Transistor Magic--Chasing the 96 Bit Pixel
rjo_
Posts: 1,825
Hi guys,
There are at least two of us out here that are just a little perplexed by even the simplest things... me and Russ. We can handle the Prop... we have good teachers and the learning materials are fabulous... but when it comes to transistors...
This is going to sound pretty stupid... so, you need to remember that I am genius.
Here's my story: I was building a 4 (or 5) Cog LED controller... to control a single full color led from Radioshack (part number 276-0029). Already this sounds stupid... why use 4 cogs?
Well... if I use counters, the newbies aren't going to understand it. And if I use counters, I'm going to have to stop what I am doing and look up some code, and I was more interested in the LED than in getting the code to look inscrutable.
Anyway... this particular LED produces red, green and blue...with varying intensities and using variable voltages.... so, for example the green/blue uses 3.5 - 4 V to produce 650 and 120mcdm respectively, but uses only 2V to control red. In my dreams I think that I am building a full color POV, a 96 bit pixel, or a flicker fusion device capable of producing seizures... but in reality... keep reading:
The way the LED works is that you put your +V into the long wire and connect the color you want to ground... a great place for resistors and transistors... but then I thought, why use resistors? "They waste energy and eventually I'm going to be taxed on the basis of how much energy I waste. I can control everything by using a duty cycle on a transistor between +V and the long wire...and between each color and ground "... that was my thought.
So far so good... I'm just using everything Chip and Andy taught me[noparse]:)[/noparse]
So, first I put an NPN switching transistor(RadioShack 276-1617) between the +V and the long wire on the LED....and simply connected the shorter wires to ground. Everything worked as expected ... but I could only get 3.3 Volts out of the transistor. Which isn't enough... but I COULD reduce the voltage by varying the frequency and duration of the signal going to the base of the transistor... just like in the learning labs.
Next I used an FET, hooked up to 5V+... MPF10 2(276-2062)... and then all of my illusions were destroyed... I could not get the voltage to drop below my gate voltage no matter how I varied the frequency and duration of the pulse from my Prop. So on the one hand... I can't get enough voltage... and on the other hand I can't reduce the voltage enough.
At this point, I'm only using one cog... and I'm stumped[noparse]:)[/noparse]
Remember... this isn't about LED's it's about transistors[noparse]:)[/noparse]
Rich
ps
Yes... I have books and I have been reading them... either they are too abstract, I am too concrete--or I have the wrong books[noparse]:)[/noparse]
Post Edited (rjo_) : 4/28/2009 12:32:32 AM GMT
There are at least two of us out here that are just a little perplexed by even the simplest things... me and Russ. We can handle the Prop... we have good teachers and the learning materials are fabulous... but when it comes to transistors...
This is going to sound pretty stupid... so, you need to remember that I am genius.
Here's my story: I was building a 4 (or 5) Cog LED controller... to control a single full color led from Radioshack (part number 276-0029). Already this sounds stupid... why use 4 cogs?
Well... if I use counters, the newbies aren't going to understand it. And if I use counters, I'm going to have to stop what I am doing and look up some code, and I was more interested in the LED than in getting the code to look inscrutable.
Anyway... this particular LED produces red, green and blue...with varying intensities and using variable voltages.... so, for example the green/blue uses 3.5 - 4 V to produce 650 and 120mcdm respectively, but uses only 2V to control red. In my dreams I think that I am building a full color POV, a 96 bit pixel, or a flicker fusion device capable of producing seizures... but in reality... keep reading:
The way the LED works is that you put your +V into the long wire and connect the color you want to ground... a great place for resistors and transistors... but then I thought, why use resistors? "They waste energy and eventually I'm going to be taxed on the basis of how much energy I waste. I can control everything by using a duty cycle on a transistor between +V and the long wire...and between each color and ground "... that was my thought.
