LED Electronics Workbench lighting - DIY

[caall99]

I have a 6 foot long professional all metal construction electronics work bench, with ESD laminated top. It also has a shelf under which i would like to install some DIY lighting. I have heard horror stories about fluorescent lighting and the noise induced into precision electronics from high frequency EMI.

I am therefore tempted to go down the LED route. I was thinking of using LED strip lights in an aluminum extrusion with a diffuser.

Here is my shopping list thus far:

LED strip light, dimmer and powersupply:
http://www.amazon.com/LEDwholesalers-Controller-2034RGB-3315-3215/dp/B0040FJ27S/ref=sr_1_4?s=home-garden&ie=UTF8&qid=1406648096&sr=1-4&keywords=led+wireless+remote+control

(liking the idea of RGB, so i can fine tune the color temperature i want)

Aluminum extrusion:
http://www.amazon.com/Shallow-Aluminum-Channel-Extrusion-cover-U03/dp/B00F9RW2UU/ref=sr_1_5?s=hi&ie=UTF8&qid=1403894926&sr=1-5&keywords=led+aluminum+channels#productDetails

(nice diffusion panel on this. also flush-mount, so it won't protrude from underneath the shelf)

My main concern is EMI noise from the cheapish chinese power supply, and the wireless dimming circuitry.

All my electronics work is high precision 7th to 8th digit stuff. Am i shooting myself in the foot here? Does anyone have any experience with this?

[mariush]

I would recommend against going with those cheap light strips simply because of the poor CRI.  Most of those cheap leds have 70-75 CRI, which screws up the colors.

I would go with 5-10w white leds from reputable stores like Digikey, Mouser, Farnell/Newark/element14 , RS-Components, TME.Eu - they're more expensive but at least you can get units with 80-83 CRI and binned properly (same lumens amount, same color temperature)... You can choose 4000k, 5000k.

I bought one 10w led (4000k and 80CRI) for testing purposes and I'm very pleased with it .. works at about 21v so it's quite easy to get a cheap 24v power supply and then use a buck driver for each led to control brightness. Alternatively, for even less noise, a classic 15v transformer and a current limiting resistor for each led would do the same job.

If you don't care about CRI, TME has some nice 25-40w leds that have very good price for the lumens.. I think they were about 20$ each if you buy 5, 15$ if you buy 25 or something like that.  Those leds do loads of light.

[thm_w]

Found this after reading mariush's response: http://www.candlepowerforums.com/vb/showthread.php?225839-CRI-and-efficacy-of-RGB-arrays-versus-white-LEDs

I've seen a few LED desk lights with adjustable temperature, but they use two banks of white LEDs behind a diffuser. It only had a few settings, but presumably you could also step between the two temperatures if desired.

[Kjelt]

Don,t use rgb leds but use two different color temperature white leds something like 2700K (very warm white) and 5000K cold white.
I would suggest making your own pcb with low power leds in series and use a controllable linear current source on the pcb or use dedicated ic,s. Check out mike electrics stuff topic about his heathrow lighting art for component ideas.
Use a lot of low power leds instead if a few high power leds. Why? High power leds have heating issuses to take care off which costs space and heatsinks. And most important you want an even ligt distribution on your workspace without hotspotting. This means use as many light sources as you can and you can get away with a simple diffuser. If you use a few high power light sources you need an expensive pro diffuser panels .  Check out how for instance Philips, Osram and chinese companies make led tl replacements. They all switch to medium/low power leds directly on a pcb without extra cooling.

[Corporate666]

Personally I wouldn't bother mixing color temperatures... it's unlikely you will get very good mixing (at least not without losing a lot of light), and it will always be inferior to just using LED's of the mixed color temperature from the start.  I realize using one kind of LED prevents you from tweaking color temperature, but temperature isn't something you would want to be changing on an ongoing basis, IMO.

I have 2700k lighting at home and I have 5000k at the shop.  I'm very happy with both, and I would never want to change my shop back to 2700k (which it was before we upgraded).  It seems like it would be an awful lot of extra work to be able to mix color temperatures, and you would almost certainly find a point you like and just leave it there.

[ovnr]

I quite like Bridgelux's Vero series of LEDs (10W-80W or so). They're available in CRIs of 90 for most temperatures, and 97 for their warm LEDs (3000k or thereabouts) - bit more expensive tho. The mounting system is also a joy to use compared to loose MCPCBs. Even the 80W LEDs are only $30 or so, and all of their products are specced at up to 2x the ordinary drive current (21000 lumen LED anyone? Just gotta watercool it.)

