EEVblog 1611 - Top 5 Jellybean 7400 & 4000 Logic

[EEVblog]

The TOP 5 Jellybean 7440 and 4000 series logic chips.



Schmitt Trigger video:

00:00 - Top 5 TTL and CMOS Jellybean chips
00:37 - Different 7400 series family types
01:52 - What makes a Jellybean component
3:27 - CD4051  / CD4052 / CD4053 Analog Multiplexer Switch
17:42 - 74HC595 8 Bit Latched Shift Register
20:58 - Used in the MXO4 oscilloscope
24:09 - Driving LED's
26:12 - 74HC74 Dual D Flip Flop
27:00 - Used in the Rigol HDO1000 Oscilloscope
28:14 - Uses for a D Flip Flop
28:53 - Special Mention 74HC374/ 74HC574
30:35 - 74HC14 Hex Schmitt Trigger Inverter
35:18 - Last chip, what about NAND NOR or XOR?
36:06 - 74HC245

The jellybean Components Playlist: https://www.youtube.com/playlist?list=PLvOlSehNtuHv268f0mW5m1t_hq_RVGRSA

[Kleinstein]

The 74HC595 shift register is common, but it still has point to whatch out for. When using more of them in series there is very little setup time for the serial inputs of the later ones. The output changes to the new signal right with the clock. This can be an issue with longer traces and not so ideal signal integrety (especially the clock).
The 4000 series has a slightly improved version (4094 that is also available as 74HC4094). The main functions are the same, but there is an additional output for the next register in the line that is hold up half a clock cycle longer. The clear pins is skiped for this. The 3 line interface is kind of a simple SPI interface.

A very nice use for the 74HC14 is to build a simple oscillator with just a capacitor to ground and a resistor in feedback.

When using a 3.3 V signal source (e.g. many modern µCs) and the logic chip has to be 5 V , there are the 74HCT.... variants with a lower threshold. Originally they are made to interface old style 74LS and similar, but the level is also a perfect match for 3.3 V CMOS.
With only a 3.3 V signal the normal 74HC with 5 V supply is really on the edge and not reliable, though it can work most of the time.

[Nominal Animal]

Us microcontroller/Arduino/Single-board-computer hobbyist types would have benefited from adding a sentence about dual supply level translating transceivers at the end (but I know the video was already long), for example noting that the 74LVC8T245 variant performs the same way as a bus transceiver, but with separate supplies for each side, so that one can do voltage level conversion at the same time.

(The NXP 74LVC8T245PW,118 variant Mouser has 100k+ of in stock for pretty cheap, 0.93€ in singles, 8€ for ten, can do 1.2V, 1.5V, 1.8V, 2.5V, 3.3V, or 5V on either side in any combination, and is thus extremely useful with 3.3V and 5V microcontrollers and peripherals, and SBCs with 1.2V or 1.8V logic levels.)

The 74LVC1T45 is the equivalent for a single data line, but in SOT23-6 (3mm×3mm×1.5mm, 0.95mm pitch) is small enough but not too small for a hobbyist to solder dead-bug style in line with the wires and cover with heatshrink, too.
 _ _ _ _ _

I see so many posts and reports of hobbyists like me burning their microcontrollers or SBCs or at least a few I/O pins on them, by not thinking about this stuff at all.  I wish someone with a wide enough viewership would point this out, because the solutions are so easy and easily available.  Starting from the current-limiting resistor (which relies on the ESD or other diodes to voltage rails), to transistor open collector outputs with pull-up resistors, to level translating transceivers, to digital isolators.

The trap for young players is the bidirectional (auto-direction sensing) transceivers, which tend to have a low-level glitch when changing direction or a difference in the low-level voltage depending on the direction, causing issues with devices with particularly low low-level threshold (sensing that glitch as a short high state).

For example, compare TI TXS0101 (bidir), TXB0101 (bidir), and TXU0101 (unidirectional) data input low voltage ranges.  TXU0101 has Schmitt trigger inputs with wide hysteresis, so basically never glitches; this is why I prefer the fixed-direction TXU0n0m models (also cheap and high bandwidth).  TXB0101 has V_IL = 0 .. 0.35×Vcc which is quite normal, but TXS0101 V_IL = 0 .. 0.15V !  In particular, chaining two TXS0101 is often problematic, because its output low can be up to 0.4V.

Observing these (using e.g. a simple Arduino sketch which toggles a pair of pins, with one of the two always an input (without pullup or pulldown enabled, Z), and the other High or Low –– cycling between ZL, ZH, ZL, LZ, HZ, LZ –– , passing through a bidirectional translator, both pins connected to a two-channel oscilloscope) would show the issues very nicely.  Unidirectional with Schmitt triggers is usually a superior solution, unless the direction does actually vary.