So far so good... I'm just using everything Chip and Andy taught me[noparse]:)[/noparse]
So, first I put an NPN switching transistor(RadioShack 276-1617) between the +V and the long wire on the LED....and simply connected the shorter wires to ground. Everything worked as expected ... but I could only get 3.3 Volts out of the transistor. Which isn't enough... but I COULD reduce the voltage by varying the frequency and duration of the signal going to the base of the transistor... just like in the learning labs.
Next I used an FET, hooked up to 5V+... MPF10 2(276-2062)... and then all of my illusions were destroyed... I could not get the voltage to drop below my gate voltage no matter how I varied the frequency and duration of the pulse from my Prop. So on the one hand... I can't get enough voltage... and on the other hand I can't reduce the voltage enough.
At this point, I'm only using one cog... and I'm stumped[noparse]:)[/noparse]
Remember... this isn't about LED's it's about transistors[noparse]:)[/noparse]
Rich
ps
Yes... I have books and I have been reading them... either they are too abstract, I am too concrete--or I have the wrong books[noparse]:)[/noparse]
Post Edited (rjo_) : 4/28/2009 12:32:32 AM GMT
Comments
First, the NPN should have worked just fine... the Emitter to ground, the Collector to the Short LED leg, the long LED leg to one side of a current limiting resistor, the other side of the resistor to your supply.
The second thing is that a MPF102 is a JFET. JFET's work a little bit differently than a MOSFET. A JFET restricts the flow between the Source and Drain when voltage is applied to the Gate, while a MOSFET does just the opposite. As a general rule, JFET's are also very low current devices and can be damaged easily if over current-ed.
Perhaps a schematic and some code that you have tried may help.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
First of all, is is about LEDs. Unlike inductive loads, which are self-ballasting in a PWM situation, LEDs are non-inductive and at the mercy of any instantaneous current pulses that are thrust upon them by overdriving them beyond their rated forward voltage. Without a way to limit current (e.g. with a series resisistor), you risk destroying — or at least severely shortening the lifetime — of your LEDs if you don't limit the instantaneous current spikes. The simplest way is with resistors. More efficient (and more complex) techniques can involve switching current regulators and inductors.
-Phil
Thank you very much.
We have moved on from bowling to Bocce ball in Special Olympics(I am a coach... not a participant). So I won't have a chance to put the schematics and code together and post it til tomorrow night. I tried so many different versions of the circuits and code and was having so much fun doing it... that I will have to largely reconstruct what I did... which may be different than what I thought I was doing[noparse]:)[/noparse]
Rich
But you seem also to be saying that·all MOSFETs are enhancement mode.· Now why would you say that?
It's certainly true that low-current devices can be damaged· "if over current-ed" -- that is, by excessive current.· That's hard to refute -- any device can be damaged by excessive current.· Any current that won't damage it isn't excessive, eh?
But I find the original post hard to follow.· A schematic is necessary; and it's a virtual certainty (as you imply)that current-limiting resistors will be needed in any case.
Let's see that schematic!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
Post Edited (Carl Hayes) : 4/25/2009 8:20:18 AM GMT
I was up late and now I'm up early and I'm just about out the door for the day. Thanks for responding... I love a good debate. I'll be posting lots of schematics... most trying to make the thing work the way I want it to without current limiting... ordinarily a ridiculous thing to do, but you never know there might be an occasional reason. One thing I've learned about Parallax... never say never. These guys are... well ya know[noparse]:)[/noparse]
Rich
Your transistor was in the wrong place. If we look at my 'BAD' circuit below I think this is how you had it wired. The problem is that a transistor will control the current flowing through it to make the voltage on its emitter 0.6v less than the voltage on its base. So driving such a circuit with 3.3v will result in 2.7v on the emitter and driving the LEDs. If you want to use a transistor as a switch you're better off with my GOOD circuit. Here the emitter is tied to ground, any voltage above about 1v will turn the transistor fully on and allow the maximum current to flow in the LEDs.
It's worth noting that LEDs aren't really voltage devices - the amount of light they produce is dependent on the average current flow. Their specification will give two figures, the maximum average current and the maximum peak current. Exceed the latter and they will probably die when most inconvenient.
