lindner@hppad.HP.COM (Steffen Lindner) (11/24/90)
I would appreciate it if any of you could help me with the following: Background Information: I am trying to built a digitally controlled light dimmer. This system should control about 24 individual channels. Each channel consists of one or more lights (incandescent). The maximum load on each channel is about 3A. The dimmer must allow a minimum of 16 brightness settings (more would be better). The system uses a microcontroller for I/O and triggering the TRIACs. All seems well except that I have not been able to come up with a simple, cost effective solution to reduce (eliminate) noise. Problem: A regular dimmer setup (controlling the phase angle of a TRIAC) produces substantial current spikes, since the load current is turned on under non-cross-over conditions. This causes a lot of interference. Another solution would be to always switch at cross-over points and just skip one or more half cycle. This works well as long as not too many cycles are skipped. At 1/16 of full brightness there would be a substantial flicker noticable. Are there any better methods of building a dimmer? Is there any way to trigger the TRIAC "gently" (ie. soft start)? Please help, Steffen
ftpam1@acad3.fai.alaska.edu (MUNTS PHILLIP A) (11/25/90)
In article <2570007@hppad.HP.COM>, lindner@hppad.HP.COM (Steffen Lindner) writes... >I would appreciate it if any of you could help me with the following: > >Background Information: >I am trying to built a digitally controlled light dimmer. This system should >control about 24 individual channels. Each channel consists of one or more >lights (incandescent). The maximum load on each channel is about 3A. The >dimmer must allow a minimum of 16 brightness settings (more would be better). >The system uses a microcontroller for I/O and triggering the TRIACs. All >seems well except that I have not been able to come up with a simple, cost >effective solution to reduce (eliminate) noise. > >Problem: >A regular dimmer setup (controlling the phase angle of a TRIAC) produces >substantial current spikes, since the load current is turned on under >non-cross-over conditions. This causes a lot of interference. > >Another solution would be to always switch at cross-over points and just skip >one or more half cycle. This works well as long as not too many cycles >are skipped. At 1/16 of full brightness there would be a substantial flicker >noticable. > > >Are there any better methods of building a dimmer? >Is there any way to trigger the TRIAC "gently" (ie. soft start)? > > > Please help, > > Steffen I once had occasion to replace the SCR's in a commercial stage light controller. That particular system had toroids (inductors) larger than glazed donuts. I don't remember what the associated capacitors were; the coils were so large that they monopolized your attention! (This system was rated at 2.4 kW per channel. Each channel had its own LC noise filter.) Philip Munts N7AHL NRA Extremist, etc. University of Alaska, Fairbanks
rsnider@xrtll.uucp (Richard Snider) (11/27/90)
In article <2570007@hppad.HP.COM> lindner@hppad.HP.COM (Steffen Lindner) writes: >control about 24 individual channels. Each channel consists of one or more >lights (incandescent). The maximum load on each channel is about 3A. The >dimmer must allow a minimum of 16 brightness settings (more would be better). >The system uses a microcontroller for I/O and triggering the TRIACs. > >Problem: >A regular dimmer setup (controlling the phase angle of a TRIAC) produces >substantial current spikes, since the load current is turned on under >non-cross-over conditions. This causes a lot of interference. While on the topic of problems, depending on the type of incandescent lights you are using, you probabbly will need some form of "preheating" current through the fillament in the bulbs if you want them to come on from some minimum setting without "popping" on. This is becuase the current required to get the filament glowing from a "cold" condition is considerably higher than that which is needed to keep it glowing. Your idea of skipping cycles to dim the bulb will work, you can also trigger on half cycles as well, to cut down on flicker. If you are expecting these bulbs to be used in any kind of lighting situation where people are not looking directly at the bulbs, then the flicker won't be noticed at all. Also, if dimming to 1/16th, chances are there won't be much light to look at anyway, since that is less than 10V and most 120V incandescant bulbs (I assume you are dimming 120V circuits) will not light up appreciably (sometimes not at all) with that kind of voltage through them. >Is there any way to trigger the TRIAC "gently" (ie. soft start)? Yes, use an inductor in series with your triac and load, this will remove a considerable amount of the spike associated with turning the Triac on during part of the cycle. The calculation of the inductance is left as an excercise for the student :-) .....Rich rsnider@xrtll
