Transistor switches for LDR "vactrols"

[nandrewjackson]

Heres a circuit I got up and running for easy switching of of multiple LDR "vactrols".


It shouldve operated at the full 5vDc but the PNP's stayed on constantly, wouldn't switch off when switch was grounded. So a simple voltage divider between the regulator and the transistors to bring the DC to about 4.7 allowed it all to function properly.


Actually, off the top of my head, I can't recall if the PNPs or the NPNs are normally conducting until the switch is grounded, or normally off until the switch is grounded.  But that's the magic of this circuit. A tiny voltage/current being shunted to ground instantly switches lots of LEDs on and off simultaneously.


I'm not sure how many vactrols (LEDs) can be on each transistor, I only made the circuit to fit one amps needs.

[PRR]

> It shouldve operated at the full 5vDc but the PNP's stayed on constantly

You've got a long sneak-path. I'm not going to work it out but indeed the PNPs want to stay on. The slight-trim "fix" is likely not stable.

[PRR]

Suggestions:

Heater power for g-amps is normally carefully balanced (two 100r resistors) for minimum hum. Making DC from it and grounding any point on the DC will seriously unbalance the AC, make hum. There is no real reason to ground the DC side. Use ungrounded jack.

(Also there's no good reason to have even minor switching transients flowing through chassis.)

LEDs (AND their switching logic) should only need steady power, not regulated power.

The NPN-PNP pair seems a clever way to get yes/no outputs but, as seen, when put together is prone to sneak-paths. And remember that whole computers were built using only 2N404, single-polarity throughout. Note that every transistor is an inverter. If you need the "not" of a function, use an inverter.

Shared resistor across multiple LEDs invites unequal currents. Especially when they are different parts (naked vs modules) doing different things (eye-candy vs switching). Your 330+100s is over-complicated; just give each LED its own resistor.

[PRR]

me> whole computers were built using only 2N404, single-polarity throughout.

Those plans seem to have missed being put on the internet. Also while the '404 filled many early transistor computers, alternate parts of similar performance were widely used.

I did find Wireless World DIY Computer from 1967. 400 Ge PNP did all the logic. (A Silicon type handled the high voltage with low leakage for neon lamp display.)
https://hackaday.com/2015/10/29/400-transistors-and-1800-resistors-form-this-1967-personal-computer/
http://www.smrcc.org.uk/members/g4ugm/Manuals/wirelessworldcomputer.pdf   (4MB PDF file)
This ran bipolar supplies, common in Olde Days, to handle the low Vbe and high leakage of GE parts. With '2222 we can do straight single-supply RTL.

[nandrewjackson]

Thanks PRR.


I figured it was some voltage leaking from NPN over to PNP that was interfering with PNP's operation. They each worked well at 5 V on breadboard when completely separate. When joined together, I tried lots of values of biasing resistors, until somehow I came up with lowering the voltage. I didn't understand the "logic" of why lower V allowed it to work, I just went with it.


Using another NPN as an inverter is new to me, I'll breadboard it out sometime soon. The only inverters we ever used in class were TTL chips. We didn't ever use any single transistors. Ever.   

[nandrewjackson]

The classes I had where we used a breadboard, we used and, nand, nor, xor, inverters, like Lego bricks. We learned a bit I guess, but seeing a single transistor with the base feeding off the collector of another one is like egyptian heiroglyphics to me. 

[shooter]

a bit

YES, you got it, 1 or 0 

[nandrewjackson]

Nice, shooter.

[PRR]

From fast to slow: TTL, DTL, RTL, and bare T and R, all work the *same*.

Find the plan for TTL guts. The input is diodes for easy gating, though implemented as an inverted "transistor" and resistor. The next stage IS your bare-naked inverting transistor. When you want high speed feeding many-many outputs (and stray capacitance), TTL adds that ugly 3rd stage for active pull up and pull down. If you only need one input you do not need the input diode logic. If you only need to pull on an LED, only "faster than a finger", you don't need the 3rd stage.

[nandrewjackson]

I breadboarded the NPNs out with A and notA.


I had to raise the A base resistor to 200 K because notA was staying on. Another sneaky leaky voltage in there.


It worked good with LEDs and H11F1s seperatly, but with them in parallel, the LEDs wouldn't fully light up. Although the diagram for H11F1 shows an internal LED, it must have different characteristics than a 3mm bulb LED.

[nandrewjackson]

Forgot to mention, I used a 9v battery and simple resistor voltage divider to get 5V.