a factor of the impedance of the load. Since this is a electronics tool it seems reasonable to leave that impedance in the hands of the tools user.
I think the led bar dispay is a good idea. Good one !!! Like!!!
Neat solution. A couple of observations...
1-- do you really need the '0' row? Could save 3 LEDs and a little space.
2-- Most of those SMT LEDs are almost painfully bright when you're looking right at them. You're going to have to spec them carefully, or power them at the lower end of their range.
I really think it's a great idea Dave, not only it's cheap and small but it also makes you double check the voltage on it, awesome! Also I think would be great to make a bigger gab after 5 so it would be more easy to orientate and not count 1,2...all the time. Maybe two color LEDs for both volts and amps on one array?
I like it.
Keeping in mind that this is a device designed for electronics people, who shouldn't have any problem interpreting displayed values.
The only thing I wonder is how are you going to distinguish between mA and V?
Maybe you can market it as new 4D display technology :)
OK, after all the ideas I've read here I must post this one.
Use a voice recorder chip & piezo speaker. For each digit record the digits 0-9, the decimal point, miliamps and volts. Then just have it read back the digits.
It would be cool as hell to have Dave reading off the readings. When you power it on Dave says "Dave's USB Power Supply." I'd pay an extra $10 or so dollars just for the fun of it.
This display idea was IMHO the classic penny wise/pound foolish trade off. I’d like both nice knobs and a readable display, and would gladly pay for that. The knobs can lay flat on the other side of the board, opposing the banana jacks. All those cost cutting measures are turning a good idea into a piece of junk…
You can put 10 RGB LEDs in a large circle, number them from 0-9, and then assign each colour to each digit, for example red for the most significant digit, blue for least etc.. or you can do 3 circles inside each other with a single colour LEDs. It will be easier to read from a distance and you get maximum of 3 LEDs on at any given time.
Before you get over excited by this, what does your board assembler charge per component placed. Last time I did this, it was between $0.10 and $0.25 per component depending on production size and number of identical parts placed. It made the time I spent ( > 1 hour ) digging through Digikey/Mouser/Newark trying to save $0.001 per 1nF capacitor totally pointless.
Interesting display solution! How about arranging the LEDs in concentric rings, like the hands on an analogue clock, or have each digit group arranged in a circle like the dials on a electricity meter?
There's an equation for any display to obtain the maximum contrast and efficiency based on refresh rate, wavelength, discrimination ratio and a few other parameters that I can't remember. I think it was linked with one of the equations in ISO-9241-3 relating to refresh rates as a function of luminous intensity.
You can limit the current lower than in specs with little loss of light output. I have used 10k resistor for 20mA LED successfully for lower power consumption on 9V. Months on single 6f22.
Cool. Did you consider making an abacus display? Put a 5 on top and four ones below for each column. Saves a few parts and just as readable (with some training).
I don't get it why are people hating it so much. I think it's a pretty neat idea, and I would like to see it in use, and try to read from it. I think it would be pretty easy and intuitive to read after only a few readings (when you get used to it).
I like that design actually, it reminds me of reading a resistor-colour value table. I think you should add indicators to tell you what it's displaying. So a LED next to V, and a LED next to mA, then you could make it toggle once or twice a second.
Would be interested in seeing what your driver circuitry will look like though, shift registers or maybe just a micro with lots of pins.
It's an interesting solution. I don't know what i would think of using it, everything new takes some time getting used to. I think i would have preferred the 7-seg in PCB solution.
If i remember things right the eye is most sensitive to green light, so maybe a green display would be more visible than red at a lower current.
I actually checked the date to make sure that it's not April fools :-D
Seriously: I guess the LCD would be a good, classical way to go. The solution doesn't rock, but I like your lateral thinking. And I like the way you interview your possible customers before something is set.
Not sure if it is "phenomenon" but what I heard it is just because LED driven at higher current has higher efficency.
If you use your independent digit as a single micro led; you will find that if you try to read it at say two feet away; it will go blurry and not readable. To take your concept to the extreme; use one 7 segment led and have a button so you could scan through each digit writhing down each figure on paper and then read the whole answer off the paper. Only making a point not to offend.
I somewhat agree with the complaints of "I'd spend an extra 2 dollar for the digits" but then again, I do think the idea is some what novel and I think that is great. especially considering the manufaturability of this solution. When I think about using a power supply as a voltage source, its really a set it and forget it operation so readability and all of that is really not too big of a deal. Monitoring current is a little more critical but honestly if you need it that bad, use a series DMM
I think it is a cool idea, works well enough and cheap!
I quite like the idea, and you could drastically lower power consumption by only lighting a single LED in each column, rather than the whole bar. But, as others have said, it may not be as quick to read; you'd have to consider if it's worth saving a bit of cash for that.
10:43pm, at customers site.. I'd rather have a display I can glance at and 'know'. jmho.
Thanks for the GREAT vblog!
