uSupply Keypad Design

[EEVblog]

As the poll says, what would you prefer for the input method on the new uSupply.
Yes, there is deliberately NOT an option for knobs, it's keys only, I won't explain why, it just is.

Separate up/down Voltage and Current control arrows with velocity control for holding them down, or a numeric keypad were you press say 3.3 and then the V or I key.

What say you.

[CopperCone]

Should be both like wavetek equipment. Especially if you can jump to different preset voltages for ease like 3.3 5 12 15

[BrianHG]

I require the step up/down when first time powering up a new prototype.  When I like to inch up the current so the supply will up the output voltage until the consumption current is cleared and the supply switches from CC to CV mode.

But, a lot of times I need to set a voltage and current value immediately.

So, I cannot answer your poll, since I need both numeric entry to set my position, and a single step up/down arrows which by default move the voltage up and down by default when in CV mode, but if the supply switches to CC mode, the same arrows should switch over to adjusting current limit until the output switches back to CV mode.  In other words, the same as my current BK Precision bench supplies.



And an optional module for an add on opto-isolated RS232 port would be handy too.
I also agree with you that a spinning rotary knob is dangerous unless it's indented single step with indents stiff enough and large enough that you can never make a mistake by accident.  Otherwise, it's a potential magic smoke moment for what you are working on...  Also, wear out and damage on these rotary encoders may achieve the same problem.

[jbb]

Oh, this is a hard one.

Can I be a pain and say “both”? (In the absence of knobs.) I’d like a voltage up/down, a current up/down and then a numeric for quick entry.

Looking forward to seeing the results.

[cat87]

Right now I'm thinking of a scenario where I need to set a certain voltage/current value with the keypad, then slowly change that in mV or hundreds of mV steps with the up/down keys.
So, just to make things harder for you, I'd go with "both"   

[TheUnnamedNewbie]

I would have to go with the "both" option. But if that is really not possible, I have to say I prefer numerical keys, though it depends on the resolution. Much easier to go from 2 V to 15 V by just pressing "Vset - 1 - 5 - enter" or something. Up-down I would mainly use to just shift tiny amounts to see what the influence of small shifts in supply is on performance.

[jbb]

Sorry Dave, we're blowing up your poll...

[BravoV]

Oh, this is a hard one.

Can I be a pain and say “both”? (In the absence of knobs.) I’d like a voltage up/down, a current up/down and then a numeric for quick entry.

Looking forward to seeing the results.

Hopefully the firmware will be hack-able, then turn the numeric keys into 4 directions arrow keys by some unique sequence like press & hold certain button to activate it for example.

[mikerj]

No encoder   I really do not like bench supplies without adjustment knobs for manual use (fine for automated test etc.).

With only push button input,  both direct entry and an increment/decrement would be preferable.  With only direct entry, multiple small adjustments become slow and very error prone, and requires you to concentrate on the PSU keyboard rather than your circuit or the instruments attached to it.  With only increment/decrement setting a specific value becomes a bit of a chore, but at least you can adjust the voltage up and down whilst looking at another instrument.

If it really has to be only one or the other then the only sensible option would be an increment/decrement system, and you should be able to choose which significant digit to operate on, and it should be difficult to accidentally change the active digit (e.g. maybe require a longer press).

[CopperCone]

  :popcorn:yes rotary encoders can be a pain in the balls especially when they malfunction and jump voltages in carefully sequenced things. I would trust a potentiometer more unless there is rate limiting code in the mcu control algorithm. I.e. halt the jump of the computed value is greater then some integer

[ArdWar]

How many digits will be available? Can we have both keypad and up-down? It can't be that hard to add 2 more buttons (the software complexity might be, though), but if I have to choose, I'm leaning to keypad input. Having up-down button might be nicer if I want to ramp things up/down, but sometimes allowing supplies change values while operating might not the best for certain scenario.

