Lynx Dimmer [Archive] - Do It Yourself Christmas Forums

06-05-2007, 11:23 AM

The Lynx dimmer I'm working on is DMX512. Hope to have a complete prototype off the breadboard in a month or so. Have to see how it goes. I had planned to create a little usb to dmx dongle that would be usable on other systems. Something with a simple rs 232 protocol to talk from the software (vixen) to the dongle and then have it create dmx512 stream out on RS495. I'm using a table to allow custom dimming curves and a different tech to dim. Looks very promising so far.

grages

06-05-2007, 02:51 PM

Looking forward to you posting more about it RJ.

DynamoBen

06-05-2007, 04:55 PM

We can always use more dimmers. How are you doing the curve? What curve are you doing?

As far as the converter I want to be careful about creating a bunch of different DMX plug-ins. What would be great is if there were only one or two different protocols. Then no matter who created the dongle hardware they would all use the same plug-in.

My dream dongle would have 2-4 DMX universes and would also communicate via RDM. RDM allows you to remotely configure DMX devices (IE no dip switches). It would be awesome to have Ethernet commnunications with the dongle instead of USB.

All of this could be possible if we come up with a fairly standard serial protocol from Vixen to any DMX dongle. Maybe we can model the protocol after the Enttec USB Pro, and USBDMX.com.

06-05-2007, 07:13 PM

My idea was to create a universal protocol for vixen to talk to the dongle and then since dmx512 is a standard any equipment could work with it. That way vixen only needs to deal with one protocol and the equipment only needs to deal with dmx. Sort of like how a three tier database app works. It speaks ODBC and then the odbc driver speaks the database engines language. This way I can switch databases with out changing code. or change the code without changing anything in the database engine.

DynamoBen

06-05-2007, 07:21 PM

Exactly!!! I agree.

We should look at the enttec protocol and usbdmx.com protocol as a starting point.

BTW an RS232 interface may be a little slow for DMX or any other high channel count interface. Most of the interfaces use a USB to Parallel connection to the microcontroller so the output from the PC doesn't act as a bottleneck. Remember there are times when all 512 channels will be changing concurrently.

07-23-2007, 04:05 PM

For those of you who do not know I have a blog posted on the Lynx I have finished it so I thought I would post the address. It is DMX using a open dmx dongle currently.

http://ledtriks.21.forumer.com/viewtopic.php?t=41

DynamoBen

07-23-2007, 08:55 PM

Very nice, I like it. What you have done is similar to what I had in mind. The biggest difference for me was channel count, I was going to go down to 4 or 8 channels and make them stackable. I was thinkin 4/8 channels would fit nicely in a 2-gang back box. Also I am running higher wattage loads (575, 750) so I can't get to crazy on channel count.

In the end nice job. I would love to see the schematic but the link seems broken. :(

07-24-2007, 08:22 PM

Thanks

08-20-2007, 10:40 AM

Heres a few pictures of the LYNX as a finished Project. This is a master it has all the componets installed.

08-28-2007, 05:34 PM

Heres a quick demo vid of the LYNX with vixen set to 512 channels and 40ms events. I'm running the new enttec DMX PRO Dongle on the new plugin I'm testing for KC. I tried to video it at 30 ms but the camera could not keep up. Needs more testing but looking very promising at this point. I definately feel the pro is worth the extra money as I see no glitches in this at all. The Open unit had glitches every so often that appears to be from timeing issues.
http://media.putfile.com/new-dongle

robogeek

10-16-2007, 03:00 PM

Some design problems with your intended design.

1. I don't see a capacitor after the Rectifier. This means you still have an DC Slope Issue because of the sine wave. It needs a capacitor, Just rectifying does not give you a Solid DC Current you need to smooth it out.
http://upload.wikimedia.org/wikipedia/commons/thumb/9/9f/Rectified_waves.png/350px-Rectified_waves.png

2. If you do rectify and smooth it out it will not put out 120 Volts DC it will actually put out 170V DC as this is the Peak Voltage from the AC signal, the 120 Volts is RMS and not peak. This might create an over current issue with some LEDs designed for 120V AC. It might not be apparent initially but over time a constant 170V across an LED string will degrade the LEDs

DynamoBen

10-16-2007, 04:43 PM

Unless I'm missing something or the design changed recently the Lynx isn't designed to do DC dimming. It is designed to do Foward-phase AC dimming, so the design is still sound.

robogeek

10-16-2007, 05:24 PM

Ok, I might have missed that. RJ described it this weekend that it did was a flat DC dimming, and that was my first exposure to the board. I just wanted to point out that the device as it stood didn't do a flat DC voltage.

