Logic Outputs Problem in 7493 IC

[_q12x_]

Hello from Romania.
I have all the counting ccts, for every number. This is counting to 16 and its the last one this 7493 can count.
I encounter a new and very weird problem. All the [Q] (QA,QB,QC,QD) outputs of the 7493 are not so good at outputting. At least to my components I tested so far.

I simulated more possibilities and all failed. I thought maybe the simulator is not that good. But when I tried the BJT in reality, it still didnt work. I really want to believe it will work with opamps in reality, but if it will not work, it means the simulator is right.
I didnt try with mosfets in the simulator yet, they were just prepared there. I didnt test them in cct.
THe problem is the (-) that 'comes out' after these buffering devices. When they are on (-) on their output.

I actually solved the problem using a tri-state buffer. They are the only component that are solving this issue.
Ive also get 10 pcs from aliespress. I start to value them now, when I actually need them. But they will arrive after 2 or so months from now.
- I want to find a solution (IF POSSIBLE) without them, or at least until they arrive.

This cct I already make and it is working fine (in the simulator) is to output a decimal number for all the 16 binary counts.
Pretty much when is 0100 it will lit the 4th LED. When is 1000 the 8th LED and so on, from 1 to 16 (including 0).
You can see in the screenshot already.
I work on this cct and numerous problems for .... about 10 or more days. Very hard to do it alone !!! But I did it. so far.

I even have an easier output method than this very complicated cct I already built.
- Simply to display in binary. Like these 4 LogicProbe from Proteus simulator that I am always take for granted.
So we can do these first !

To make a simple binary counter [LED display], I will still need some sort of buffer, like the ones I specified already.
I wait for your response. This is a very complicated cct and a big problem for me.
Thank you and hope I will get some interesting answers! (mister @marconi , my Logical friend?haha)
And also Happy New Year 2023 to everyone that are reading my posts !

 

[marconi]

I will answer soon but not today.

 

[_q12x_]

Hello,
Very interesting observation that you made with the current through the 7493 !!!!
"The Q outputs of the 74LS93 can only source -800 uA (Ioh) whereas they can sink 16mA (Iol) See page six at this link."
I couldnt see it myself, without your help, not in a million years. Yah, very good catch there. Respect for that !
- - -
I've --simulated-- YOUR (private) suggestion with the PNPs. And I made it work !
Of course from the 100th time, but it did.
I had to include those NOT gates to normalize the output from the PNP's. Also I had to tweak some resistors values as well, to make it work at full capacity. 10k on the Base of the tr's was too much and the voltage was too small on those bus wires. I made them 3k and I got 5V per PNP output, on each colector.
Sourcing and sinking the current through the transistors... Very interesting tweak !
Now Im very curious how it will actually work in reality. The simulator is giving (sometimes) false positives. I'll update you soon.
Thanks so far !

f

 

[marconi]

Seems we make a good partnership.

 

[_q12x_]

Seems we make a good partnership.

Yes....heh....dont say 'hop' yet.
---in reality--- it didnt work. I told you, I get false positives from the simulator.
Is a bit of explanation of what I did.
I used c556b for the PNPs. I have them brand new from 2000's. Rarely used, once or twice.
For the NOT gate, I used 74LS04

I got 150mA through the entire cct (read from the PSU) because I used the 100R on the gnd. And I change them to 300R and I get 80mA through the entire cct. Without them in the cct, I get 10mA. The 100R were getting very hot as well, while the 300R not so much.
I marked as -Ds1- and -Ds2- the 2 sets of Diodes I used on my breadboard. I didnt build ALL of this, only parts to see if its working.

What all this means?
Ds1 are set to count to a fixed number. Ok? In this case I set it to count to 8 for testing purposes.
Ds2 are set to --display-- number #1. It's their only job. When counter pass through #1 from its counting program(to 8), this part will display #1 is activated. Very simple.
I hope it is very clear now.
The problems:
- #1 LED is not lighting at all ! It should, as in the simulator.
I probe with osc the base of the tr next to #1 and no visible signal.
I probe the 300R, all of them , and no visible signal.
I probe the base of all 4 PNP tr's , and no visible signal.
I only get a ~3Vish(or so) signal on the NOT output to the base of the first PNP. No signal on the input of the NOT gate.
And the probes of the osciloscope I had to inverse them. I connected the (+) probe to the (+)rail of 5V. And the (-) probe on the output and I could get a reading - only this way !

I got a very-very-VERY low reading, almost imperceptible on the #1 diode in Ds2 group. And only when using the 300R !!! When I used 100R, nothing. Very perplexing ! Also with the osc probes inverted (probing with the (-) on the pins there, to actually see a reading !!! ).

Ive also did an experiment. Initially I put a white LED on #1. And I thought, well, maybe is not sensitive enough (and I was very right) and I change it to a red LED as you can see in the picture. I get crazy sometimes and try everything. You know. And from my MANY permutations I did, I reach this stage when I eliminated some 300R and some diodes in Ds2. The effect was interesting to mention to you. The LED was LIT but like 50% lit. Ok? But it was pulsating. It was getting even more lit, to 75% LIT. Ok? And back to 50 and back to 75 and so on. So it is a signal on the line !!!!! Thats my observation.
The VERY small imperceptible signal confirmed this LED LIT exception experimentation.