:popcorn:yes rotary encoders can be a pain in the balls especially when they malfunction and jump voltages in carefully sequenced things. I would trust a potentiometer more unless there is rate limiting code in the mcu control algorithm. I.e. halt the jump of the computed value is greater then some integer

Wouldn't these design use a incremental encoder? If values jumped around then 99% of the times that's a software problem, and if it jumped around because of a dodgy intermittent signal then it maybe still a software problem.

[SiliconWizard]

Both entry modes are useful in different situations IMO. So ideally, that would be great to support both.

As an alternative to a plain numeric entry that would be a compromise, you could allow up/down operation either on the whole value or on a specific decade (that one could select with the push of a single button or with arrows). In this mode, the selected decade digit could be blinking. This would allow quick entry while keeping the number of keys at a minimum.

Just my 2 cents.

[GeorgeOfTheJungle]

Lab equipment without knobs?? Are you kidding me? Up/down buttons are a poor substitute for proper fine and coarse knobs imo. The keyboard is a must only for the plus/pro model.

IOW, WRT knobs, the more the better :-)

[Lightages]

I voted for keypad, but both would be much better. Maybe a function like hold the I or V button while pushing 2 for down increment and 8 for up.

[GeorgeOfTheJungle]

I also agree with you that a spinning rotary knob is dangerous unless it's indented single step with indents stiff enough and large enough that you can never make a mistake by accident.  Otherwise, it's a potential magic smoke moment for what you are working on.

In all my life I never ever had this problem, not even once. (I'm 54).

[madires]

I'd go for the numeric keypad and add an alternative up/down mode, like the numpad of a PC keyboard.

[GeorgeOfTheJungle]

Then if you want to step up once and it wasn't in incremental mode you type in an 8 and the magic smoke escapes...

[madires]

That's easily fixed by a "Set" button:
- enter the value or push up/down keys
- uSupply shows the new current/voltage but keeps the old value still set
- confirm with Set key -> new value gets applied
- if not confirmed within x seconds -> update display with old value

[The Soulman]

Both entry modes are useful in different situations IMO. So ideally, that would be great to support both.

As an alternative to a plain numeric entry that would be a compromise, you could allow up/down operation either on the whole value or on a specific decade (that one could select with the push of a single button or with arrows). In this mode, the selected decade digit could be blinking. This would allow quick entry while keeping the number of keys at a minimum.

Just my 2 cents.

What he said.

[BrianHG]

Dave, do you have a photo of the keypad/up-down buttons you intend to use?
Or will you be making a custom membrane keypad?

[GeorgeOfTheJungle]

That's easily fixed by a "Set" button:
- enter the value or push up/down keys
- uSupply shows the new current/voltage but keeps the old value still set
- confirm with Set key -> new value gets applied
- if not confirmed within x seconds -> update display with old value

...and then the UI is a disaster :-) fine+coarse knobs are better imo.

[max_torque]

I don't care (about having direct voltage input or Up/Down keys).

Because, if it hasn't got knobs, i'm not buying it anyway............   

[Maxlor]

If it has to be just one or the other, give me up/down keys. While those make just setting a voltage more cumbersome, the advantage of being able to gradually change the voltage whilst not looking at the keypad (but instead at, say, a scope screen) is worth it.

[mikeselectricstuff]

Up/Down keys should probably not have any acceleration on a power supply - should just be a simple auto-repeat style at a constant speed, with some way to seperately control that speed ( like a digit select).

Given the choice of numeric or up/down, then it would have to be the latter as there will always be times you want to increase/decrease the output, and numeric only would be useless for that, but there are also times you want to enter a specific voltage numerically.

A possible third way, given the "buttons only" constraint, would be an up/down button pair (with auto-repeat) per digit.
This would allow any voltage to be quickly reached, as well as up/down adjustments.
This would need fewer buttons than numeric entry, but still provide a quick way to get to a specific voltage

[plazma]

Up/down + left/right for selecting the digit.

