EEVblog #298 - Dave's Decade Digit Display - USB Supply Part 3

[george graves]

I totally forgot about those.  They would bake an excellent display for a PS - but I think dave is going for something a bit more minimal.

[EEVblog]

Yes, I want something pretty minimal I think.
The other battery powered PSU is a better bet for a big graphical display if desired.
Do those OLEDs work outdoors?
(think Agilent OLED meter)

I found a source for low-ish cost 5W isolated converter (still much more expensive than the 2W though), and am now considering a 1A upgrade option, but the cost is rapidly increasing on this project way above what was originally intended.
I know everyone says, "I'll gladly pay X extra for X feature", but it's getting out of hand for such an ultimately fairly limited USB power supply.
So I think a better display and the extra power capability are mutually exclusive. I need the reign in the cost whilst trying to give maximum capability.

Dave.

[caroper]

The original spec was fine.


a) Low Cost
b) Isolated


if it is isolated then 2 in series/parallel  == More current || More Voltage || Split Rail


PC control / display, whilst nice to have is a pure luxury driven by the fact that it is connected to USB, I think the ability to use it with a USB Charger is more useful.


Personally I think any display is overkill if it pushes the price up, it is a field PSU, think battery substitute, use a DMM to set voltage and current and have LED's to indicate status.


To be honest I am not sure where you make the split between the USB uSupply and the battery uSupply but the discussions seam to be merging into one.


You could even start adding the Reference capability that Richards Design is heading towards, but that again is going to far off track (for your project that is, I do like Richards design too).


Follow the advise you gave in today's walkabout, keep the cost down, the spec reasonable and meet the functional goals.


But above all else, keep up the Dialogue, it is entertaining and educational.


Cheers
Chris

[ejeffrey]

So I think a better display and the extra power capability are mutually exclusive. I need the reign in the cost whilst trying to give maximum capability.

I think a better UI is more important than more power.  A device that is inconvenient to use will never be used.  That could be either in the form of an LCD display or an isolated computer interface.  The computer interface provides more features (multiple displays, logging), but is still somewhat inconvenient and can't be used when powered from a cell phone charger.  Still, I would be probably be willing to fall back on the bargraph display as a last-resort option, as long as the computer interface is available if needed.  The LCD provides less functionality, but it always works.  The LCD also requires less development effort on the software side...

[tom66]

Yes, I want something pretty minimal I think.
The other battery powered PSU is a better bet for a big graphical display if desired.
Do those OLEDs work outdoors?
(think Agilent OLED meter)

I found a source for low-ish cost 5W isolated converter (still much more expensive than the 2W though), and am now considering a 1A upgrade option, but the cost is rapidly increasing on this project way above what was originally intended.
I know everyone says, "I'll gladly pay X extra for X feature", but it's getting out of hand for such an ultimately fairly limited USB power supply.
So I think a better display and the extra power capability are mutually exclusive. I need the reign in the cost whilst trying to give maximum capability.

Dave.

There's still the option of designing your own; I know you don't want to mess around with it, but there's a lot to be considered in such a design (so great for beginners) and it will ultimately cost far less than a DC-DC module. You can then also add a tracking pre-regulator automatically to it because the feedback control is freed up. If you use a decent LC output filter, you can make a truly variable ~0.5V - 10V switching PSU with ~10mVp-p noise - no need for an LT3080 regulator. 10mVp-p is likely fine for the target market - I imagine small MCU boards being a common load. (You'll still need a separate isolated 5V though if you go for this; but a 0.5W DC-DC converter is a cheap and readily available component and that could easily power an OLED/LED display and micro.)

[amspire]

There's still the option of designing your own; I know you don't want to mess around with it, but there's a lot to be considered in such a design (so great for beginners) and it will ultimately cost far less than a DC-DC module.

I started down this track with a DC-DC converter for my General Purpose power supply design, and I found myself enormously frustrated. For a start, it is almost impossible to find a source of the right cores with the right material, right gap, right former and right mounting hardware at a decent price for an open source-type design. The builder would also need to get the right gauges of enameled copper wire, the right tape, and then if they do not build the transformer perfectly, it can ruin both the performance and the safety. A poorly constructed switching transformer can easily result in the supply just destroying itself after a few minutes.