Thats it.

 

[marconi]

I will take a closer look over the next few days.

 

[marconi]

Are the LEDs in an array with their cathodes all connected together like this:

Or are the LEDs all separate and you have connected their cathodes together and then to 0V?

 

[marconi]

For circuit around #1 transistor the diode connected to the left of R199 is incorrect. Have you done the same for the transistor circuit below numbered #2?

 

[marconi]

How are the 16 LEDs to be lit? One LED of 16 or n of 16 where n is the count state?

eg for count state n=3 where count starts at state 0 and goes up state 15:

off off off on off off..off

or

on on on on off off off….off

 

[_q12x_]

Here is the full cct. Zoom in it to see the details. Its a big file.

or here where you can really see it in original resolution https://www.deviantart.com/q12a/art/CountingOutputLEDs-v1A-944060484
also you can download the image and inspect it in your favorite image viewer in your windows.
Hmmm.... I am using some custom LEDs in my drawing and you cant see the direction they are pointing. They are towards GND. All of them.

Hold on I actually show their direction ...I guess was not that clear to others...it was very clear to me. You see that green bar next to the led? Like a diode that has a line designating the (-) side. I really thought is intuitive enough. Hmmm. is not looking like a conventional LED, that is very true....

 

[_q12x_]

How are the 16 LEDs to be lit? One LED of 16 or n of 16 where n is the count state?

Ive also shown this part as well. In #1 post I made here in this thread. Look with attention on the picture there and you will see a single green light glow. Ive also explain it in words what it does. You should re-read my #1 post.
I will prepare a video and showing that is indeed working but only here in simulator. Not in reality. Hold on.

 

[_q12x_]

 

[marconi]

Are the LEDs in an array with their cathodes all connected together like this

View attachment 105082

Or are the LEDs all separate and you have connected their cathodes together and then to 0V?

Answering my own question by studying your bread board in #5, you are using discrete red LEDs, each of which has their cathode connected to 0V and their anode to +5V via a 1kR resistor. You turn them on and off by a transistor in parallel with the LED, a way I find unusual.

Now I have been able to study your DS and # circuitry, you have built your own version of diode transistor logic (DTL)gates and by connecting its inputs appropriately to the four busbars for QA, QB, QC and QD you can use these gates to carry out a 4-term logic operation appropriate to selecting one of the 16 LEDs to illuminate eg D* . C* . B . A to light the fourth LED when the 93 counter is in binary state 0011 or count n = 3; n is in the range 0 to 15.

These logic scheme will light only one LED at a time, 0 to 15 and then repeating.

I reckon that only one of your DTL gates connected to the logic busbars will probably work fine. The problem arises when you have more than one because they interact with each via the logic busbars. This is because they are current driven not purely voltage driven. Let me pause there on my critique of your circuit.

It appears to me that your project aim is to construct combinational logic to convert a 4 bit binary code into a one of 16 code. You could do this using logic ics which you could buy - NANDs being typical because you can make all logic functions using these. You have decided not to. Instead you are enjoying building from scratch your own logic gates using DTL technology.

https://www.niser.ac.in/sps/sites/default/files/basic_page/logic gate circuits.pdf

I suggest you study the section in this tutorial on DTL. And furthermore that you switch the current to an LED rather than divert the current away from the LED. And, that you connect the 16 LEDs anodes to +5V and use a transistor to sink each to 0V.

Aim to make a DTL NAND and a DTL Inverter because with these you can do any logic operation.

PS: (In my projects I use a NAND or an inverter gates to sink an LED to 0V via a 1kR resistor. A logic 1 at their inputs turns on the LED which is an example of positive logic - if something is true then something else is true.)

 

[_q12x_]

Did you watched my movie?

 

[_q12x_]

Instead you are enjoying building from scratch your own logic gates using DTL technology.

Im not enjoying anything. Im only finding the best intuitive ways I can make this work.
My aim is to get 4 wires (the bus wires) and output from them every combination. Up to 16. But is impossible to make it with only 1 one BCD. I tried 3 of them and I failed on cascading them. I can use 1 single BCD and the rest, from 9 to 16 using this DTL way.
Then I thought, for testing purposes I will use everything made of DTL starting from 0. And thats the history I reach this stage.
Hope is clear enough now. Watch my movie and you can skip through it. Im explaining there a lot of details.
Thanks.

 

[marconi]

Yes I have watched your video which was clear. Construct these two circuits. The left one will light the led when all inputs are logic 1 and the right one inverts the input.

First of all do a decoder for 0011. You will need one copy of the NAND and two copies of the NOT.

Do you have 8 n channel power mosfets?

 

[_q12x_]

Do you have 8 n channel power mosfets?

