With just one tiny little addition that could do so much more... One XOR gate with a couple of random taps taken from the outputs and feeding into the data input would create the classic LED waterfall with a continuous stream of random data.
The is actually no issue with the clocks being connected together, this is pretty much a requirement for must synchronous systems to operate. From the Texas Instruments data sheet for the 74HC595 - "† This setup time allows the storage register to receive stable data from the shift register. The clocks can be tied together, in which case the shift register is one clock pulse ahead of the storage register."
+Julian Ilett It's not immediately obvious but if you think about it the flip-flops in the shift register are connected in exactly the same way and work without problem and of course also give the required one clock delay. This is in fact how most synchronous digital circuits are arranged; a topic for a video perhaps?
Thanks for this Julian - I too was inspired by Big Clives sign (although I'm not sure I'd be bothered with quite the number of LEDs he soldered into that sign!). I've some 74HC595s sat in my component box for a couple of months. I must find a suitable project to utilize one or two. Thanks for the explanation - I'm sure I'll been viewing your video again once they come out of the box.
You can´t imagine how much i appreciate this one! Some 30 years ago i played around with 4000er series shift registers, but i wasn´t able do set up what you just done...so shift-regs had beed like some sort of evil electronic things to me. I used binary counters and decoders instead. Now i go for these 595s, may with the arduino, 4017 or just like you the 555...) Lesson learned, thanx Julian!
What would happen if you connect the output of the last register to the input of the first one? Would it be like a loop in wich you can program pulses?
Cool! I made one out of SPST(single pull, single throw) 5 volt relays and led's for relay-diode logic many years ago. I was expecting a sequence generator with the circuit I designed but then it turned out to work as a shift register... that was a moment! I have been into this stuff more ever since. Thank you for this because you show how easy it is to make this exactly as you have or otherwise.
Race condition can be essentially discounted (from a visual point of view) in this setup, because your clock rate is fast compared to the data rate. If the race condition occurs on the very occasional data transition it will be corrected on the very next clock pulse when the data input will have reverted to steady state (at which point it matters not if "new" or "old" bit gets clocked in to latch as both are the same value).. And you have very long runs of steady state input when you use a switch as input. However if you were to drive the input of the second SR from the latched output of the first, and so on, you may see a different result? Or reduce the clock rate and you may see an occasional latch error.
Thank you for your statement 'bit like a scope'. I wonder if anyone else ' latched' on to that idea. Mmm! Another five rows of LEDs and how many chips? I'm nearly 80 years old and can't get my brain around that. Over to you. Hehe
Didn't understand half of that (my lack of knowledge, not your explanation) but I still enjoy watching. I do like the macro filming, even if you only achieve it with a magnifying glass :-)
would it be possible instead of a 555 timer but use an audio source to create a spike that could be used to clock the circuit with audio and be set up as a small sound operated light display of some kind.
the shift register contains a current limiting resistor at each output. many 74h series do. that is the reason the amperage isnt ruining leds. this is misleading.
Some of these 8 segment bar graphs would work well with this circuit. www.amazon.co.uk/Pins-Segment-Display-Digital-10x20mm/dp/B00H8QG6AW/ref=sr_1_3?ie=UTF8&qid=1476639538&sr=8-3&keywords=8+segment+LED+bargraph+display
You "could" actually use a 595 for multiple inputs. Add a push button to every output of the 595 and have them all connect to 1 input on an arduino, send a bit to every output quickly, one at a time and check the input to see if a button has been pressed. Wherever the bit was, that's where the button was pressed. Not efficient but an interesting experiment.
I've done something like this using a 595 as an address register and 8 other similar registers as display drivers (ucn5841 I think it was, open collector shift registers designed for driving coils instead driving common anode LED displays) using 4 io lines and a serial input I made an 8 digit led display that could display numbers, beep and display 21 other characters
9:09 I don't understand the part with the 7-segment led display. There are only 2 of the shift-register chips on the board, hence 16 bit can be stored, but the board has 4 digit display, with 7 leds each (if you don't use the dp). So it needs to shift 28 bits around, with only 16 bit registers. How does this work?