By the way... Your Idea is really awesome!
Have you considered that the LEDs in those displays you're using are aren't very efficient, and that if you did create your own LED display solution that you could use more efficient LEDs? The LEDs in those displays put out 6.5mcd at 10mA. These green LEDs for example would cost only a little more, but put out 2.5x as much light, so you could run them on less than half the current: LTST-C191KGKT And these red ones put out 4x as much light and cost the same: LTST-C191KRKT
I think you should still go with the slotted PCB style seven seg display. Less LEDs and lowish consumption if you PWM the LEDs so they are bright when selecting values, but then dim/off when nothing is changing.
The bar graph idea is novel, but in this case it's all a bit of a faff to read it quickly while using the thing somewhere inconvenient.
The other problem is how the data is displayed, for a beginner, it's not an easy way to view the data as you have to link the LEDs. There could also be the factor of what if one of the LEDs in the left most column burning out (theoretically, even though highly unlikely) the data displayed could make someone think it's currently 3.3v output but could be 13.3v output due to a fault LED. It could cost more in terms of current/power, but you could have a little scrolling LED array display.
Love the display Idea. How about a dual color led and show both voltage and curant at the same time?
I saw ePaper price tags at a tradeshow once. Basically 7seg cut-out with ePaper behind it. See if you can find that. It was targeted at supermarket shelf tags so they could reprice shelves via infrared from a central computer.
I feel there is going to be a problem not just in the display of things, but also in knowing which mode that you're currently viewing. How do you tell when it's powered up for example if it's showing the voltage or current (I know it would most likely show the voltage rather than the current)? If you leave and someone plays with the buttons, do you know if it's showing the current load or the current output voltage? You need some way of telling whether it's showing voltage or current.
Have to agree with (electricstuff) Mike. Clusters of LEDs in a seven segment format would not boost your LED count by much. In fact, if you used a wide "font" you could have the vertical parts be just 1 LED and 2 for the Horizontal, for a total 10 LEDs per digit. Also, is this really going to be able to push 0.01V at 0.1 mA? For some reason that's a much finer gradation than I was expecting. Anyway, this is really fun and instructive. Keep on!
if you are going to mount an acrylic top anyway, why not etch/engrave the 7-segments into that and make it backlit?
I don't like that the decimal point is in different places depending on the mode (V or I). I think it would lead to people misreading the display more often.
I think lining up the LEDs in seven segment format would be better. The main problem I have with a decade display is that it can't easily be read at a glance or out of the corner of your eye.
Couldn't agree more. This isn't good UX. What about going with your matrix idea, but then shutting the display off after adjustments are made?
This I agree with. Although to me, if you lose the knobs and familiar 7 seg display, you're taking away the consistency between almost every power supply out there (except programmable types). Push buttons and funny bargraphs aren't ideal.
What about building a 7seg display right onto the pcb out of surface mount led's? Then code it to illuminate each led for a split second to draw the numbers and that way only one, maybe two led's are lit at a time. This would cut cost down and reduce the current draw as well. your looking at a total of 21-24 leds.
Also, it could be really hard to read in a low light environment. You get a bit blinded by the LEDs and cant read the numbers next to them.
If you used LEDs that put out light sideways they could all have numbers next to them to light up.
I kinda like the binary idea. 5 leds for the whole number up to 31 above each led you screen print 16,8,4,2,1. Then for the fractional part you have 10 leds with the numbers 512,256,128,64,32,16,8,4,2,1 which would read up to .999.
29.999V would be shown as...
11101
1111100111
5.100V
00101
0001100100
Dave, you could just as well form the 7segments out of single LEDs and multiplex them for digits AND segments. It's a bit more work for the µC (its just for measuring and display anyway, so no biggie), but saves pins (compared to the LCD solution), price (compared to the premade 7 segment display AND is easily readable in not-that-perfect light. If your bar display shows 22.22, how the heck are you going to see its not 11.11? Same for 4.44, 5.55 and so on.
Dave, that few cents that you managed to save might shave off some ease of usability as well. Why not stick to the idea of making your own 7-segs using leds BUT instead of just multiplexing the digits, multiplex the segments too (provided of course that the micro manages to blink all of them in ~15ms).
That way you'll have current consumption of just one led AND the practical usability of a 7-segment display. Best of both worlds.
Why don't you have it emit a perforated strip of paper with the voltage readings and also get the settings from a similar perforated strip of paper? It would totally give that WWII feeling to the user that too many products lack today.
What about using the rectangle led's (as used in LED bar graphs) laid out in a 7 segment configuration
It's kind of old school but could work
That's a not-too-bad idea, but it will take some time to learn how to use it.
You can actually use any microcontroller you want to control the LCD modules. What the LCD capable controllers do have, is a bunch of shift registers, with one or more common pins, as required by the LCD. What happens, is that the shift register outputs and the common pin are inverted every couple ms. I've tried that before with a status LCD from an old laptop and an ATmega88 in software. It worked out just fine
It will work with "only" 500mA, but if you require more current (which seems unlikely most of the time you would be using a little thing like this) you can get a 1.5A USB wall wort and call it a day.