johnf@hpscdc.scd.hp.com (John Flowers) (11/27/90)
A friend of mine works for Teatronics, Inc. Teatronics makes multi-channel analog and digitally controlled dimmer "packs", from 4 to 24 channels. In the models that carry a 2.4 kW load they have LARGE inductors on the outputs of the TRIACS. These inductances are not measured in Henries but in Seconds (of rise time). For a typical 2.4kW (20A) channel, the inductor is spec'd to not be less than 500 uSeconds (micro). I hope you can figure out what inductance this translates to for your system. ****************************************************************** * John "Microwaves are for cooking" Flowers /_/ \/\/ * * johnf@scd.scd.com, hplabs!hpscdc!johnf / * * HP Santa Clara Division, Microwave Counters Manufacturing * ******************************************************************
hgw@julia.math.ucla.edu (Harold Wong) (11/27/90)
In article <1990Nov24.205337.18959@hayes.ims.alaska.edu> ftpam1@acad3.fai.alaska.edu writes: >In article <2570007@hppad.HP.COM>, lindner@hppad.HP.COM (Steffen Lindner) writes... > I once had occasion to replace the SCR's in a commercial stage light >controller. That particular system had toroids (inductors) larger than glazed >donuts. I don't remember what the associated capacitors were; the coils were >so large that they monopolized your attention! > Could someone explain the uses of toroids? Specifically, For what purpose should they be used for. Where should they be used? How should they be used? What are the differences between the donut vs bar version? How do you determine the size you need? Should I wrap all my wires around toroids? Any inspiration would be appreciated. Thanks. Harold ------------------------------------------------------------------------------- Harold Wong (213) 825-9040 UCLA-Mathnet; 3915F MSA; 405 Hilgard Ave.; Los Angeles, CA 90024-1555 ARPA: hgw@math.ucla.edu BITNET: hgw%math.ucla.edu@INTERBIT
bender@oobleck.Eng.Sun.COM (Michael Bender) (11/27/90)
In article <2570007@hppad.HP.COM> lindner@hppad.HP.COM (Steffen Lindner) writes: >I would appreciate it if any of you could help me with the following: > >Background Information: >I am trying to built a digitally controlled light dimmer. This system should >control about 24 individual channels. Each channel consists of one or more >lights (incandescent). The maximum load on each channel is about 3A. The >dimmer must allow a minimum of 16 brightness settings (more would be better). >The system uses a microcontroller for I/O and triggering the TRIACs. All >seems well except that I have not been able to come up with a simple, cost >effective solution to reduce (eliminate) noise. A few years ago I built an 8-channel digitally-controlled dimmer. I used an 8749 for the CPU, and Intel 8254 (right number??) counter/timer chips to actually control the firing of the Triacs. A zero-cross detector would feed the CT chips' reset (clear?) input, a higher-frequency clock would actually clock the CT chips, and the 8749 would load values into the CT registers, which determined the duty cycle of the Triac-controlled load. It worked quite well. The CT chips would be used in auto-reload mode, so that every time they got a reset (clear??) from the zero-cross detector, they would start counting again; after a programmed number of counts, they would trigger the Triac and it would turn on for the remainder of the cycle. mike -- Won't look like rain, Won't look like snow, | DOD #000007 Won't look like fog, That's all we know! | AMA #511250 We just can't tell you anymore, We've never made oobleck before! | MSC #298726
ftpam1@acad3.fai.alaska.edu (MUNTS PHILLIP A) (11/28/90)
In article <3450@exodus.Eng.Sun.COM>, bender@oobleck.Eng.Sun.COM (Michael Bender) writes... > >A few years ago I built an 8-channel digitally-controlled dimmer. I used an >8749 for the CPU, and Intel 8254 (right number??) counter/timer chips to >actually control the firing of the Triacs. A zero-cross detector would feed >the CT chips' reset (clear?) input, a higher-frequency clock would actually >clock the CT chips, and the 8749 would load values into the CT registers, >which determined the duty cycle of the Triac-controlled load. It worked >quite well. The CT chips would be used in auto-reload mode, so that every >time they got a reset (clear??) from the zero-cross detector, they would >start counting again; after a programmed number of counts, they would >trigger the Triac and it would turn on for the remainder of the cycle. > Better yet, use an 8751 and do it all in software. An 11 MHz 8751 can support 16 channels at 128 brightness levels. It also has a serial port for easy interfacing to a host computer. What did you use for a zero-cross detector? My latest is an AC input optocoupler, but the current transfer ratio is critical for proper operation. (8751 port pins require a fair bit of current to pull down.) Philip Munts N7AHL NRA Extremist, etc. University of Alaska, Fairbanks