In all this the led always needs some current limiting somewhere, usually directly in it's current path. Even with narrow pulses the peak currents can destroy the led (or transistor) due either to hot-spots or some other thermal problem developing in the device. Bear in mind what happens when your software behaves badly and pushes excess currents through the led for longer periods than it can handle. If you can absolutely guarantee that the peak currents are within limits and not for excessive duration and duty cycle etc (such as ambient temp) according to the device limitations then you could certainly go for some rather low value of current limit resistance in series with the led, otherwise you will be replacing burnt-out leds usually at most inconvenient times.
*Peter*
You really need to use a current-limiting resistor in series with the LED.· This can be either above it or below it (as the GOOD schematic is drawn); most people put it above.
You expressed a desire to eliminate that resistor in order not to waste power.· That's noble, but the power will be dissipated anyway, whether in the (unsaturated) transistor or in a current-limiting resistor, and resistors are better at dissipating power.· Besides, if you saturate the transistor (by providing more base drive as Peter suggests) you will have much better control of the peak and average currents (peak current is controlled by the resistor, and average current is controlled by the duty cycle).
Why do you get better control?· Because you can easily get resistors with 1% tolerance, but the current gain of transistors is much more variable, not only from unit to unit but also with current and temperature and the whims of the gods.
Use the resistor.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
Post Edited (Carl Hayes) : 4/25/2009 2:36:40 PM GMT
The current to the led is controlled by the output voltage to the base of the transistor and the emitter resistor value.
The power dissipated is split between the transistor and the emitter resistor.
The unregulated voltage can be used to power the led without having the light intensity vary since the transistor is a constant current source.
I would recommend a slight refinement to your circuit: Instead of a series resistor to the base, use a divider. The reason for doing so is to reduce the base voltage — and, hence, the emitter voltage — so that more voltage is available to the LED. In your circuit, assuming the base resistor value is low enough that it doesn't rely heavily on the transistor's beta (almost always a bad idea), the emitter voltage will be close to 3.3V - 0.6V = 2.7V, leaving less than 2.3V for the LED with a +5V supply. If the LED's forward voltage is more than this, it won't light up.
-Phil
There are also a number of ways to wire up transistors wrong but they will partially work.
Re saving power, if you run a led off a 5V supply at 20mA you are using 5*0.02=0.1W. Not a lot of power. For perspective, the human body uses about 100W just sitting at a computer typing. But I guess if you want to use 1000 leds, you might need to think about power saving.
The 'perfect' led power source would be a switch mode constant current source. But it is hardly worth the expense considering the power savings, unless you are using the newer ultra high brightness leds for house lighting etc.
I use kwinn's circuit for leds but with a modification - the resistor is above the transistor and the transistor emitter goes to ground. Ball park 1k to 2k7 on the base, and 220R to 1k in series with the led.
There definitely needs to be a resistor in series with the led, and I wouldn't go anything under about 100 ohms. There is an exception to this with a circuit used for pulsing leds hard where you charge up a capacitor via 100R, and then dump all the capacitor's charge through the led with no resistor. This gives pulses of 1A or more, but then you have to wait till the cap charges up again. The 100R charging the cap keeps the average current to about 30mA.
Controlling the gate current of a transistor can be tricky as you would need matched transistors otherwise the brightness might change. Maybe do some experiments with 5V-> led-> 220R ->collector -> emitter -> ground, and vary the base resistor from 1k to 100k. Perhaps set up two or three next to each other and see how consistent the brightness is.
The standard way to change brightness is to pulse the led, and of course, this does save power too. So 5v-> led->220R->collector->emitter->ground, maybe 1k into the base, and try pwm from 0 to 100%.
Post Edited (Dr_Acula (James Moxham)) : 4/26/2009 4:03:19 AM GMT
my comment about MOSFETs vs. JFETs was meant to be simplistic... for the most part people think of MOSFETs as being enhancement mode.