[andyturk]

What say you.

Bring back the knob. 

[BrianHG]

A possible third way, given the "buttons only" constraint, would be an up/down button pair (with auto-repeat) per digit.

Up/down + left/right for selecting the digit.

Dangerous.  When looking at my scope, if I hit the 10s digit +, I can go from +5v to +15v in 1 step, or the same with current if my cursor is on the wrong digit, or, I hit the wrong 'UP' button.

To move the large values, a keypad with numbers + enter, and cancel is great.
To increase and decrease voltage and current in steps, the addition of up/down arrows is also needed, but, keep those steps at a safe manageable size.  Acceleration here is ok if the up/down button is held for more than 3-5 increment steps.

I prefer both the keypad + at least 1 dedicated set of up/down arrows + 1 enable/disable output button.

[tocsa120ls]

*Bugs Bunny voice* Imagine, asking for it

No buttons at all, make an iOS-only app that functions as input, but has no two-way comms, so the actual values are displayed on the LCD only

[Mr. Scram]

Whatever the case, I'd prefer rubber buttons rather than metal dome buttons.

[mikerj]

A possible third way, given the "buttons only" constraint, would be an up/down button pair (with auto-repeat) per digit.

Up/down + left/right for selecting the digit.

Dangerous.  When looking at my scope, if I hit the 10s digit +, I can go from +5v to +15v in 1 step, or the same with current if my cursor is on the wrong digit, or, I hit the wrong 'UP' button.

Agreed, but if it's really only direct entry or inc/dec then there's really no choice; direct entry by itself is useless for bench work.  Digit change could be protected to some extent, e.g. long push or perhaps require an additional function button to be pressed at the same time.

[fable]

With numeric keypad you can have both

[Jeroen3]

With numeric keypad you can have both

Num lock.

[Avacee]

Assuming the button + screen sizes are appropriate then one option is to use a 2x5 layout for buttons 0-9 that line up with the voltage and current set values.
e.g.
VV.VVV 
AA.AAA 
01 234
56.789

4+9 do Up/Down for the right most digit (1mV/A) and so on with a clear indication whether the keys are in Up/Down or Value Entry mode.
Don't have to faff around moving a digit selector left+right then.

[TheUnnamedNewbie]

That's easily fixed by a "Set" button:
- enter the value or push up/down keys
- uSupply shows the new current/voltage but keeps the old value still set
- confirm with Set key -> new value gets applied
- if not confirmed within x seconds -> update display with old value

...and then the UI is a disaster :-) fine+coarse knobs are better imo.

Why would the UI be a disaster? I think most of the supplies I work with on a daily basis work like this, and I'll take them over "Fine+coarse" any time of the day.

[GeorgeOfTheJungle]

Good for you. But knobs are better :-)

[0xPIT]

For a PSU, a rotary encoder with momentary button (or two) will feel much more natural than any keyboard.

Remember what you can do:
- Rotate slow/fast (acceleration)
- Click
- Double Click
- Long Click
- Hold
- Rotate While held
- and even enter digits using a binary sort keyboard

I vote for encoders

• .

[ammjy]

i hope uSupply have + and - voltage together.

[xani]

There is a reason almost every power supply with numeric keyboard also has a knob

#bringbacktheknob

[madires]

It seems Dave would have to offer three models with different UIs to make everyone happy.

[GeorgeOfTheJungle]

There is a reason almost every power supply with numeric keyboard also has a knob

#bringbacktheknob

And a knob is not necessarily an analog potentiometer, 2 gpios suffice to make an incremental rotary encoder work.

#bringbacktheknobS 

[JPortici]

I imagine that the usupply will be as flat and compact as possible, which would be the reason why he doesn't want knobs.

I voted for numeric entry, because that's what would make me the least mad if i had to have one or the other. however numeric + up/down would be the best. why not combine them, use another button as "shift"?

