Interested in Apex/RK controlled d120's?

[Rosco's Reefs]

Because we can (or want to!) Why else do we own and play with aquarium controllers?  

[AlanM]

Made some progress.  So I used some parts from work to look at the little circuit that the potentiometer is attached to.  It's a 500k Ohm pot that Evergrow supplies on their board.  The number doesn't really matter, just that it's really really high compared to the internal resistance.  When it's that high no current flows through the board (actually, a couple microAmps) and the light is fully on. 

 

When you turn the pot all the way down to 0 a resistor on the board prevents it from going all the way to 0 and keeps it at about 280 Ohms which is just a bit up above the dimmest level allowed.  If you go all the way dim at 0 Ohms you start to see the lights flicker a bit which would annoy the heck out of people. 

 

The problem is that it's not linear.  With a 10k pot I can go from all the way dim at 0 to almost all the way bright at 10k with most of the dimming happening between 0 and 2k, but then if I pull the pot off, to make the jump from 10k to infinite resistance, it gets just a bit brighter. 

 

So all of that circuit is for linearizing the dimming function.  It gives finer control at lower Ohm resistance and much coarser control at high resistance so that you can go from full off to full bright while maintaining a useful range of dimming over the entire travel of the knob.

 

Would it be useful for folks if I had it go from like 0 to 90% brightness under control and then full on for the last 10% (by opening the circuit)?

[monkiboy]

Would it be useful for folks if I had it go from like 0 to 90% brightness under control and then full on for the last 10% (by opening the circuit)?

sounds like it would make the most sense that way indeed. i really appreciate your time and commitment to this endeavor. it's great just reading your updates - fantastic!

[DaveS]

The point about it bottoming out at 280 Ohms isn't a surprise.  My Meanwell drivers also have flickering problems at the equivalent of 0 intensity so I keep the lowest setting at 5%-10%. 

 

If it makes it much simpler, instead of an uber complex circuit, to go linearly from the lowest setting to 90% and then a step function to go from 90% to 100% I think most would be ok with that.  If nothing else, it's a start until someone else figures out a better way. 

 

I would caution to check that the settings your circuit provides and make them consistent with what the original pot provides.  In other words, staying above the level where flickering occurs and also staying at or below their current level for 100%.  While you may be able to get the circuit to get the driver go higher, there may be heat and other long term stability issues that caused Evergrown to stay below the max.  I don't think we want the "Red Ring of Death" or whatever other equivalent due to the mods. 

 

Thanks for spending the time on this!!

[AlanM]

I think I don't have to worry about blowing the driver.

 

If I short the dimmer wires together, the highest current observed, with 2.5mA going through the dimmer wires, is at all the way dim and that's when you see a flicker.

 

If I then open up the dimmer wires, so current is 0mA it goes as bright as it can.  The driver maxes out at 530mA, btw.

 

Wonder if an audio or log taper instead of a linear taper one would do something good.

Edited June 24, 2013 by AlanM

[AlanM]

mA	V	Ohm 	LED Current	LED Voltage
0.2	9.81	481000	0.59	82.7
0.35	9.77			
0.38	9.76	467000		
0.57	8.99		0.57	
0.59	8.87	460000	0.56	
0.76	8.04		0.51	
0.77	8.01	446000		
1.03	6.79		0.42	
1.11	6.44	397000	0.39	
1.3	5.52		0.33	
1.41	5.03	301600	0.3	
1.52	4.51			
1.63	3.96		0.23	
1.67	3.79	241900		
1.75	3.419			
1.84	3.002	194300	0.18	
1.86	2.868		0.16	
1.96	2.43			
2.02	2.15	143000	0.12	
2.13	1.631		0.09	
2.16	1.47	104200		
2.18	1.366		0.07	
2.26	1.016		0.05	
2.33	0.672	60000	0.04	65.4

Here's a table of data collected on the little dimmer board and on the wires going to the string of LEDs on one channel.  The LED current goes from 40mA at all the way dim up to 600mA at full brightness.  The voltage dropped across the 28 (or maybe 27) LEDs goes from 65.4 up to 82.7.