Whether clocking both SH_CP and ST_CP simultaneously causes a race condition or not, the datasheet shows that the clock pulses are 180° out of phase. Therefore if you connect the SH_CP directly to the 555 timer and ST_CP to the 555 through an inverter, any problems with instability within the 74HC595 are resolved. This also matches the clocking pattern in the datasheet.
Julian ilett have you ever played with pld's ? I've recently been given an old printer and I found a... atmel ATF16V8B pld I've downloaded the datasheet as I didn't have a clue what it was. and as you seem to like playing with logic!!! cool video; -)
Nice. It's given me an idea to make what I can only describe as 'organ pipes' or 'bubble tubes' of LEDs to be triggered by a MIDI xylophone I've been meaning to finish. Would look great having 8 of these 'tubes' driven by your penny organ.
Instead of two shift registers use a 7447 to light up 16-- 7 segment displays by using a 74154 to pull each 7 segment low or maybe its' high? and then use a 74193 as a clock and a 2114 memory chip to store the number for each of the segments. This may work better for a 7448 I cannot remember But I know this works as the 74154 is running so fast you cannot see the change I am thinking a 555 timer and slow that clock down and you can see each display light up at a time. Interested? Let me know I can draw it out. Or I used to could I am 61 now and a tenth grader dropout. So
I just read the '595's datasheet yesterday and then this popped up in my YT recommended list. Spooky. Anyway, according to the datasheet: "If both clocks are connected together, the shift register always is one clock pulse ahead of the storage register." So it doesn't sound like there's a race condition but working as expected. Cheers.
4 times 10 segment LED bargraphs would look good here if you crave neatness, but you would need to put them on the breadboard below that one and then use 40 link wires (ouch!) or, since you are going to need 40 you might as well go with the resistors anyway. Of course this only really makes sense if you already have them and are that OCD. He he he!
Great stuff. This afternoon I got three tpic6c595 shift registers. They have a different pinout from regular 595 chips but they can sync quite a lot more current.
You should have fun with a digispark. I actually managed to create software for the digispark that hacks any current windows system and the anti malware program cant do anything about it. It abuses the usb keyboard functionality to inject download scripts into various programs. Its basically a usb stick arduino, giving you great new possibilities.
Very interesting, I too watched the Big Clive video and was quite curious to try them. Well I paused this video to go to Ebay to order and some are on their way form China now; just have to give them a go, As always interesting and has to be one of the neatest breadboard layouts I've seen,
Two minutes in you say you are worried that it is wrong to connect the shift clock and the register clock together. At 4:01 into the video you show the data sheet, and in section 1 it states that "If both clocks are connected together, the shift register will always be one clock pulse ahead of the storage register". So it is OK to connect the clocks together.
Just wanted to share, there is a young man Kevin Darrah, that has a channel and has really good info on using these chips, and he built a massive 8x8x8 rgb led cube with them and an arduino.
One thing that would be fun to try is to tap off one of the shift outputs and XOR it with another shift output and take that XOR output and feed it back into the input (or'd with the switch input). That should create a pseudo random number generator.
Unlikely to get metastability issues, because the edge triggered registers are presumably very similar in timing and they work for the chained shift register. Just the extra clock until the shiftregister is transfered in the output register is the issue.
You could skew the clock pulses using a resistor with CMOS. You could play with jumpers and resistors to get to a point where you do see a very obvious race condition. Use a resistor to slightly delay the shift register clock pulse perhaps? The resistor plus gate capacitance will introduce a delay according to the r/c time constant. The "right" values could make it interestingly unpredictable!