[Mr. Scram]

It seems Dave would have to offer three models with different UIs to make everyone happy.

There's the classic mistake of assuming that what people indicated is preferred will actually sell. You wouldn't be the first one to get hurt that way.

[mikeselectricstuff]

Assuming the button + screen sizes are appropriate then one option is to use a 2x5 layout for buttons 0-9 that line up with the voltage and current set values.
e.g.
VV.VVV 
AA.AAA 
01 234
56.789

4+9 do Up/Down for the right most digit (1mV/A) and so on with a clear indication whether the keys are in Up/Down or Value Entry mode.
Don't have to faff around moving a digit selector left+right then.

<enter> before and after a numeric entry, per-digit up/down ( with auto repeat), though probably not on the 10s

Or a V and an I key :
<V> 1.23<enter>
<I> 2 <enter>
To avoid needing a  clear key, timeout if <enter> not pressed within  a few secs - maybe pressing V or I also cancels.

[David Chamberlain]

completely impracticable idea here. How about a capacitive touch screen with a knob on it that is if you are really hell bent on not using a real knob.

If this really is a binary poll then I would say a numeric keypad only.

Does uCurrent also mean micro physical size constraint for the UI?

[Fire Doger]

It would be easy to include both, you can have full numeric and with a function key or a combination of 2 simultaneously keys to change to up down left right mode for digits 2,8,4,6.

So a numeric can work as up down but up down can't work as numeric.

[xani]

Looking at the size and format of uSupply's LCD (53x57mm) I think we can say it wont be that flat that size of front panel would exclude encoder, so there must be other reason.

completely impracticable idea here. How about a capacitive touch screen with a knob on it that is if you are really hell bent on not using a real knob.

Every single app or device which tried to do whole "let's just emulate knob on a screen" felt utterly fucking terrible in every way. Sliders work much better for touch.

[GeorgeOfTheJungle]

In general, yes, but that touch scroll thingy on the first iPods worked wonderfully.

[Jeroen3]

Every single app or device which tried to do whole "let's just emulate knob on a screen" felt utterly fucking terrible in every way. Sliders work much better for touch.

Yes. But sliders can be bad as well.

In general, yes, but that touch scroll thingy on the first iPods worked wonderfully.

That wasn't emulated on a screen. It was really there.

[raxpox]

How about a touchpad with numbers and sliders printed on the pad surface?  The user could enter the voltage directly, slide the V/I, or use 'gestures' (pinch/zoom, etc.) to change output parameters.

Another possibility is the use of force-sensitive resistors, as used on some gestural-input musical instruments (Linstrument, etc.) and recently discussed on Embedded #236 (https://www.embedded.fm/episodes/236).

A 3D-touch gestural interface on a power supply would certainly be innovative... terrible at first, probably, but innovative.

[raxpox]

Referring to my previous post about using a printed touch-panel, here is a suggested layout:

[David Chamberlain]

Referring to my previous post about using a printed touch-panel, here is a suggested layout:

I like the ON button being as far away as possible from the change value buttons. Also you are missing a method to select the decimal place for the up down selectors?

[BrianHG]

Referring to my previous post about using a printed touch-panel, here is a suggested layout:

I like the ON button being as far away as possible from the change value buttons. Also you are missing a method to select the decimal place for the up down selectors?

It's a generic step up/down, if you want to change the V or I by a large value, just type it in.

[JS]

How many buttons are you willing to put on the thing?

Keypad is nice, up down is a must, typing 5 digits when you want to increase 1LSD is a BUGBIG NO.

Would be nice to have a set button when you go up or down, but you should probably be able to disable it and have direct increments. If no keypad this is a must, if there is a keypad it minght not be there.

The thing very quickly ends with ton of buttons if you don't want to have multi function buttons, OVP and OVP settings, V, I, keypad, bs, incremets, set, clear, on, off. Over 20 buttons there.
NumLock might save a few, increments, OVP, OVC, and other options might be in the numbers.