Cheap but could be greatly improved quite easily. Create a negative image of some numbers, in the same positions as the LEDs, on a sheet of acetate. Place the acetate under the transparent enclosure lid (LEDs would need to be close up to avoid light leaking into the other numbers). Numbers could then be read directly without having to jump to the side to see. A further improvement would be a diffusing matt film between LEDs and acetate.
I think those LCD's have only 1 common pin for all the segments rather than a common pin for each digit. Maybe they do that to save the pins on the module? dunno
I agree with others that a USB micro and using the PC for instrumentation with no physical display would be neat and really useable. It is connected to a USB port after all! How many people are going to accidentally fry something misreading the LED's?
Except I think the brain will intuitively recognise digits faster than a bunch of arbitary dots in an unfamiliar format, so the damping time could probably be set faster with 7-seg.
To me it comes across as a school project, I would like the msb to have a LED to remove any doubt if an LED has failed (however unlikely) if the 2 msb led failed you would assume under 10v Cheap is good but everyone has an extra $1 for a significantly better solution even if they have to go without their can of V that day?
He also mentions the current usage. Remember that this is a USB power supply and there's little amp to spend.
If you need to save some mA and pins but keep the 7seg display, why not use some charlie-plexing?
I definitely think it is a novel idea. i would like to see it prototyped though. I, like some others on here, think once the distance from the surface is factored in it might be less readable. definitely go for it though. Cheers!
your led design is very good.
but how can we measure the current or voltage and send this datas to the led's and show what is current current or voltage?
Oh and Dave, can you build it on the breadboard so we can see working ?
I would prefer LCD as well.
The reason for saving ~40 mA is because it's 8% of your 500 mA. So your compromising your maximum power output of the PSU.
1) If you're going to use discrete LEDs, why not lay them out in a 7-segment fashion?
2) Why limit yourself to a micro with a built-in driver when writing your own is just a few hundred LOC?
3) Aren't there 3-digit 7-segment displays wired up in a matrix, that don't need as many pins? If not then did you consider 3x individual 1-digit 7-segments? Three of them could be wired up in a matrix
Great now if I buy it every time I use it I will spend 30 minutes looking at light "moving" up and down the columns as I turn the knobs :D
I love your solution, but I was wondering why you couldn't mux the LCD?
Sure you are right but I think setting with dedicated buttons will be quicker and better. Also you won't gat problems with new/old OS support. I like controlling devices from PC but only when it is necessary (some scheduled tasks, automatization...). Lab power supply is better with dedicated voltage display and config buttons/knobs...
damn 500 char limit. back to monitoring current, from a distance, I would think looking at a bar graph would be more sematic than some digits bouncing around so we might just need to see it to believe it.
Now being able to log current readings via a PC app and a graph, that would be awesome but talk about cost and your goal to isolate this thing... that would be tough. Let's hear more about the supply, whats the topologies that will offer up isolation, step up and step down functions cheap?
I had the same idea as takati68... stick with what you have and turn them off after a time. Blink the decimal point every few seconds to show that it's on. With a push button to wake it up. I really like the idea, but I like the 7 segs better.
That is not a very compelling reason because 500mA is the MINIMUM specification, not the maximum or required spec. I have at least a dozen power supplies with a USB end and > than 500mA.
is 290.9mA enough?
you could get 400mA @3.3V with a 2W DC-DC converter
also you may add a led to indicate if you are displaying Voltage or Current
I think it REALLY needs separate labeling just next to each LED row just not to swipe your finger over it all the time looking is it say 7 or 8. Apart from that it is a kind of OK. As others mentioned, 3-color LEDs could be cool for displaying current and voltage at once, that would be a feature that a normal display doesn't have.
But... I'd like an LCD a little more I suppose, just for comfort of use.
uknow u could paint the transparent cover, leaving number shapes over the leds clear.
it would slightly improve readability not having to reference the numbers on the side...
I can't see anything better than 7-seg led display.
If you *really* wanna go cheap, you could do it all in software and let the µController emulate a USB keyboard. Just add a switch labeled 'output on/off', and as soon as the switch is turned on, it prints the values in Notepad or whatever, so you could even use it as an easy data logging solution. (google for 'usb business card' for someone who implemented this)
A set of RGB LEDs displaying it as a resistor colour code? I do like the idea though. Keeps it thin enough to fit in a laptop bag too!
Long term availability won't be a problem for 16x2 displays, they're all over the place.
Why not display the voltage and current on the PC through the USB port.
You can try multiplexing single led in display instead of multiplexing single digit. I will be more readable than table of small leds and also draw a current of one LED. Of course you can make display from LEDs or use off the shelf one.
Interesting and cute idea - you can later design companion device that will translate the decade display into human readable format:p