The best example would be a standard USB-to-serial (TTL) converter cable/dongle, but with an extra TI TXU0204-Q1 or TI ISO6742 at the end, plus two 0.1µF supply bypass capacitors, one for each supply side; and maybe current-limiting resistors on the TX and RTS outputs.  It would expose VCC, GND, RX, TX, RTS, and CTS on female pins or pin hooks.  RTS and CTS are obviously optional, and not often used.  VCC would go to whatever logic level the device uses for these signals, and GND to device ground.  TXU0204-Q1 has Schmitt triggers and supports any VCC between 1.1V and 5.5V; ISO6742 is an isolator that supports VCC = 1.71V–1.89V or 2.25V–5.5V (i.e., 1.71V–5.5V except 1.90V–2.24V).
The TXU0204-Q1 costs less than 1€ in singles at Mouser, and ISO6742 less than 2€, so we're talking about minimal extra cost.  But how many devices would have been saved from excess voltages and currents, if such a cable were used?  Why don't SparkFun, Adafruit, et al. offer those?  Us sausage-fingered hobbyists would greatly benefit from these!

Or is it that having parts of ones own circuits at different logic level voltages is just too complex to think about?

It would be useless for me to make such a video.  Even my posts here and e.g. at PJRC/Teensy forums don't get traction.  I myself have learned this the hard magic-smoke-escaping way; I just wish others could learn about it earlier and easier.

[julian1]

Really enjoyed this video. Good part choices, rooted in practical building-block components.
A video covering all the different series would be great. 

[ballsystemlord]

A video covering all the different series would be great.

I second the motion.

[ledtester]

A very nice use for the 74HC14 is to build a simple oscillator with just a capacitor to ground and a resistor in feedback.

I like using a NAND gate with Schmitt-trigger inputs - then you can easily gate the oscillator signal with another logic signal or tie both inputs together to get just an inverter.

[madires]

Other classic CMOS logic ICs are 4066 (analog switch) and 4093 (NAND, Schmitt-trigger).

[ballsystemlord]

At about 3min in, Dave pulls up the first datasheet. It's for a CD405xB. It says, " Maximum input current of 1uA at 18 V over full package temperature range, 100 nA at 18 V and 25*C."

Now how can it be 1 uA over the full temp range and then 100nA at 25*C? 1uA is much bigger than 100nA.

Thanks!

[SeanB]

1uA is an absolute maximum, the 100nA is a typical value, which most parts will have. The larger number includes the operation at 125C, where the die will be the most leaky.

[Kleinstein]

The leakage current is usually very temperature dependent, about doubling every 10 C. So leakage at 125°C can be 1000 times higher than at 25°C.

With the cheap logic series chips the specs are rather loose as the they are not tested to a strict level. Even some of the typical values are rather conservative (e.g. to leave room for porcess changes).
The actual leakage is often even lower. There is a chance to find cd405x series chips with actual leakage in the 10s of pA range (at room temperature and moderate supply).

[Sensorcat]

The 74xx4053 can do even more: If some glue logic is needed, it can often replace the 74xx00 NAND and 74xx02 NOR. And since design with transmission gates is less common, it is also less boring!

In fact, the transmission gate is more versatile than any other gate. In principle, any gate logic can be made from 74xx4053s only. While a pure NAND or NOR design requires anything other than NAND or NOR to be made from more than one gate (except an inverter), a single transmission gate can be wired for 6 different functions, see table below. This means a lower number of gates is required for a given task. Usually, this should also result in lesser PCB space and cost.

It is important, however, to pay attention to the effects of chaining the gates: If a signal has to pass through several stages, the propagation delay increases or it may not work at all. Intermediate inverters, whether made from transmission or other gates, will fix this.

Y0, Y1 = MPX inputs
S = select input
Z = MPX output

A, B, C = inputs of logic function

Y0      Y1      S       Z = logic function
------------------------------------------
0       1       A       buffer A
1       0       A       not A
B       1       A       A or B
0       B       A       A and B

B       0       A       ~A and B
1       B       A       ~A or B

Bonus, the general 3-input function:

B       C       A       (A and C) or (~A and B)

[PathLoss]

1. List of 7400-series integrated circuits

https://en.wikipedia.org/wiki/List_of_7400-series_integrated_circuits

2. List of 4000-series integrated circuits

https://en.wikipedia.org/wiki/List_of_4000-series_integrated_circuits

Have fun...   

[cosmicray]

I see so many posts and reports of hobbyists like me burning their microcontrollers or SBCs or at least a few I/O pins on them, by not thinking about this stuff at all.  I wish someone with a wide enough viewership would point this out, because the solutions are so easy and easily available.  Starting from the current-limiting resistor (which relies on the ESD or other diodes to voltage rails), to transistor open collector outputs with pull-up resistors, to level translating transceivers, to digital isolators.

74ls07, buffer, open-collector, very useful for translating voltages.