JS

[Jeroen3]

No touchpad please. If a wire comes near, touch keys go ballistic. Plus if you're on an isolated workbench/chair/mat it doesn't work at all.
And there is no tactile feeling, you'd have to emulate it by clicks or beeps.

[mikerj]

No touchpad please. If a wire comes near, touch keys go ballistic. Plus if you're on an isolated workbench/chair/mat it doesn't work at all.
And there is no tactile feeling, you'd have to emulate it by clicks or beeps.

Yep, a bench instrument with capacitive touch controls is right down at the bottom of the pile in terms of usability, with proper buttons some way up and proper rotary controls at the top.

[raxpox]

How about a force sensitive resistor, like the ones used on the Linstrument and as talked about on Embedded #236?  The gesture control (slider) would be easier to implement, since that's a predominant feature of the technology.  I suspect that FSRs would have the same external voltage sensitivity issues (unless properly shielded) as capacitive touch panels, and the force required to actuate the buttons may be higher than people expect from a 'touch' device.  That said, the FSR gestural instruments seem to be very tunable to a human-range of input pressure.  Since this is a printed panel, it doesn't have to be transparent, so the dependence on ITO or other exotic materials is not required.

Regarding the button layout: I'm not a UI designer, it was just for illustrative purposes.  A button change here or there will enable the required functionality.  The point was to illustrate how a slider could be nestled in among the buttons to enable a scrolling (knob-like) function without a knob.   Personally, I prefer turning knobs over pressing buttons, since I grew up with a Tek 545 as one of my first toys, but I realize that physical knobs are not practical in a compact device.  Maybe when we're all wearing AR glasses/contact-lenses and have haptic-feedback neural implants, we can revisit the knob as an effective virtual input method.  And no 'knob' jokes, you bunch of knobs.

With respect to comments (this goes for the whole EEVBlog forum): It is much easier to come up with reasons why a technology doesn't work than it is to just try something out.  A good example is touch-screen technology in aircraft avionics - this was considered impossible for *many* reasons, until somebody implemented it, then it magically became possible.  There are HIRF/Lightning/EMC concerns with capacitive input screens, but careful engineering and understanding of the limitations will guard against 99/9% of the undesired results.  Avoiding a technology due to a 0.1% chance of false inputs is very much "putting the cart before the horse."  There are certainly boondoggle ideas out there and Dave (et al) does a good job of implementing critical thinking to prevent people from wasting money.  No worries, though - if an idea has merit, somebody will profit from it.

Thanks for the feedback, everyone.

[mikeselectricstuff]

No touchpad please. If a wire comes near, touch keys go ballistic. Plus if you're on an isolated workbench/chair/mat it doesn't work at all.
And there is no tactile feeling, you'd have to emulate it by clicks or beeps.

Yep, a bench instrument with capacitive touch controls is right down at the bottom of the pile in terms of usability, with proper buttons some way up and proper rotary controls at the top.

Yes -one reason touch buttons are a non-starter for testgear is you may well be wanting to press buttons while looking at something other than the keypad - e.g. watching for smoke, looking at other instruments.
A membrane keypad should have sufficient tactile feel ( domes on buttons or ridges between) to be able to find buttons without looking.
 

[Kalvin]

I would like to see Dave making a review of his own power supply and its ergonomics in the similar manner he has been reviewing the power supplies and their usability over the years. At least he should have a good knowledge on how *not* to design a power supply user interface and thus he should be able to avoid typical UI goof-ups.

[Maxlor]

Or maybe the whole buttons thing is something of the past. How about no buttons, but instead cellular connectivity, settings that are stored in the cloud, and a UI in a phone app? You could make the buttons configurable then to make everyone happy. You could use swiping with velocity control! And best of all, since it goes through the cloud, settings could be saved and transferred to other devices easily!