In fact actually since its just rectified it makes it much simpler since you can use Transistors instead of Triacs and you still get a slight safety benifit since it is not a constant DC source.

DynamoBen

10-16-2007, 06:39 PM

Hey I could be wrong, I didn't design it. ;)

10-16-2007, 10:06 PM

Yes it is a DC dimmer. and if I were to do it like that without taking it in to consideration it could be too much voltage. The max voltage out of the Lynx is 117 volts on normal line voltage with even the slightest load on it(open circuit is as you would expect slightly above 120 volts). The dc ripple is of no concern as the lights are normally operating on massive ripple if you think about it so I have removed the vast amount of this. Remember that the LED strings already use half or full wave recitification without filtering it with a capacitor. The design is sound and has months of operation in testing and will have a full season on it before it is released to the public so to say. Thank you for the input as we love to have lots of people look over our stuff for flaws and no one has brought this up to me yet so you my friend are on the ball. This is how things get better.

robogeek

10-26-2007, 06:07 PM

I am not questioning that it was a sound design, i just wanted to let you know that you are not giving the LEDs a flat level, you do still have times where it hits the Zero volt point and you also have times that it will hit approximately 170 Volts, If you are using a multimeter to look at the signal it will be averaged and therefore giving you what might seem to be a 117 Solid voltage. But if you were to look at it with an O-scope I can guarantee you that without holdup caps it will show as a half wave rectified signal with a tiny bit of deadband.

I believe that is why CrazyLightLady's LEDs Flickered at the miniPlus.

One thing i would do is put a Capacitor on each channel to smooth out the voltage for the particular channel. the only thing i would try to avoid would be the point where there is no voltage from the rectified sinusoid.

If no capacitor is used then the board still needs to be phase locked with the rectifed sinusoid to produce the best possible results

11-04-2007, 11:36 PM

Entropy

12-24-2007, 04:59 PM

Thank you for your input. Do to the method in which it functions a capacitor on this setup would create problems. The crazy ladies leds were ones she gave me to see if I could get them to work as she had issues with them. I have tweaked the firmware to allow even these to be dimmed successfully.
I'd be curious what this tweak was.

Achieving consistent dimming without phase locking the PWM waveform to the AC ripple (or simply filtering out the ripple with a capacitor after the rectifier) is an incredibly impressive achievement. How'd you pull it off?

DynamoBen

12-25-2007, 12:04 PM

So you say this is DC dimming, I'm not sure that I follow you. I went through your forum to try to dig up a schematic so I could understand what you meant but couldn't find anything. Any chance you could clear this up for me?

P. Short

12-25-2007, 01:03 PM

For some reason RJ is very reluctant to discuss it.

My guess is that he is using a full-wave bridge rectifier to create pulsing DC, then using SCRs or MOSFETs to perform PWM control.

--

Phil

12-25-2007, 04:59 PM

Didn't think there was any secret there, I have said it at least half a dozen times.

for example not long ago here in the Lynx go here first post.

http://www.doityourselfchristmas.com/forum/viewtopic.php?t=1626

"The Lynx is a DMX dimmer that just works a little different than a standard PIC dimmer. It does not require a zero cross signal as it does not time to the AC. In fact it doesn’t actually use AC at all it uses pulsing DC to power the lights. It also does not use triacs it uses mosfets to switch the load."

I wouldn't say that was reluctant to discuss it. I just hadn't seen his post as I often miss them. This is why I asked about a "post in the last 2 days" kind of search from brian phil. I sometimes come in and have to leave or get booted and the "post since you were on" resets. When someone else post in then it comes back up as a current post and I find it. Sorry for the delay Entropy.

There is no magic in anything I do. Its not incrediable at all. It's just different from the pic dimmers. With a fast enough pwm you can average out any waveform.

DynamoBen

12-25-2007, 05:11 PM

I've only ever done phase angle dimming any disadvantages?

12-26-2007, 02:18 PM

I would not say disavantages to the pic dimmers. It just dims nice and it removes some of the issues with LED's which was my target.

DynamoBen

12-26-2007, 02:39 PM

It is an interesting method. I can see where it might be easier from a code side. The only downside I can think of is the requirement for a hefty rectifier to power the outputs.

jwells

12-26-2007, 03:44 PM

RJ from all the boards I read people have issues with dimming the 1/2 wave led lights with LOR and AL which it I read correctly are most of the led's at the local stores like Lowes Home Depot Target. I read the boxes and I cannot tell if they are 1/2 or full wave. On PC the 2008 group buy led's are supposed to be full wave and not have issues dimming. Can you shed any light on this area since I think you are using led's ? I wanted to get some that are on sale but most are gone from what I see today.

