Discussion: LEDs, AC power, rectifiers, Caps, and DMM readings.

[VanIsleDSM]

I've just converted my bench grinder to have LED work lights, originally they were 10watt incandescent bulbs running off 12VAC, if you'd like to see more of that look here:
http://www.candlepowerforums.com/vb/showthread.php?p=2417478#post2417478

I used a full wave rectifier and an LM317 with a 1.2ohm resistor, which on straight DC current would be 1.25/1.2 = 1042mA.

Because a full wave rectifier doesn't put out straight DC current, with my DMM I actually only measured 314mA through the LEDs.. I'm not sure how exactly the DMM averages the 1042mA spikes to reach 314mA.. but I know there are true RMS meters.. and I know my is not one of those fancy ones.. so I don't exactly know how it's getting that figure.

Adding a 920uF smoothing cap brought the DMM reading up to 610mA.. I tried a 4700uF too (which would have been too big to fit) and it reached just over 900mA..



Now I also have another project going on some pond lighting.. the original NOMA power supply is again.. 12VAC, I used a full bridge rectifier but no smoothing cap this time.. using LM317s with 1.2ohm resistors I get 680mA on the DMM.. I suppose it must have more internal capacitance.. but then you'd think adding more load would decrease the current reading through each series set of LEDs.. but it doesn't.

And something else I can't understand.. for the pond I also have some standard LEDs wired in series with resistors.. but in this case.. unlike hi power LEDs with the constant current LM317 drivers.. I get results just like as if I was using real DC current... 9.5VCD on the pond I'm using 33ohms of resistance for 2 ~3.5v 100mA LEDs in series.. getting about 96-97mA on the DMM.. using straight 9.5VDC from a new 9v battery gives me virtually the same reading on the DMM.. shouldn't the rectified signal provide less power?

..while typing that I just had a revelation, I suppose the 9.5v I read on the DMM is obviously also an average, and that's why I get an average of about the same current too... I know LEDs can usually handle quite a bit of ripple current, but it can create thermal problems if there's too much.. hopefully I'll be fine.. all the LEDs are heatsinked very well.

One more confusing thing.. on the pond the output was ~12VAC under load, but when rectified drops to ~9.5VDC under load.. while my grinder was about 10VAC and went up to ~12VCD when rectified.. I thought the voltage went up when rectified.. but the NOMA power supply went down? Some internal reason I suppose.

Anyway.. just thought I'd start this discussion to get more insight into all of this...

 

[HKJ]

It is not really a good solution your have made.
The LM317 is supposed to work with a "nearly" stable DC input voltage, i.e. with a capacitor after the rectifier.
And the LM317 will get very hot, when it is running with 1A and a couple of volts.
A capacitor and a power resistor might do the job, but your will get some heat.

It might be a good idea to get a driver from DealExtreme, it can drop the voltage without a lot of heat (At least if it is a switch mode regulator).

Your must also be careful only using stable DC on the LED, no pulsing current, it might be dangerous. Depending on the frequency the grinder might look like it is stopped, while running at full speed!

 

[JB01245]

I think this is on topic. Is there any reason these drivers from Kaidomain http://www.kaidomain.com/WEBUI/ProductDetail.aspx?TranID=2982 wouldn't work on AC with a 12V wall wart? Trying to add some better lighting to my mill.

 

[HKJ]

Most electronic circuits will only work on DC, but a bridge rectifier and a capacitor will make any AC to DC, but at a higher voltage!
The 18 volt maximum input voltage of the driver, might be a bit to low for safe operation.
A 9 volt transformer will be better.

With this kind of driver your will delivering 755 mA to the led, but only drawing maybe 250 mA from the capacitor. The transformer must deliver something in between these values.

 

[VanIsleDSM]

I've seen many capacitive drop circuits using LM317s without a constant DC input.

With a half decent little heatsink the LM317 doesn't get very hot.. and you'll find even some of the drivers on DX are linear and use an LM317 or very similar component.

You say no pulsing current.. yet you tell me to use a switching driver, this is impossible.

I'm not worried about the grinder looking like it's stopped though.. if the lights are on then it is on..

This discussion isn't really about LM317s and driving LEDs with them.. I have done this many times.. I just want to talk about the DC current readings from the DC waveform created by the rectifier and smoothing cap (if present)

I just ordered a TRUE RMS clamp style meter so I can measure high current AC/DC.. now I shall have the tools to figure it all out.. other than an oscilloscope to actually see the waveform... I'll be able to average it properly at least. Here it is: http://cgi.ebay.ca/Tekpower-Auto-Ra...ryZ73163QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

That's one thing I found out.. the difference between my cheapo DMM is that it just assumes the waveform is a sine wave.. and calculates the average as such..therefore if the waveform differs at all from a sine wave then the readings won't be correct..

As soon as I get the meter I'll post the new readings to compare.

 

[HKJ]

The LM317 is designed to smooth a varying input voltage, but not handle a input voltage that goes below the output voltage. It might or might not work.

The driver on DX uses DC input and the LM317 can get very hot, also with a little heatsink. In fact, I have made a program to design LM317 circuits, including the heatsink.

The current is a very complex to calculate in your circuit, a circuit simulation program can do it, but it is complex to setup, if your do not know about electronic.
The switcher cad program here can be used: http://www.linear.com/designtools/software/

Can the clamp meter measure low enough current for your purpose and does it measure RMS? Personally I use a Fluke 189 and a few lesser meters.

Cheap meters have that problem.

When your have a rectifier and a capacitor, the current on the AC line and the current after the capacitor will not be the same!

I forgot:
It is not a problem using a switch mode driver, the frequency is to high to give problems with a "slowly" rotating grinder. But also: Many switch modes has a stable output voltage.

To give your an idea why the current it not as expected:
The rectified voltage is going between 0 and 17 volts, when your add a capacitor your will change the 0 to a higher value.

Before the LED will give any light the voltage must be:
Led Vf: 3.0 volt
LM317 min voltage: 2 volt (About that value, I have not checked the datasheet).
Rectifier: 1.2 volt
Total: 6.2 volt

Before your have a full 1 amp current the voltage must be:
Led Vf: 3.6 volt
LM317 min voltage: 2.5 volt (About that value, I have not checked the datasheet).
LM317 regulation: 1.25 volt
Rectifier: 1.4 volt
Total: 8.85 volt

I.e. your will have the following phases for each half cycle on the mains (voltage goes from 0 to 17 and down to 0 again):
Voltage below 6.2 volt: Led off
Voltage between 6.2 and 8.7 volt: Led on partial power
Voltage above 7.3 volt: Led on full power
Voltage between 6.2 and 8.7 volt: Led on partial power
Voltage below 6.2 volt: Led off

The light from the led will depend on how much time is spend in each phase, with a capacitor added you spend less time at the low voltages and will get more light.

Here is a curve with a full mains cycle, rectified without a capacitor:

And rectified with a big capacitor:

White is transformer output, yellow is capacitor voltage.
These curves are calculated with a 0.5 A constant current load and a 1000uF capacitor.

 

[JB01245]

HKJ - Thanks for the reply. That would explain the smoke when I tried it. Thought I may have had something wired wrong, but didn't want to risk another driver.