Only issue I have with them is that they're too high voltage to be able to comfortably run a lot of them in series, and finding low-cost drivers so you can run one driver per LED is pretty much impossible too. I've been meaning to set up 4x 15W LEDs as workbench lighting, and will probably run them in series on a horrible deathtrap of a home-made driver (120V output). I'll probably get some plexiglass caps lasercut to protect them from flying debris and unsuspecting fingers poking around.

[TerraHertz]

Hah. We're all thinking alike. Pic is of two reels of LED strip, and aluminum L sections for improvising lighting on my soldering bench.

Strips are 1 warm, 1 cool white. ebay seller lemon-best
Color:5050 Cool White  Item Id:141300252362 - Price: AU $9.74  5m roll
Color:5050 Warm White  Item Id:141300252362 - Price: AU $9.74  5m roll

Presently thinking about the power supply. I want to have one regulated current supply (well filtered) to the common high side, plus two FETs and current sensors, one on the low side of each strip. Two control pots: brightness and tint. Brightness feedback to the main supply using the sum of the two currents, while Tint sets the ratio of currents via the two FETs. This way only needs 3 wires to the light fitting.
I started working out a circuit using op-amps and stuff, but then realized this is much better done with one of the postage stamp arduinos. So that will be fun - I haven't used them previously. Have a pack of 10 of them I bought a while ago, for all of $2.50 each.

[TerraHertz]

Oh, and one reason I'll do my own supply, is I don't want this to happen again. Pic is of a nice cheap LED worklight bought from Jaycar last year. I thought it was great - it's bright, doesn't get hot, and at last a portable worklight that doesn't die if you bump it too hard. Had been planning to buy a few more of them. Then recently it was working one day, but when turning it on the next day it was dead. Pulled apart, and the inverter module is dead. It's also hard-epoxy potted. Grrrr.... And out of warranty no doubt, not to mention that I'd already disassembled it.

The other reason is that I'm sure there won't be any kind of two-LED-strip driver with current ratio control between the strips. RGB colour strips yes, but then the controls will be wrong for this use.

[caall99]

Thank you for the responses everyone. I proceeded to purchase the LED strip lights. I bought a variant that alternates RGB LEDs with cool white LEDs. I think the "infinite" color control of this combination should give me something that i feel is pleasing. I also bought the 2.4G/wifi remote to control them. Still worried about noise induced from the PWM in my other electronics. Will report once the light fixture is built and installed.

[SeanB]

I bought a few myself, and the inner module is not potted, but just in a shrink sleeve. I took one apart to use in an uplighter, and it works with the 2 parts separated and on wires.  10W lamp is not as bright as the 70W MH it replaced, but it does use a lot less power. 70W lamp definitely is brighter than the 500W halogen that was in the fixture before, and lasts a lot longer as well.

[David Hess]

If I wanted both high efficiency and low EMI, I would use linear fluorescent lamps and a passive ballast but I guess they are getting difficult to find.

With modern lamps that use an electronic ballast, I would consider adding external filtering to the AC line to reduce conducted noise and running them on a separate circuit.

[mij59]

If I wanted both high efficiency and low EMI, I would use linear fluorescent lamps and a passive ballast but I guess they are getting difficult to find.

With modern lamps that use an electronic ballast, I would consider adding external filtering to the AC line to reduce conducted noise and running them on a separate circuit.

The electronic ballast is not the problem, the lamp becomes a hf transmitter.

[David Hess]

If I wanted both high efficiency and low EMI, I would use linear fluorescent lamps and a passive ballast but I guess they are getting difficult to find.

With modern lamps that use an electronic ballast, I would consider adding external filtering to the AC line to reduce conducted noise and running them on a separate circuit.

The electronic ballast is not the problem, the lamp becomes a hf transmitter.

Extra filtering at on the AC line side still helps but if a passive ballast is used, then there is no high frequencies for the lamp to transmit barring some interesting exceptions; old incandescent lamps with coiled filaments operating in a vacuum can radiate RF at 1 MHz and higher.

I find the whole situation rather annoying since at some point I will be faced with replacing fluorescent tubes with lower efficiency halogen lamps if I want a low EMI work area.

I also prefer linear format fluorescent tubes because their large surface area inherently yields illumination with low shadows.

[NiHaoMike]

Use a linear PSU and select just the right number of series LEDs to allow the regulator to be very efficient.