12-27-2007, 01:52 PM

They often are not marked. On your lynx it will not matter any at all as the power is fully rectified before going to the lights. If you plug a set of led's in and they don't light turn the plug around. If they light then you know they are half wave. Full wave will not matter which way they plug in. On the lynx they will all be Full wave lights. So don't worry about which you buy. Just make sure they are ones that dim well. Out of 11 different sets I bought and tested only one did not dim well and it was some C9 led's with capacitors in them. They dimmed without flickering they just had a shorter range of dimming.

Entropy

01-04-2008, 08:24 AM

I've only ever done phase angle dimming any disadvantages?
If it's being done the way I interpret what he's saying (a high frequency PWM waveform, much higher than the 60 Hz AC frequency or the 120 Hz ripple of rectified AC, which removes the need to phase lock the dimming waveform to the AC), then it will produce a hefty amount of EMI. High frequency PWM is OK as long as you don't push it through a long cable (like a string of lights) - In which case the long cable begins to behave as an antenna.

How much EMI? No idea, probably not enough to be a big problem as long as a ham radio operator doesn't live near you - they won't be happy if you splatter harmonics over the HF bands. Highly unlikely that you will ever be able to take the project commercial if you decide to do so - FCC approval for such a device probably won't happen.

Trepidati0n

01-04-2008, 10:05 AM

Entropy,

It might not be as bad as you think. Long strings of lights are fairly inductive and thus tend to take choppy things and make them smooth. This same principle is used in buck converters. This noise is also further reduced because the currents return the same path.

01-04-2008, 01:14 PM

mrpackethead

01-04-2008, 04:48 PM

This is an interesting design in deed.. However without schematics and perhaps code its hard to make truely informed comment.. Perhaps RJ might like to post the detail of his design sometime.. I'm sure RJ is confident that the design will stand up to being scrutinized by other people in this forum.

The Cheap LED strings i looked at, where simply half wave rectified, with a current limiting resistor..

An interesting side-bar thread to start is looking at the lifetime and efficiency of LEDS when subjected to high ripple currents. Manufacturers of 'cheap' led strings make a considerable trade off between cost and lifetime and light output. Above approx 40% P-P ripple, LEDS can experience more heating during the 'peaks' than 'cooling' during the valleys, resulting in higher die temp and a reduction in life time and output

In some circumstances there can be a measurable degregation in light output after just a few hours of operation, and a reduction of lifetime by 90-99%.. However many leds now have expected life times of 100k hours.. If you reduce that to even 1000 hours, blinking on and off, its going to be a lot of christmas's before they flinally blink off for the last time.

As with everything, theres design comprimises.. If long life time and high are issues, then you need drive circuitry that meets or the operating parameters of the device.. If you are prepared to throw away light efficacy and lifetime, but get the cost right down, you build them like the walmart/home depot super cheap strings.

From what i have read, it looks like the Lynx design follows the later design.. its all about driving cost down, ( RJ you might like to comment? ), but comprising on life time and light out of the LEDS. Don't get me wrong, the design is valid, it just depends on your requirements if it will meet your needs. I think its good to be aware of the benfits ( low cost ) and limitations ( short lifetime ) of the systems.

For me, i'm primarly working with higher power leds ( 1/3 W ), and they really have to be treated properly, if you want them to last any time at all.. For example a 3W white Lumiled operated on rectified AC, ( and current limited with a resistor appropriately ) like the lynx will do, will within 30 mins be at less that 1/4 of its specificed output..

RJ i look forward to your responses, and hopefully some detail of what you are doing... Keep up the good work.

Trepidati0n

01-04-2008, 05:32 PM

One thing RJ has mentioned is he is looking towards the future. I think we can assume that 1/2 rectified LED's will be gone or at least a minimum in a short period of time.

mrpackethead

01-04-2008, 05:41 PM

half or full wave rectified, there is still large current ripple across the LEDS.

Trepidati0n

01-04-2008, 05:54 PM

Then maybe I'm confused....or you may be confused.

If you talking AC lamps...then where is the problem?

Now if youare talking typical DC lamps...that is easily solved with the lynx...you just run all 16 channels as DC channels and power the Lynx outputs with DC (except for the digital part which is powered seperately).

