Trying to make a transistor cancel another transistor

[hevans1944]

You can replace the 2N2222 NPN transistors with the 2N5401 PNP transistors if you reverse the LEDs and substitute a negative 4.5 V supply for Vcc.

 

[David Lacroix]

Thx again for all the tips! I do appreciate everybody taking on their time to draw schematic, it helps a lot!

As for the LM3914, I think I'm going to start slow with transistors, I want to stay as analog as I can and I would like to do all my wiring in the case something is wrong I made it all myself so I can untangle the mess. 8 LEDS might be a lot,I just came up with the number, could be 6, could be even 4, it does not need to be precise, it's not a thermometer for the NASA. As long as it can make the difference between -10 Celsius and +25 Celsius it should be good. The project is basically just an excuse to play around with components. I like CDRIVE's idea of getting the LED progressively brighter, I will look into that also! I plan to use a NTC-MF52 10K thermistor. The scale would probably be from -30 Celsius to +30 Celsius.

I don't have much requirements for the thermometer. As long as all the red LEDs are not lit up when there is snow outside and secondly, as long as the damn thing is not on fire, it should be all fine!

 

[CDRIVE]

A lot simpler than mine. Can you do it with PNP's like he wanted?

Sure, just reverse the battery, and all LED polarities.

Chris

 

[dorke]

After you have played around with discrete transistor and learned what you can.
It would be time for a more general lesson:
Comparators 101.
These are very basic and important building blocks with numerous useful applications.

The thing you are trying to build is basically a classic job for comparators.
The solutions you were given are actually implementations of comparators with transistors.

Here is a "rough" 4 LED thermometer implementation using a quad LM393 ,a very common IC.
+V can be 4.5V ( 2 -36V is allowed).
Learn it, play with it,complete it- (R1 and R2 values) and you will be greatly rewarded.

 

[CDRIVE]

Ah "Compactors"! Another example of Spell Check fail?

Chris

 

[dorke]

Yes @CDRIVE, I 'm correcting that

 

[dorke]

Here's a discrete concept you may like or not. It uses transistors in their linear mode. It should not be considered a finished circuit but does demonstrate a discrete method of lighting LEDs in a bar graph fashion. IE, when the next led lights the previous led does not go off. This better mimics a mercury thermometer anyway.

All transistors are configured as Common Collector Amplifiers (Emitter Followers). In this configuration a transistor cannot provide voltage gain but does provide current gain. They also provide a low output impedance (Emitter to GND) as well as relatively high input impedance (Base to GND). P1 is set at 100%. P2 is set to 80% and P3 is set at 60%.

As you can see from the LED current curves each LED gets progressively brighter until the next LED up the ladder begins to turn on also.

If learning to design Transistor circuitry is what you desire then give it a go. As I said earlier, I'm not claiming it's the best choice around but it certainly demonstrates transistors as current amplifiers.

Chris

View attachment 29389
View attachment 29390

@CDRIVE ,What "tool" are you using for drawing and simulating?
Looks like a nice one.

 

[CDRIVE]

I use Tina Classic but it's not freeware. A freeware version of Tina called Tina-TI can be downloaded from the Texas Instruments website.

BTW, I see your spellcheck nailed you again.... "Drowning"?

Chris

 

[hevans1944]

I got all that pretty-print schematic drawing stuff too, but I think @dorke got it right: I drown trying to find components, place them, and wire them up pretty with all the right labels. Only do this as a last resort if I want to build a <gasp> SPICE model or lay out a printed circuit board.

 

[CDRIVE]

That's why I like Tina. Finding parts and drawing schematics is fast. Very fast!

Chris

 

[hevans1944]

The thing you are trying to build is basically a classic job for comparators.

No, it's basically a classic job for an analog-to-digital converter and a microprocessor.

Since the NTC thermistor is more or less linear over the temperature range in question, it shouldn't require anything more fancy than an 8-bit ADC. Test the range of the A/D conversions and bin them into eight outputs according to temperature, one for each LED you want to light up. Program it so they light in sequence and stay lit (like a real thermometer) or just one lights up depending on what bin the input falls into.

You won't learn much about electronics, but Microchip PIC microprocessors are tons of fun in their own right... and very inexpensive once you have laid out the fifty bux or so for the PICkit 3 programming pod. Make sure you purchase this pod and the chips from Microchip or an authorized distributor because there are a lot of Asian counterfeits floating around.

And if you are really lazy, or just anxious to get something up and working, just purchase an Arduino Uno and program it from your desktop or laptop computer. All the gazintas and gazoutas you need are right there on the Arduino. Just add some LEDs and a thermistor and a few resistors. Voila! An LED thermometer to impress and amaze your friends with your newly acquired electronics prowess. Plenty of "sketches" available on the Internet for this sort of thing, so you don't even have to roll your own code. What's not to like?

 

[dorke]

I use Tina Classic but it's not freeware. A freeware version of Tina called Tina-TI can be downloaded from the Texas Instruments website.

BTW, I see your spellcheck nailed you again.... "Drowning"?