The amount of design involved in a  good isolated supply can be an order of magnitude more then a boost or buck converter. When I have done it in the past, it is not hard to end up with about 30 different transformer designs before I get an optimized design I am happy with. An isolated switching DC-DC converter usually works on the bench the first time you power it up, but to get the right power/input/output voltage ranges, the right efficiency, the right noise performance, the right level of switching transients due to the winding leakage inductance, etc is a job that can take weeks or more.

Even more frustrating is seeing the modern core types that could give a much smaller, and much better shielded transformer, except the cores are unavailable if you are not ordering quantities well into the thousands per month. In many cases, the moderns cores require custom pre-formed windings made from flat rectangular-profile enameled copper wire. Impossible for the hobbyist.

All this means that if you try and design an open source isolated DC-DC converter, you are likely to end up in the transformer manufacturing and supply business. As there is safety isolation involved, this can easily be a task you want to avoid.

If you go down the track of using an off-the-shelf transformer, the reliable supply of cheap transformers is pitiful,  they are usually very poorly specified (so you have to get them and test them to see if they may work as required) and never exactly what you need. I found that to get the performance I needed, instead of going for a custom 10mm core design, I had to choose a 25mm footprint transformer that was way to big for any compact case. You can be lucky, but usually not.

The DC-DC converter module is just a more practical way to go, and for most people, it ends up being a cheaper option.

Richard.

[EEVblog]

The DC-DC converter module is just a more practical way to go, and for most people, it ends up being a cheaper option.

I was not silly enough to consider anything else!
The modules are by no means perfect, but you simply work around their limitations.

Dave.

[ftransform]

how about you control the voltage by adjusting the brightness and blinking a single LED. You could have a little printed reference table to judge what voltage it is outputting and how fast it is blinking. I just saved you a $, I expect 50% royalty. MAYBE a multicolored LED if you wanna get fancy but jesus we are not made of money man.

I was going to suggest a little speaker to read the voltage out loud but that might be too expensive. You could corner the blind market.

[rbk17c]

I have now finished a one off PCB with 26, 1206 SMD leds arranged as:
01
01234567890 ss
01234567890 ss
(a 200 count display)

Its 2 sided, 100X39mm PCB, in a charlieplexer arrangement (28 leds, using only 6 cpu-pins).
while it works, and the display is quite stable, I must confess that it is extremely hard to read.
I even added some 'back light' (switch on all Leds for something like 1/100 of the time), AND changed the colors, I still find myself counting.:-(
This is almost as geeky and hard to use as a BCD clock. - Sure - after 1/2year of frying your projects, you WILL learn to read it fast and precise, but...
I might still be saved by some clever front panel, but for now...

/holger






 

[bookaboo]

I think it's coming down to either:
Nokia 5110 display 84x48


or a standard 8x2 text display:


The 8x2 text display is about half the cost of the Nokia one.

Snap poll please!

Dave.

Wow the nokia one wins on so many levels.

- It will look good so more sales (if thats the goal)
- You can do whatever you want with bargraphs etc,
- Fantastic learning for everyone

[MickM]

Hi;   
   My vote is for the Nokia display.
Adafruit has them back in stock.

Mick M

[calin]

I reckon I did not had the time to read the whole thread so if this idea came up please disregard.

This is an "USB power supply" right? supposed to be hooked by a laptop or PC. So ... you have one BIG DISPLAY right there. Having a device implement USB is not that hard and does not require a super hefty MCU; and a simple application that sets and displays the voltage and amps is theoretically "free". Plus you can have adjust both ways, made from the PC side and/or using keys/encoder etc on the supply itself.

Sincerely, I find this solution while not as fancy as a LCD display cheaper and practical. Plus I am 100% sure that USB communication will not use up any significant power.

Just a thought.

And yes, Dave no need to get it fancy; isolated and normal power. Display ca be made of a nice app on the PC and done.

[Rufus]

Been said before, No, doesn't work; muxing is optically equivalent to reducing the current.
Want to keep brightness, then you must increase the current.

This is not true at all.  A muxed display will look considerably brighter than a steady-state display of the same power. 

Don Klipstein http://donklipstein.com/ledp.html doesn't think so.