Now if you are talking current sensitive lamps (one that like constant current)...then the lynx won't do what you want it to do directly. What you are talking about is current regulation which is a whole different animal....just not a bad animal though ;) However, the Lynx could be "hacked" to look like a buck converter where you bascially power hte lynx outputs with DC, PWM it...then put an inductor on the output. If you add a small cap to the output of the inductor, you now have a very stable and clean current source for the lamp that can be controlled via PWM. If you want fast response time from the LED..then add a burden style resistor. The lynx is EXTREMELY flexible from what i've seen so far..and the reason why I have coming on this coop and four more for the next.

Right now I see almsot no limitations with the lynx.. My only limit right now is my imagination.

Also if you look at the Lynx in DC mode...it could be used to create an 8x8 array 8)

01-04-2008, 06:29 PM

RJ i look forward to your responses, and hopefully some detail of what you are doing... Keep up the good work.

I'm to busy using all my stuff that I was told didn't work later!

:lol:

mrpackethead

01-04-2008, 07:31 PM

If you talking AC lamps...then where is the problem?

Im not talking about AC lamps.. I've been talkign about LED's..

Now if youare talking typical DC lamps...that is easily solved with the lynx...

I'm not talking about LAMPS.. I'm talking about LEDS... If we are talking about lamps with a fliament, then they can run on either ac or dc.. but we are not, we are talking about LEDS.

Now if you are talking current sensitive lamps (one that like constant current)...

Mmm, finally we get close.. We are talking about current devices.. not lamps, but LED..

then the lynx won't do what you want it to do directly. What you are talking about is current regulation which is a whole different animal..

Well, from what i have read, the Lynx does dim strings of LEDs.. The current regulation is done by a resistor in series with the LEDS though... not the lynx itself.

However, the Lynx could be "hacked" to look like a buck converter where you bascially power hte lynx outputs with DC, PWM it...then put an inductor on the output. If you add a small cap to the output of the inductor, you now have a very stable and clean current source for the lamp that can be controlled via PWM.

Mmm.. sounds like a major hack. How are you going to provide feedback to the buck converter to know when to turn and off... ( fundemental to making it work.. ).. putting a cap across the output, does have the function of reducing the ripple, however it renders the circuit not suited for PWM... ( a cap big enough to reduce the ripple, will render the PWM useless. ).. So in order to reduce ripple you need to use a bigger inductor...

As far as i can see the the lynx will dim led strings.. just not in a way that will keep them running brightly for a long time.

Right now I see almsot no limitations with the lynx.. My only limit right now is my imagination.

How woudl you propose dimming 9 3W Luxeon Stars.. ( I=700mA, Rd=1 ohm, Vf = 3.7V ).. Die temperture never to exceed 70C, and ripple current less than 30%?

just remember, that rarely does one solution fit all problems.

mrpackethead

01-04-2008, 07:35 PM

I'm to busy using all my stuff that I was told didn't work later!

I dont' think i said it did'nt work.. Based on the little information i have ( no schematics / diagrams / code ), its hard to know how it works.. and so we have to make some conclusions based on our general knowledge of electronics and from what you've said..

A schematic would put all to rest woud'nt it?

Does the lynx provide current limited dimming with a less than 30% ripple suiable for LEDS?

mrpackethead

01-04-2008, 08:03 PM

After thinking through how and what the lynx is, i think what RJ has done is quite clever, though perhaps not orignal. . its one of the new techniques in industrial/commerical dimming called 'Sine Wave Dimming'..

Google 'Sine Wave Dimming' - theres a lot of references about it..

A sine wave dimmer can best be described as an electronic autotransformer. The output is a sine wave with continuously variable amplitude. The deformation (distortion) of the sine wave is very small, typically 1-2%.

Sine wave dimming is really a misnomer, as this kind of dimmer will output whatever waveform supplies it. Feed this dimmer with a square wave, and the output will be a square wave, too. “Variable amplitude by means of pulse width modulation” is a better description, but that hardly qualifies as a marketing buzzword hence, “sine wave dimming.”

In a sine wave dimmer, the incoming power is chopped up into many slices of varying width to create an output that is essentially a true sine wave with variable amplitude. This sounds simple but requires very precise control of the IGBTs and constant monitoring of currents, voltages and temperatures to ensure reliable operation.

In a sine wave dimmer, the IGBTs are being used as fast switches, and the losses are much smaller than in FPC or RPC dimming. Sine wave dimmers are almost as efficient as SCR dimmers, around 98% at 230VAC and 96% at 120VAC.

The switching of the IGBTs is done at a high frequency, typically 30-50kHz, and while this creates electrical noise, the filters needed to eliminate this noise are a lot smaller than the bulky filter chokes needed for effective filtering of a comparable SCR dimmer. Sine wave dimming practically eliminates the main problems of FPC and RPC dimming: noise and harmonics.