bender@oobleck.Eng.Sun.COM (Michael Bender) (11/28/90)
In article <1990Nov27.201653.4865@hayes.ims.alaska.edu> ftpam1@acad3.fai.alaska.edu writes: >In article <3450@exodus.Eng.Sun.COM>, bender@oobleck.Eng.Sun.COM (Michael Bender) writes... >> >>A few years ago I built an 8-channel digitally-controlled dimmer. I used an >>8749 for the CPU [...] > > Better yet, use an 8751 and do it all in software. An 11 MHz 8751 can >support 16 channels at 128 brightness levels. It also has a serial port for >easy interfacing to a host computer. > > What did you use for a zero-cross detector? My latest is an AC input >optocoupler, but the current transfer ratio is critical for proper operation. >(8751 port pins require a fair bit of current to pull down.) At the time I was doing the dimmer I had an 8048 ICE for my S-100 system. The 8051 ICE that my school had was under the control of some anal-retentive type that couldn't see it being used to develop a "mere" light dimmer. I think that guy is selling used cars now or something... I used a transistor/resistor off of the p.s. transformer, and as I remember it wouldn't really switch at the zero-cross point, but just a little after when the transistor turned on (this was way back before I knew anything about engineering!), so I couldn't get the triac on for the complete cycle, but it was close enough that the lights looked like they were full on. I had plans to build this great digitally-controlled light board, but never built more than the first 8 channel card. Looking back on it, an 8051 would have been a better processor to use, especially because of the built-in serial port - I was using an 8-bit parallel connection between my 8048 and the controlling computer. I guess you just learn these things as time goes on (this was in 1984 I think). mike -- Won't look like rain, Won't look like snow, | DOD #000007 Won't look like fog, That's all we know! | AMA #511250 We just can't tell you anymore, We've never made oobleck before! | MSC #298726
rando@ginger.dfrf.nasa.gov (Randy Brumbaugh) (11/28/90)
I also have done some work with light dimmers, mostly trying to fix them. (This newsgroup was very helpful). Anyway, with all this talk of digitally controlled light dimmers, I thought I would throw in a couple of things: 1) There is a standard for serial, digital dimmer control, USITT (US Institute for Theatre Technology) DMX-512. It is a 250kbit/sec RS-422/485 serial link. I have a copy and will elaborate if there is any interest. A light board which uses this protocol will control most dimmers on the market, and a dimmer can be controlled by most light boards. 2) Consider the dimmer curves. There are four, I think: - The control position vs digital output value (In the control board). I think this is usually linear. (i.e. halfway up = 128 if the range is 0-255). - The digital input to the dimmer control circuit vs. the firing of the SCR/TRIAC. - The firing of the SCR vs. average voltage output. The voltage varies throughout the AC cycle, so this is not as straightforward as it might seem. - The average voltage to the lamp vs. light output. This is surprisingly non-linear. Something like a 10% voltage drop gives a 50% light output decrease. I can look up the exact curves if anybody is interested. ( I guess there is also light output vs. perceived brightness in the human eye, which is not linear.) There are several "standard" curves - "linear voltage", "linear light", "linear brightness", etc. Some are preferred by stage, some by TV. My point is, if you are building a dimmer to mate with an existing control board, you can only control the second curve. Maybe some kind of EPROM, table-look-up would be a good idea. Maybe a switch to select from 2-3 curves. Maybe using a timer/counter directly happens to work out to a reasonable curve? Does anybody have any experience in this area? Randy Brumbaugh rando@skipper.dfrf.nasa.gov
stevem@specialix.co.uk (Steven Murray) (11/29/90)
rando@skipper.dfrf.nasa.gov (Randy Brumbaugh) writes: > My point is, if you are building a dimmer to mate with an existing > control board, you can only control the second curve. Maybe > some kind of EPROM, table-look-up would be a good idea. Maybe a > switch to select from 2-3 curves. Maybe using a timer/counter > directly happens to work out to a reasonable curve? > Does anybody have any experience in this area? I've been interested in digitally controlled light dimmers for a few years. I worked for a New Zealand company a couple of years ago, and actually got around to MAKING one. The unit we made was based on the 80C552 which is a derivative of the 8051. The first version was just 4 channels with 256 light levels. The software I wrote allowed 3 phase operation, and the plan was to make a 12 channel 3 phase unit, but it got vetoed by the boss. I had the microprocessor trigger the triacs directly, just as you are thinking of. I had a lookup table in EPROM consisting of 256 16 bit