+Nick Hill or go the opposite way, when you insert a capacitor in series with that data out clock pulse it will be delayed in order to avoid that race condition....;)
Or you could use a simple npn transistor with a base resistor to limit the current and a collector resistor to make a negative amplifier used as a not gate dlb.sa.edu.au/rehsmoodle/file.php/282/kpsec.freeuk.com/trinvert.gif
+TubiCal A capacitor would AC couple the clock when in series. This would create a charge pump which would likely change the time at which the signal crosses the detection threshold of the input. This arrangement could also potentially damage the input if Vcc were higher since a charge pump can double the voltage. A resistor on the latch bus if selected correctly should on it's own guarantee the usual objective of no race condition. I was thinking to put a resistor on the shift clock bus to guarantee a race condition for fun and effects :-)
I had a set of these arrive this week, and I wired up a circuit to test them, and so far as I can tell, either the chips I received have a 100% failure rate, or I'm doing something wrong, and neither makes sense. Source input is 3.3v, clock input on both RCLK and SRCLK is 10hz, OE and SRCLR verified tied to VCC. Pushbutton on SER similar to what Julian is using here. I've used a meter to check every input, both on the breadboard and directly at the chip pin, and I've checked and rechecked the values, but no matter what, it behaves as if the SER value is tied to ground. (Sometimes the outputs are all high when I plug it into the circuit, then they cycle low immediately and stay that way no matter what I do with SER.) Sanity check dictates that there's no way the company sent me 250 faulty shift registers, but I've verified every pin both with the datasheet(which, by the way, specifies that the "race condition" isn't an issue, tying SRCLK and RCLK together simply puts the output latch 1 clock cycle behind the internal register) and verified with a probe that it's getting the right value. The only way I've managed to get them to work as advertised is by placing the chip on top of the breadboard and then holding it in with slight pressure. This doesn't always work, and even if it did would be inadequate for obvious reasons, but checking the signals directly at the chip pins proves that it can't be a breadboard problem, because the signals *do* appear on the indicated pins. I'm out of ideas, and I'm torn between dumping the entire shipment in the trash or giving the seller a 1-star review, both of which don't make sense because as I said already, while 1 or 2 faulty chips is unlikely but possible, 250 of them borders on impossible.
Started testing every single shift register in the set individually, and I discovered what the problem was once I found a few that worked: some of the registers the vendor sold me actually are faulty. (between 30 and 50% so far.) They probably didn't think the buyer would actually test them before putting them to use.
Interestingly, the datasheet for the NXP version of the 74HC595 does discussion l specify the behaviour if SHCP and STCP are connected - it says the output is delayed by one clock cycle exactly.
Hi Julian, I've been following your RU-vid videos for some time now and found them extremely interesting, Thank You. I've come across this video and I've tried to duplicate the setup but trying to trace the circuit from the image on screen it's been a little difficult, Is it possible to email me a circuit diagram and component list so I can build this project. Thanks.
выглядит отлично! но по видео не удается разобрать все подключения. есть возможность выложить схему? видимо, какая-то ошибка у меня в разводке 555. у кого получилось, можете показать схему? и номинал резистора между 13 и 14 ножками. looks great! but the video fails to parse all the connections. Is it possible to lay out a diagram? Apparently, some kind of error in my wiring 555. who did it, can you show the circuit? and the value of the resistor between 13 and 14 legs.
"Julian's Logic" Something about that title made me laugh :) I have recently been playing around with LM3914 LED bargraph drivers but my chinese source of 500 LEDs didnt ship my order :(
+Luke Den Hartog Oh well. But that could be the responsibility of the post of the other country too. There was a guest on EEVblog not so long ago who was talking about these issues. Post is a mess.
yeah was wondering the same thing so I found this: www.ebay.co.uk/itm/2860-tie-point-solderless-breadboard-Including-jumper-wire-E1U4-G8K4-L5E3/322930676246?epid=1582356529&hash=item4b302b1216:g:ydkAAOSwxixaK4nZ Is this what you are looking for?
+Lan Party Hosting Einstein says no. Long wires won't have much effect. You could throw in a few inverters to add some gate propogation delay, but that's a bit pointless since a single inverter solves the problem anyway.
+Lan Party Hosting A typical coax cable gives you ~5ns/m. And if you just use a wire of a few meters length you may get the additional feature of a cell phone rf power sensor or something.
Sir, you could just use one NPN transistor and 2 resistors (NOT gate) to inverse the clock amplitude for the Storage CLK (pin 12) - maybe to teach young people good habits, and not to be so careless?