I was also basing my answer in that JFETS are physically different in their layout construction than a MOSFET device, not just doping differences.
With a JFET, the Source and Drain are directly connected to the well in a way that does not form a PN junction. The gate connection sits on either side or both sides of the well.
With a MOSFET, the Source and Drain are connected to the well but form a PN junction at their connection. The gate connection in this case is over the well.
For both, the Well can be either P+ or N+ ... for an NMOS device, the well can be the substrate, and most often is, but that doesn't always have to be the case. A PMOS device requires an additional well to isolate it from the substrate.· For that, an NMOS device can also be constructed within an additional well.· The purpose is usually to provide·additional isolation from the ground plane if there is a circuit particularly sensitive to noise.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe
IC Layout Engineer
Parallax, Inc.
Post Edited (Beau Schwabe (Parallax)) : 4/27/2009 3:54:49 PM GMT
The reason I use this method are that it is simple, works well, and does not put additional stress on the power supply regulators. Of course when switching power supplies are used there is no unregulated voltage so no point in using this circuit.
The resistor belongs in the high side, VCC for the following reason: Most Practical.
It protects the rest of the circuitry from over current conditions from
any source, mainly those fat fingers and probes that are bound to
short out something on a pcb under development.
By placing it at the closest point to the supply, it also protects
the 'failed' transistor from becoming an issue to the rest of the
circuitry if a catastrophe does happen, aka any shorting on the PCB.
The worst case for the shorted transistor this way, is that the the led is ON always.
The base-emitter is then protected by the input resistor limiting base
current and the base-collector is protected by the series resistor in the
collector circuit.
The emitter resistor is only needed when you are trying to reflect the beta (gain)
back into the base region for amplification.
( Yes, it does need to be swamped by the passive circuitry elements
for proper use as an amplifier circuit. )
I think that most people have seen the emitter resistor in an amplifier circuit
and figure it must be in there by default during their use, which is not the case at all.
This is a full on, switching use, It has a limited need of an AC analysis.
jack
One of my favorite techniques is to use an nMOSFET (common source) or NPN (common emitter) with an LM317L+resistor in series with the LED to regulate the current. It's a bit overkill and drops 1.2V, but the current is always spot on.
-Phil
On that note I was looking at the circuit I posted after reading the suggestions and concerns everyone posted with a view to updating and improving it. The first thing I am going to do is replace R1 with a diode and see how that works. That should prevent problems if the transistor shorts and makes calculating the value of R2 for a desired current simpler.
By the way, I came up with this circuit in order to drive a large number of multiplexed 7 segment displays (28). The problem was that the display was too dim since each one was on for only about 3.5% of the time, the peak currents were already at maximum, and all the leds were soldered on a PCB making changes difficult. What I ended up doing was to enable each segment and the decimal point in turn and then turn on every digit that had that segment lit. A bit more complicated to program but now each segment is on for 12.5% of the time which makes a big difference in the brightness.
sigh.
I know that people tell you how fabulous you all are all the time...But it is really true. You aren't just a bunch of guys acting nice and informed... you are actually the nicest group of people I've seen for quite a while... and the information that oozes out of your pores is phenomenal.
I've decided to be a moderator more than just a guest here, with the goal of producing the most useful thread available for working with LED's and transistors.... looking after the interests of people just like me... really, really interested guys but just a bit befuddled. So, nothing will be wasted.
Yesterday, I went to RadioShack and bought two of every transistor they had in the drawer... which came to just under $22. The packaging is sometimes very vague, so if it doesn't work the way I expect... I can through it away and put the next one in the other way.
I also bought a red green and blue LED of every kind that RadioShack carries. That came to just under $60. So, for you budding electronics manufacturers out there... the money seem to be in the LED's.
One of the things that I noticed is that virtually all of the transistors... regardless of type... specify 5V as the base voltage. I've been pretty much ignoring this particular specification... figuring that this was a maximum. But now I'm thinking that sometimes it is a maximum... sometimes it is a minimum... and sometimes it is smack dab in the center of the functional range. ???