And of course, there could be different models with a price point appropriate to their functionality! Like, say, you need 1mA resolution? You can rent that feature for just $.99 a day! Data logging and graph display? Just $.19 per screen width!

 

[EEVblog]

How about a force sensitive resistor, like the ones used on the Linstrument and as talked about on Embedded #236?  The gesture control (slider) would be easier to implement, since that's a predominant feature of the technology.  I suspect that FSRs would have the same external voltage sensitivity issues (unless properly shielded) as capacitive touch panels, and the force required to actuate the buttons may be higher than people expect from a 'touch' device.  That said, the FSR gestural instruments seem to be very tunable to a human-range of input pressure.  Since this is a printed panel, it doesn't have to be transparent, so the dependence on ITO or other exotic materials is not required.

Not practical when you have say 1024 positions you need to set to precisely, and limited space. Pretty much guarantee that any such solution will suck.

[xani]

I would like to see Dave making a review of his own power supply and its ergonomics in the similar manner he has been reviewing the power supplies and their usability over the years. At least he should have a good knowledge on how *not* to design a power supply user interface and thus he should be able to avoid typical UI goof-ups.

Like lack of knob...

[Abbas]

*the feeling when you want to answer the poll but your choice is not there, what should i do!!?? *
In my opinion I would like to go with both(probably the third option  ) key for individual Voltage and Current adjust button and with separate direct numeric keypad input. It's always handy to input directly with keypad the voltage and current settings, and slightly adjust up and down when you need/feel to.

[joey120373]

Skimmed this thread, haven't voted just yet, but i like the idea of a numeric pad with UP-DOWN buttons in addition to the numeric buttons.

Of course this is an issue when it comes to adjusting the output on the fly by say +1v in .001v increments.

Velocity control would be good here, but as Mike pointed out, it could easily under/overshoot. However, ( spit balling here ) what if there were two velocity modes, one ( realtime ? ) that sets the output automatically,  and another ( Safe ? ) mode that allowed the setting(s) to be changed but the output would not be active (change) until either
A: a "set" button was pushed, or
B: a specified delay time passed, allowing the user time to correct a mistake.

Kinda spit-balling here.

One thing i haven't seen proposed is a cap-touch option ( unless this is what you were referring to raxpox ), this would allow the keypad to be directly on a PCB, and cost of extra buttons would be essentially nothing.
I'm still a fan of tactile buttons, but a cap touch PCB offers almost infinite flexibility and probably cheaper in the end vs tactile or membrane buttons ( please no membrane buttons, i hate those). 

Another approach would be an interface that could be completely customized by the user.

the Cap touch ( or whatever style ) front/switch panel could be its own system, with, say a cheep arduino compatible micro on board to either read the cap touch pads, buttons, pots or encoders . This would allow the user to write his own code, or even design his own front/switch panel. Simple Uart comms to the power supply controller. And if you don't like the design,  just roll your own  with buttons, pots, encoders,  or whatever floats your boat.
Add a port for a cheep ebay bluetooth serial interface and now you could have a remote controlled power supply...

   
 

[technix]

I like a numpad personally. It is technically downwards compatible with the up/down keys through repurposing some keys.

I don’t like a keypad with a lot of buttons though, as it can get confusing fast. Here is the keys I think is needed, with my intended layout and repurposes:

7/Home | 8/Up | 9/PgUp | Escape/Stop
4/Left | 5 (nub) | 6/Right | Fn
1/End | 2/Down | 3/PgDn | Backspace/Cancel
Period | 0/Clear | Enter/Start (bigger button)

This layout is inspired by the numpad on a PC keyboard. It is, at the same time, a numerical keypad and a navigational keypad.

The key cap of number 5 have a small nun on it for touch typing and assisting the disabled.

The Escape button is also the emergency stop button: when pressed seven times within two seconds all outputs are disabled until power cycled.

The Fn key allows other keys being further modified, like enabling a dangerous feature requires pressing Fn-Enter instead of just Enter.