  I have done lots of experimentation with 555 oscillators and various LEDs and solar cells in various room lighting conditions. Regardless of frequency and duty cycle, a pulsed LED that gives the same solar cell response as a continuously operated one of the same model will almost always match the visually apparent brightness of the continuously operated one. I have sometimes seen slight gains - less than 10 percent when I get any gain at all - in apparent brightness if the pulse frequency is barely fast enough to make the LED not visibly flicker. This frequency is usually around 60 Hz.

I haven't seen anything but an ancient HP application note claiming otherwise. LED efficiency usually drops off at higher currents and I^2R losses mean pulsing the same average current results in higher die temperature.

[SteveyG]

Been said before, No, doesn't work; muxing is optically equivalent to reducing the current.
Want to keep brightness, then you must increase the current.

This is not true at all.  A muxed display will look considerably brighter than a steady-state display of the same power. 

Don Klipstein http://donklipstein.com/ledp.html doesn't think so.

Who the F is Don Klipstein and why should we care about his opinion? There's lots of literature out there showing how the eye responds to certain frequencies differently, which has the effect of making an LED driven at a low duty cycle at these frequencies to appear brighter than an LED driven with an equivalent average current. The wavelength of the LED is also important. Just because 'Don Klipstein' dicked around with a 555 timer and an LED doesn't make it gospel.

I haven't seen anything but an ancient HP application note claiming otherwise. LED efficiency usually drops off at higher currents and I^2R losses mean pulsing the same average current results in higher die temperature.

I²R losses on a non ohmic conductor?



You can get 3 digit 7 seg displays for 38 pence from Rapid in the UK. These would be ideal IMO.

[george graves]

Here's a nice set up if you wanted to keep the design all pick and place-able.




Taken from: http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=4436&p=43546#p43520

[Zad]

The issue isn't the cost of a 7-segment LED array, but the current. Imagine 3.88V at 88.88mA. That is 96 illuminated LEDs. Even at 1mA each, it is a lot of power overhead.

[poptones]

I'm new here so my opinion counts squat, but I like the idea of the PC enabled power supply. Why not make it usb enabled? Not only does this make it easy to handle the settings and display, but you could even use software to operate it as a data logger! Make a circuit and see how it performs over a range of voltages easy peasy. To me, this would make it considerably more valuable than a small box with some numbers and buttons that just happens to use my very expensive pc as a mere source of current.

[McMonster]

As someone probably mentioned already, the problem would be in carrying the signals over the isolation barrier. There are chips that can do that, but they're expensive.

[poptones]

OK, what about this? Rememebr the old BCD thumbwheels? THese provide a readout and provide a means to set values. So what about making a modern LED version of the thumbwheel? Use the strings of LEDs, but provide an up and down button for each decade. Then you're not just turning a knob and counting the dots in your head - if you want to go up by a tenth of a volt, you press the tenth of a volt setting. No need to count as your brain instantly associates the button press with the led change to be a tenth of a volt. If you want to go up a volt, go up a volt. etc. Put white lines on the panel to give quick association with certain common levels like 3v, 5v, 12v, 15v, 24v etc.

[sonicj]

source: decadecounter.com

[andyturk]

@sonicj: I loved using those as a kid, and drove them with 7448s. Even got pretty good at reading hex values using the weird segment combinations created by the 7448. 

[Zad]

I had to post this. Your photo reminded me and I had to go and look for them! I bought them when I was a kid, in a box of mixed 7-segment displays from Tandy. This would be around 1980 I think. They were great for messing around on the breadboard. Good times!

[EEVblog]

I probably still have some of those in a draw somewhere...

Dave.

[poptones]

LOL gonna go retro? Make a version with nixies for those who want some nostalgia with their USB power supply?

Oh, I do have a question: regarding the current capabilities of this thing - why not have it battery powered and charged by usb? One could run much more instantaneous current, you'd have a better power budget, and it would generally be a more flexible device - like a battery with adjustable output and current limiting.

[Pilot3514]

Dave,

I give you full marks for creativity but I can not score this a easily usable.

As with so much in design, you have trade-offs.  Power consumption, cost, availability, size, weight, and so on.  On your scale, you balanced everything and came up with this solution.  No one can fault you for that.  Someone else will weigh the factor different and find a different balance point.

When I see this design, I see something in the computer bag or I have a computer near by.  Therefore, I see computer control as a valuable option.  Therefore, having another display (DDDD) is a viable secondary alternative.  It function as both computer controlled and stand-alone.