Brian... I love your posts. You are another one of those guys that is always right. But... for this thread, I refuse to use resistors. That doesn't mean that you shouldn't post correct solutions... everyone that reads this will want the right answer. Just that my purpose isn't just getting the right answer but having a fairly deep understanding of it.
Capacitors are fine... just no resistors. Remember, I'm trying to understand everything ... and if you guys throw resistors, capacitors and diodes into the same equation... you just lost me.
Beau... can you believe it? Being challenged on your own web site... of all the nerve[noparse]:)[/noparse]
Carl... keep it up[noparse]:)[/noparse]
Rich
My latest contribution is to follow... a little code... some salesmanship and a little less humor[noparse]:)[/noparse]
Seriously, Rich, transistors are not magic, and simply hoping that some sort of voodoo will help you get by with breaking the rules will not make it happen. Instead of shelling out for extra components, like so many clay pigeons destined for certain demise, get a copy of The Art of Electronics. It's both approachable and thorough, and it will be well worth your money and the time spent studying it.
-Phil
Post Edited (Phil Pilgrim (PhiPi)) : 4/28/2009 1:56:02 AM GMT
CON
_CLKMODE = XTAL1 + PLL16X
_XINFREQ = 5_000_000
pub main
dira[noparse][[/noparse]16] := 1
outa[noparse][[/noparse]16] := 1
repeat
I turned it on, and low and behold, an incredibly dim led. Well this was promissing, as the led was showing some light. So i change the outa instruction as follows: outa[noparse][[/noparse]16] := 0. loaded it into ram, and yay light went off. Since that was working as intended, except for one thing, why is the light so dim? so i start checking voltages with the light in the on state. Using a multimeter and the ground on the ppdb, i got 5v on the collector, ~3 on the base, and like 2.7 on the emitter. I sort of expected this because of the voltage drop that i've read is unavoidable in transistors. So i figure that all i need to do is increase the voltage to the collector and i should get increased brightness in the led. I hooked up a 12v battery because it was the only other thing i had lieing around, and i got no change in the light. When i used the multimeter, i got very conflicting results. When i use the ground on the ppdb, i don't get anywhere near the 12v on the collector pin, but if i use the ground on the battery i do. On the base pin i'm still getting the same, but here's what i don't quite understand, although i do have a hunch. The emitter pin, when using the ground on the ppdb shows approx 4v. Thats a huge voltage drop, if i understand everything right. But if i use the ground on the battery, that does go up to 5v. I'm not quite sure why all this is happening. Is there anybody that can explain? Keep in mind, other than whats built in the PPDB, i'm not using any resistors, capacitors, or anything like that. I really don't understand these devices (i know, blasphemy), so if i need something in specific places, let me know. I don't want to screw up anything, and it looks like i haven't yet. I appreciate the help ahead of time, and look forward to a response. [noparse]:)[/noparse]
Russell
That exchange is classic!
Hey Rich, how about a really long lead on the LED?
(bada, boom!)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
Chat in real time with other Propellerheads on IRC #propeller @ freenode.net
Safety Tip: Life is as good as YOU think it is!
Transistors are always refreshing, a breath of fresh air; simplicity itself.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
I notice the lack of mention of the current limiting resistor from the prop pin to the transistor. Transistors are great devices that get to fail it all kinds of exciting ways. like LEDs, no current limiting resistor and quicker than you can say "whoops I forgot a resistor" your transistor is now a device that will never fully saturate/never fully discharge. I learned long ago about that, USE RESISTORS!
Hmmm -- I think I'll go play a game or two·of pool.· But I refuse to use a cue stick.
Dang, it's way past my dinner time, but my roast turkey doesn't seem to be done yet, and it's been hours and hours.··Of course I refuse to use any heat.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
· -- Carl, nn5i@arrl.net
Russ
Cheers!
Paul Rowntree