 

The other measurements were taken across the pot board.  The Ohm column isn't what the circuit actually sees.  It's an offline reading on the output of the board where I disconnected it at those settings and did a separate measurement.

[AlanM]

So I know now I can control it from 0-100% with just a 20k digital pot, but after playing with it some more today with a bench-top CC/CV power supply it seems like it could be as stupidly easy as just putting the Apex 0-10V outputs across the dimming pins.  It dimmed nicely from 0 to 100% just by supplying 0-10V.  It never drew more than 2.5mA of current in either direction from the bench supply, so I don't think it will hurt the Apex modules, but I posted a question about it on the Neptune forums just to be sure.  If I don't hear anything back in a day or two I'll just give it a try and try not to blow anything.

[Tink]

You're my hero!

[DaveS]

The apex 0-10v outputs are super sensitive.. Just shorting the ends when you supply voltage can cause the fets on the board to fry. Curtis admonished me once when he replaced the entire board for me under warranty but said "never again"... I would be super carefuke to ensure this wont cause problems and/or void the warranty.

[AlanM]

The apex 0-10v outputs are super sensitive.. Just shorting the ends when you supply voltage can cause the fets on the board to fry. Curtis admonished me once when he replaced the entire board for me under warranty but said "never again"... I would be super carefuke to ensure this wont cause problems and/or void the warranty.

Thanks for the warning. I will try to be careful. Hmm. Maybe I should borrow someone else's VDM to try it with, heh.

[Origami]

Thanks for the warning. I will try to be careful. Hmm. Maybe I should borrow someone else's VDM to try it with, heh.

Heck, buffer it with an op amp if it's an issue. Nice to know that it only takes a 0-10V signal to provide the dimming range. Where did you pick up the signal injection point? Is it just one of the leads on the stock trimpot? Are you taking pictures?

[AlanM]

Yes, it's one of the leads on the board that the pot is mounted on.  The stock pot is a 500k Ohm pot with a little board on it with a transistor and a few tapering resistors on it to make the dimming more linear as you turn the pot.  The board has a 4-pin connector with 4 wires going to it from inside the driver.  Picture below:

 

 

Two of the wires are the on-off switch activated by the internal SPST switch on the pot.  Those two wires are still connected in the picture above. 

 

The other two wires get a 10V signal sent from the driver.  The driver then measures the voltage drop across those two wires.

 

When it's 10V (meaning the resistance is high and no current flows across the dimmer wires) the LEDs are on with full current of 590mA. 

 

When it is 0V (meaning the wires are shorted and about 2.3mA flows across the dimmer wires) the LEDs are nearly off.  The stock board has the low resistance at around 400 Ohms, so it never really gets down to maximum dim because they start to flicker in a very visible way.  It happens at about 1V when I'm supplying a voltage rather than a resistance.

 

Here's a plot of pot resistance vs LED current:

 

 

Lately, instead of using a pot, I supplied a voltage and was able to see that the dimming works really well when just putting an opposing 10V across the dimmer wires, essentially dropping the voltage across an opposing voltage rather than across a resistor.  It worked really well.

 

The problem with this approach is that current still wants to flow, and the device supplying the opposing voltage might not like current going in to it's positive terminal and out it's negative terminal.  The Apex folks, RussM in particular, said that under no circumstances should current flow back in to the VDM ports.  I'm wondering if a diode would fix the issue and still let the voltage get through.  I don't really know about op amps or how I'd use one to buffer the voltage.  Do you have a pointer to read?

[Origami]

Ugh. I can't see the images from work machine. Blocked. Maybe later.

 

A decent op amp with good rail-to-rail performance, configured with unity gain and powered by whatever that transistor is drawing power from (I'm guessing that it's around 10 volts but you may want to check), may do the job. It offers isolation and should be able to sink a few milliamps without trouble.