[T3sl4co1l]

Last year, I picked up a nice bag of those cheap 3W LEDs (they were on sale besides!).  Which of course are more like 1W, not really even 2W if you shrug at them, but, whatever.  Anyway, I built two circuits, one a sort of fairly standard flyback as a sconce sort of thing, as well as a strip design twice as powerful:



(Ooh, that #26 core got really toasty.  That got changed to a Kool-Mu, I forget which part number exactly, but similar size.  It still runs hot, but not dangerously so.  A ferrite cored choke would be even lower loss, but not really worth it.)



(The red strings are a much, much earlier creation.)  The windows are receiving afternoon direct sunlight, so you can tell they're pretty bright.

Fair warnings: line operated circuitry, don't touch it, or build it yourself, blah blah.  Also, no EMI filter.  Properly, this sort of design should be filtered, and boxed in plastic (to qualify as double insulated equipment).  In my defense, I did mount the LED strip up high, out of most "random groping" range.

Tim

[jlmoon]

Oh, and one reason I'll do my own supply, is I don't want this to happen again. Pic is of a nice cheap LED worklight bought from Jaycar last year. I thought it was great - it's bright, doesn't get hot, and at last a portable worklight that doesn't die if you bump it too hard. Had been planning to buy a few more of them. Then recently it was working one day, but when turning it on the next day it was dead. Pulled apart, and the inverter module is dead. It's also hard-epoxy potted. Grrrr.... And out of warranty no doubt, not to mention that I'd already disassembled it.

The other reason is that I'm sure there won't be any kind of two-LED-strip driver with current ratio control between the strips. RGB colour strips yes, but then the controls will be wrong for this use.

So how much noise do you think the LED Driver produces?  I think one of the original points in this thread was avoidance of the noise issues pertaining to lights fluorescent in nature.   >> just curious

[mzzj]

If I wanted both high efficiency and low EMI, I would use linear fluorescent lamps and a passive ballast but I guess they are getting difficult to find.

With modern lamps that use an electronic ballast, I would consider adding external filtering to the AC line to reduce conducted noise and running them on a separate circuit.

Fluorescent lamps are quite noisy by nature because of the electric discharge.

But you can use the old fashioned iron ballast to power the leds. Wont get much quieter(emi) than that! 
Couple of bridgelux modules to get the voltage around 60-90 volt or so, bridge rectifier and  old iron ballast.

[TerraHertz]

So how much noise do you think the LED Driver produces?  I think one of the original points in this thread was avoidance of the noise issues pertaining to lights fluorescent in nature.   >> just curious

There are multiple potential noise issues from lighting, in the context of an electronics development lab, where you might be working with sensitive circuitry. The sources are:
1. The ordinary mains wiring in the room. Radiates at 50Hz plus harmonics due to any sine distortions from nearby loads, plus noise conducted in on the wires from noisy loads in the neighborhood.

2. Noise radiated from the light fitting/tubes & wiring. Mains fluorescent tubes are inherently noisy due to being a long arc. And that's ignoring the high voltage spikes during starting. Cold cathode tubes as in LCB backlights use drive voltages of several KV, so are extremely bad.

3. Switching noise radiated from any light that uses a switchmode supply. This can be direct from the supply, and also coupled into both the input and output wiring as antennas. With commercial units you take what you get. With a properly designed and constructed unit this can be eliminated.

For most electronics development none are these sources are a concern. But if you want to do very low voltage high impedance work, it is an issue. It's one reason I recently added old fashioned transformer-driven 12VAC halogen ceiling lights to my workspace - now if I want to I can turn the fluorescent battens off, and just use the halogens. There's still the 240VAC to the power points, but that can be isolated by turning off the breaker elsewhere. That said, I'm in no way trying to achieve full Faraday Cage shielded room effect.

With the LED striplights, they're for a bench to be almost exclusively used for soldering, so I can't see noise there being a problem. But just on principle I'll design the drive circuit to not radiate RF. Good mains filter on the input, metal case, PF correction input stage, then all LV wiring well filtered before leaving the box. The wires to the LED strips will carry only clean DC. Same with the wires to the two control pots, if they are not on the actual drive box.


Incidentally, I don't know about contemporary digital scopes with LCD screen backlights, but a couple of Tek ones I used about a decade ago were appalling. Unusable for low voltage analog work, imo. The screen backlight was cold cathode and apparently unshielded in the plastic scope case, resulting in multiple millivolts of 30-40KHz noise pickup in all circuits on the bench anywhere near the scope.