values, where the 16 bit values were reloads for the timer counter, and the 256 values were the desired brightness levels. To tell you the truth, we have only ever used one curve - linear brightness of the lamp. This is what the operators wanted and what the control desks put out, so it is what we delivered. All adjustment for the non-linearity of the bulb and triac triggering time was done in the dimmer. To generate the look-up table to produce the linear brightness curve I wrote a flash basic program that worked out the amount of energy left in each half cycle at each point in time ('area under the curve') and then allowed for the basic non-linearity of the bulbs - when you give them just a small phase angle they glow red, but produce little useful light. The work we were doing needed 256 light levels. You can get away with 64 and still have a professional product, as long as you don't intend to 'Fade' the light - with just 64 levels, it seems to 'Jump' on slow fades. On 10 minute fades with 15 watt bulbs in a dark room near the lower light levels, my boss noticed that my 256 levels 'Jumped' so I had to redo my software for 1024 levels in the end (by that stage the boss had also got me to change to a 68705 CPU because it was $1 cheaper than the best 8051 price at the time). Soon after that I left the company because I felt the boss was wasting my time! Anyway, I wouldn't advise people to try skipping half cycles to dim lamps - unless the bulb has got a high thermal inertia, it looks really bad. All the professional dimmers have good, big torroids for filtering, too. A 500us time constant is fantastic - you can get a good quality system with just 140us torroids. We used to use EI type chokes, but they are more than twice the weight and size. Also, if you want to drive halogens (through transformers usually) you have to ensure both positive and negative half cycles have the same trigger point - different by more than a couple of percent and your transformers will be very unhappy indeed. DMX-512 is neat, but the 250kbaud rate makes the software a mite tricky, especially if you intend to be doing real time triac triggering as well! - Regards Steven Murray -- Steven Murray uunet!slxsys!stevem stevem@specialix.co.uk I am speaking, but | If these are your opinions, then we are in agreement!! not for my employer.| Flames, spelling errors, complaints > /dev/null
tell@oscar.cs.unc.edu (Stephen Tell) (11/30/90)
In article <1990Nov26.181619.9855@xrtll.uucp> rsnider@xrtll.UUCP (Richard Snider) writes: >In article <2570007@hppad.HP.COM> lindner@hppad.HP.COM (Steffen Lindner) writes: >While on the topic of problems, depending on the type of incandescent >lights you are using, you probabbly will need some form of "preheating" >current through the fillament in the bulbs if you want them to come on >from some minimum setting without "popping" on. This is becuase the >current required to get the filament glowing from a "cold" condition is >considerably higher than that which is needed to keep it glowing. I note that its fairly common for lighting designers to build this into their cues, preheating lamps that are to come on in the next cue at 10% power. Unless the first cue is very dim, the soft orange glow is hardly noticable on stage. This actualy makes the transition happen faster, since the filament is alreay somewhat warm. The difference is definitely visible, and for theatrical work, it depends on the effect you want. It might also affect bulb life, I suppose. Tuning instructions for an analog dimmer I worked on once had you set things so a 10% control input (1 volt) produced an 18 volt AC output, resulting in a fairly dim but visible output from a 120 volt lamp. When turning the control up from 0, the dimmer output stayed off untill around 8 or 9%, and then came up quickly to the "preheat" value. I'm not sure if this was to ensure that if the analog stuff drifted that 0% was always definitely off, or if it is related to the cold-filament problem > .....Rich > rsnider@xrtll -------------------------------------------------------------------- Steve Tell e-mail: tell@wsmail.cs.unc.edu usmail: #5L Estes Park apts CS Grad Student, UNC Chapel Hill. Carrboro NC 27510 Former chief engineer, Duke Union Community Television, Durham, NC.
oparadis@oracle.uucp (On G. Paradise) (12/12/90)
Most of the problems with el-cheapo dimmers in those $40 halogen lamps can be solved by: 1. Adding a resistor in series with the triac gate, to get rid of potentiometer backlash. 2. Getting the diac trigger voltage through a double RC network, to obtain zero through 360 degrees of conduction. 3. Adding a pi (CLC) network to reduce RFI. Way back in the '70s, RCA had schematics and values for these in one of their triac data books. They also mentioned: 4. Connecting a snubber network across the triac (MT1 - MT2) for dV/dt control of inductive loads. My old RCA databooks are long gone. Can anybody look thses up and post the values? On