The advanced technology of sine wave dimming makes it possible to offer many new and useful functions and features on dimmers. For example, sine wave dimmers can offer instantaneous overload cut off and the ability to handle indefinite shorts on the load with a return to full function as soon as the short is removed. Circuit breakers are fitted in order to comply with electrical regulations but should only trip in case of serious malfunctions.

So RJ is this close to what you've done?

DynamoBen

01-04-2008, 09:14 PM

Well I hate to burst your bubble but Lynx isn't a sine wave dimmer. How do I know, I worked for one of the leading manufacturers of theatrical dimming and one of the newest products in their line was a sinewave dimmer.
http://www.silentdimming.com/

What RJ is doing is DC dimming, still very creative.

mrpackethead

01-04-2008, 10:17 PM

Hey ben, have you got the low down on the lynx then? Details are a bit sketchy so, i'm trying to put 2 and 2 together to get 10.. If RJ would simply produce a schematic, then we could put doubt to rest...

On the topic of sinewave dimming, i first ran into a similar technique not with lighting dimmers, but with UPS's which produce proper sine wave outputs, running full double conversion.. similar but quite different..

01-04-2008, 10:37 PM

Hey ben, have you got the low down on the lynx then? Details are a bit sketchy so, i'm trying to put 2 and 2 together to get 10.. If RJ would simply produce a schematic, then we could put doubt to rest...

I'm sorry!, I was unaware there is any doubt to anything. I do not believe I have asked you for anything. I do not believe I require your approval of my design nor do I want it. I am for sure I do not owe you any explianation of any kind.

For someone wanting something from me on the other hand, you could not have appoached it in a worst manner.

RJ

DynamoBen

01-04-2008, 10:39 PM

RJ has said in the past that he isn't interested in posting a schematic of his design at this time; and may never. Some details exist in the forum (and off forum) but you will have to do some searching.

In the end he has come up with a design that works well. At this point he isn't interested in sharing the gorey details about how it works and I respect him for that.

mrpackethead

01-05-2008, 03:39 AM

the doubt is about how the lynx works.. But if you dont' want to share you design, your quite entitled to.. Maybe i missed the point of what DIY christmas was all about....

daviddth

01-05-2008, 04:54 AM

I have been reading these pages with interest. While the internals of the grinch, 595, and even Renard are well known, the Lynx is not. After following the Renard changes over the last year, I can actually follow why RJ does not want everyone having input into his design, or the way it works.

RJ your doing a great job, and while I dont necessarily agree with you keeping the code and the way things work close to your chest (Yes I know some of the info is out there and I have been reading it), I can accept that its your choice.

If people dont want to build this design because of the way it has been offered, then they should just chose another design.

Trepidati0n

01-05-2008, 12:20 PM

Mmm.. sounds like a major hack. How are you going to provide feedback to the buck converter to know when to turn and off... ( fundemental to making it work.. ).. putting a cap across the output, does have the function of reducing the ripple, however it renders the circuit not suited for PWM... ( a cap big enough to reduce the ripple, will render the PWM useless. ).. So in order to reduce ripple you need to use a bigger inductor...

You do not need feedback to make a buck conveter work. The standard switching converter is open loop as a fundamental topology. The only reason to close it is handle things like input ripple, load changes etc. If you are running with a stiff input and a fairly fixed output...closed loop regulation has very little benefit.

As for cap/ripple reduction comment...again I remain confused since it does not make sense in how switching converters work unless you are going to extremes. Remember, we only need a time constant in the range of 5 msec.

There are many ways to skin this cat...some ways better and some just different. The Lynx freestyle might be more what you need since the output could be tied to an RC time constant and piped into an opamp where you could implement DC current control. This would limit the ripple and allow you to safely drive the device in question.

DynamoBen

01-05-2008, 02:07 PM

Maybe i missed the point of what DIY christmas was all about....

We all need to remember what DIY Christmas is about. Its not about necessarily sharing every detail of a design. It’s about making projects (like lynx) available to folks who can build them at home or making a less expensive equivalent to a commercial product. In some cases that means schematics and source code but in others it might just be a kit.

In the future we may see designs that are entirely SMT. For most folks SMT is not something they would want to do themselves at home so we may have projects that come fully assembled. Only time will tell if this will be in the case. Regardless it’s still DIY because you are not spending thousands of dollars on a commercial product.

Everyone here shares the same goal, from those that design the projects to those that build them; we are all just trying to make interesting displays and save some money doing it.