 

I can't tell you if the diode idea will work. It may not if the driver is looking for a place to sink current which it sounds like it probably is.

[AlanM]

Yeah, I read somewhere that LED drivers commonly regulate output current by watching the current across some resistor to ground and when you put a little bit of voltage on the dim wires it changes the current measurement on the calibration resistor and fools the driver into giving the LEDs less current.

[monkiboy]

Ugh. I can't see the images from work machine. Blocked. Maybe later.

is it just photobucket or all major photo-hosting sites? i can email you the photo if you like or post it for alan on imgur, flickr, etc?

[Origami]

is it just photobucket or all major photo-hosting sites? i can email you the photo if you like or post it for alan on imgur, flickr, etc?

It's personal photobucket accounts. I can see it on my phone if I go get it. I'll look later.

 

Alan, can you trace out the little auxiliary circuit card enough to generate a schematic? I suspect that it's very simple - possibly just a voltage follower (common collector topology).

[AlanM]

Here's a little sketch I made when I was trying to figure out what the 500k pot was doing at high, low, and middle after tracing the circuit out.  I didn't draw the pot right, but you can figure it out by looking where the 500k and 250k resistors are.

 

 

Does this help?

[AlanM]

Also, here are the other two pics attached instead of linked from photobucket.

 

 

[AlanM]

And one more showing the board that I unsoldered the pot from.  I blew the transistor on this one with a battery.  Doh. 

 

The top two holes are the switch, the bottom three are the two sides and wiper of the pot.   the four pins on the right are top two power switch, bottom two dim+ and dim-.

 

[Origami]

Yes. It does and it doesn't.

 

It looks like it's a common collector voltage follower. You don't show it, but is there a connection to another point between the 56K and 10K resistor or is it floating? Even if it is floating, it's acting as a virtual ground.

 

I'm also wondering if you've captured it accurately. For example, I'm trying to resolve the apparent 9.87 volts across the base-emitter junction of an NPN transistor in the "Max Bright" configuration. My initial look says that this should be closer to the VBE(sat) voltage (like the 0.64 volts that you show in the other "Max Dim" drawing.

 

Do you have a part number for the transistor?

[Origami]

Alan, where do they pick off V+ and V- that are shown in your drawing? Is it what looks like a 4-pin header on the right of your second picture? If so, it looks like one pin might be carrying 10 volts, another may be carrying circuit ground, and perhaps the last two may be carrying the voltage across the collector emitter pair? I see the VDIM+ and VDIM- near the perf-edge but can't relate it to any particular point. The capacitor looks like it's there as a transient suppressor that's directly across the switch contacts but it's hard to say. I'm trying to read into the shadows as traces on the board.

[AlanM]

No part numbers on any of the parts including the pot except for the surface mount resistors.  The transistor just says HF on it.  I may have drawn it wrong.  Here is a sketch of where the board traces are going after looking at it under a microscope.  The one coming from the third pin on the pot goes underneath the 10k Ohm surface mount resistor, it doesn't connect to it.

 

[AlanM]

Alan, where do they pick off V+ and V- that are shown in your drawing? Is it what looks like a 4-pin header on the right of your second picture? If so, it looks like one pin might be carrying 10 volts, another may be carrying circuit ground, and perhaps the last two may be carrying the voltage across the collector emitter pair? I see the VDIM+ and VDIM- near the perf-edge but can't relate it to any particular point. The capacitor looks like it's there as a transient suppressor that's directly across the switch contacts but it's hard to say. I'm trying to read into the shadows as traces on the board.

VDIM+ and VDIM- are the two bottom pins of the 4 pin header.  The header is repeated twice on the board.  the one on the left actually has the header on it.

 

Here's what the top of the board looks like:

 

[Origami]

Sorry, had to head home. Just catching up. Where does pin 2 of the header go?

 

(Sent from my phone)

 

 

[YHSublime]

I can't wrap my head around most of what's being talked here... but I do have a question for the brains. Can my Apex control at IT2080?