Chris

Yep, we did it again.
I can say the spellchecker(and the late hour) is Drowning me indeed
BTW,If you are around,what do I need it anyways

Will correct it ,thanks.

 

[dorke]

No, it's basically a classic job for an analog-to-digital converter and a microprocessor.

Since the NTC thermistor is more or less linear over the temperature range in question, it shouldn't require anything more fancy than an 8-bit ADC. Test the range of the A/D conversions and bin them into eight outputs according to temperature, one for each LED you want to light up. Program it so they light in sequence and stay lit (like a real thermometer) or just one lights up depending on what bin the input falls into.

You won't learn much about electronics, but Microchip PIC microprocessors are tons of fun in their own right... and very inexpensive once you have laid out the fifty bux or so for the PICkit 3 programming pod. Make sure you purchase this pod and the chips from Microchip or an authorized distributor because there are a lot of Asian counterfeits floating around.

And if you are really lazy, or just anxious to get something up and working, just purchase an Arduino Uno and program it from your desktop or laptop computer. All the gazintas and gazoutas you need are right there on the Arduino. Just add some LEDs and a thermistor and a few resistors. Voila! An LED thermometer to impress and amaze your friends with your newly acquired electronics prowess. Plenty of "sketches" available on the Internet for this sort of thing, so you don't even have to roll your own code. What's not to like?

Don't agree on that one.
That will totally miss the point of learning analog electronics and concepts.
It will move the all thing into the "software realm" which isn't the point here at all.

One thing I do agree on,
the implementation I suggested is indeed a kind of A/D...a flash A/D is build that way.

 

[CDRIVE]

If Dave is going make that monumental leap to micro's I'd advise him to go Picaxe! Every aspect of the Picaxe system, initial cost, programming hardware, chip cost vs arduino, learning curve and support will be the most painless.

That said, it's complete nonsense to make a giant leap like that when we're talking about building a basic electronics foundation.

Since (as Hop stated) Thermistors are linear devices David could demonstrate its action with an inexpensive analog (D'Arsonval) or digital panel meter and a few resistors. Back in the day I used to disassemble movements, remove the faceplate and mark them with dry transfer numeric/letter kit. I had to do this for the "S Meter" on my old Hallicrafters SX-42.

Note! Absolutely no coffee should be consumed prior to doing this kind of work! In fact a shot of Jack Daniels might be beneficial!

Chris

 

[hevans1944]

David could demonstrate its action with an inexpensive analog (D'Arsonval) or digital panel meter and a few resistors.

I hadn't thought of THAT! But in keeping with the analog idea, a digital panel meter is out. So all we need to do is convert the needle deflection into signals that will drive eight LEDs... maybe a tiny mirror glued to the needle to deflect a beam of light onto eight LDRs each wired to turn on an LED when the light illuminates them? Where the hell is Rube Goldberg when you need him?

 

[dorke]

@CDRIVE, my time to correct you.
Thermistors are not linear devices !
They are in fact exponential ones in regards to temperature.

 

[CDRIVE]

@CDRIVE, my time to correct you.
Thermistors are not linear devices !
They are in fact exponential ones in regards to temperature.

Touche!

Can I claim that Spell Check confused "linear" with "analog"?

Chris

 

[dorke]

Yes you can...

 

[hevans1944]

@CDRIVE, my time to correct you.
Thermistors are not linear devices !
They are in fact exponential ones in regards to temperature.

@dorke don't blame Chris! See my post #51. NTC thermistors exhibit an exponential inverse relationship, actually. See this ancient papyrus from the 1960s. If you read this paper through to the end you will understand why I never aspired to be a computer programmer. However, everything is linear if you confine your inspection range to a short enough interval! I learned that little tid-bit while taking a calculus course. However, sometimes external "compensation" is needed.

There is a little company, YSI Incorporated, located near Dayton in Yellow Springs, OH, that got their start making "linearized" thermistor-based temperature probes for the U.S. Government (among other customers) many years ago (1960s IIRC). They had a lot of patents that are probably expired by now. These are the folks who also invented the glucose blood-sugar monitoring strips, one version of which I use on a daily basis to monitor my Type 2 diebetes. They are now heavily involved with environmental monitoring, present day.

It's a very "green" company located on the outskirts of a well-known hot-bed of radical, liberal, socialist, maybe even Communist activity. I love that little town! Great food and art there. It's politics, not so much. It is a gateway to John Bryan State Park, where my second wife and I were married in 2000, and to Young's Dairy where real Jersey cow milkshakes are served. Anyway, YSI went on to bigger and better things after milking linearized thermistors for all they were worth in government contracts, and Omega Engineering (and perhaps others) picked up the baton. See this web link for what Omega is doing today. Thermistors can be made "linear enuf" for guv'ment work. And so much easier to do so with a microprocessor. The OP should hurry to catch up.

Edit: Corrected post #50 to post #51 per Chris' comment below. Corrected spelling for John Bryan State Park.

 

[CDRIVE]

That would be post 51 Hop. Yes, I did parrot you when I posted that misspeak but since Constantine didn't catch it when you said it I took the heat solo. ... You owe me for not snitching.

Chris