I have.
But the thing is.... I progressed until then.
I measured everything with my osciloscope. I wasnt getting a clear HIGH on the 7493 outputs. In fact, I wasnt getting ANY SIGNAL AT ALL. I was getting a very lousy signal, but VERY close to gnd but I think that was some other type of signal, not the real one, because my osciloscope is catching multiple signals at once. So I decided that 7493 either its outputs are not good - because all this diodes arrangement, compromising somehow the outputs to gnd ...
OR
the signal was so fast that the LED #1 didnt had time to react. Maybe everything is working as it should, but if the signal is fast like 1us, like a spike, that doesnt tell anything to the LED what to do and how much to stay lit. Right?
---
I then, did the following experiment: I put a 7474 DTFF on one of the output of the 7493 and run the entire thing.
Very simple setup:

I was counting to 4 I think. So the Led on the 7474 output, was litting and staying lit until the next count. (UNFORTUNATELY)
like so: 1234 - led ON and staying ON, 1234- led OFF and staying off, and again....
It should have be: 12 -led ON - 3-Led OFF 4 and repeat the same sequence again. But it wasnt.
BUT
This prove a point. That the outputs of the 7493, in the current configuration, with those diodes there, can be expanded to other inputs IC. And als othat the signal is ok and also that is as I suspected, VERY-VERY FAST. Too fast for the LED to know what to do. Thats a problem !!! And I intend to find a solution.
Momentarily I am thinking on a DELAY solution using a 555 per 1one line for the moment. But if it will work, I will have to put 4 of them for each line.
Also VERY important to mention ! If until this experiment right now, I was NOT getting any clear osciloscope readings on any of those 4 lines (when they were all connected to the NOT IC and after that to the LED...) NOW, I get a CLEAR 5V HIGH and 0 LOW. Using the Flip Flop version. But Im starting to see the picture. I hope you will agree with my observations.
Pretty much what Im doing right now is a binary display for ONLY these 4 wires here.
IF you know any way of DELAY-ing the signal, to slow it down to... I dont know, 100ms maybe? enough that the LED will flash ON, that will be great. I am planning to use a 555 for 1 line for now. If this will work, we're in bizniz.
Tell me what you think.

 

[marconi]

If you have 8 n channel power mosfet then but these can be connected common source to the 0V rail and a 1kR resistor inserted between each drain and the 5V rail.

Four of the mosfet have their gates driven by A, B ,C and D outputs of the 7493 and the other four by A*, B*, C* and D* derived from the four inverter gates you currently use. You can dispense with the 4 pnp transistors.

Connect the first 4 mosfet drains to the current logic busbars and the second 4 mosfets drains to a new set of 4 logic busbar. Now you can connect the 4 input dtl NAND with led as required to 4 of these 8 busbar to select the right 4 term logic operand term for the led. The eight busbars provide A, B, C, D, A*, B*, C* and D* the logic signals.

With this scheme you can avoid using dtl NOT gates.

 

[marconi]

I have.
But the thing is.... I progressed until then.
I measured everything with my osciloscope. I wasnt getting a clear HIGH on the 7493 outputs. In fact, I wasnt getting ANY SIGNAL AT ALL. I was getting a very lousy signal, but VERY close to gnd but I think that was some other type of signal, not the real one, because my osciloscope is catching multiple signals at once. So I decided that 7493 either its outputs are not good - because all this diodes arrangement, compromising somehow the outputs to gnd ...
OR
the signal was so fast that the LED #1 didnt had time to react. Maybe everything is working as it should, but if the signal is fast like 1us, like a spike, that doesnt tell anything to the LED what to do and how much to stay lit. Right?
---
I then, did the following experiment: I put a 7474 DTFF on one of the output of the 7493 and run the entire thing.
Very simple setup:
View attachment 105101
I was counting to 4 I think. So the Led on the 7474 output, was litting and staying lit until the next count. (UNFORTUNATELY)
like so: 1234 - led ON and staying ON, 1234- led OFF and staying off, and again....
It should have be: 12 -led ON - 3-Led OFF 4 and repeat the same sequence again. But it wasnt.
BUT
This prove a point. That the outputs of the 7493, in the current configuration, with those diodes there, can be expanded to other inputs IC. And als othat the signal is ok and also that is as I suspected, VERY-VERY FAST. Too fast for the LED to know what to do. Thats a problem !!! And I intend to find a solution.
Momentarily I am thinking on a DELAY solution using a 555 per 1one line for the moment. But if it will work, I will have to put 4 of them for each line.
Also VERY important to mention ! If until this experiment right now, I was NOT getting any clear osciloscope readings on any of those 4 lines (when they were all connected to the NOT IC and after that to the LED...) NOW, I get a CLEAR 5V HIGH and 0 LOW. Using the Flip Flop version. But Im starting to see the picture. I hope you will agree with my observations.
Pretty much what Im doing right now is a binary display for ONLY these 4 wires here.
IF you know any way of DELAY-ing the signal, to slow it down to... I dont know, 100ms maybe? enough that the LED will flash ON, that will be great. I am planning to use a 555 for 1 line for now. If this will work, we're in bizniz.
Tell me what you think.

Qc changes every 4 clock pulses. This is correct. See count sequence table bottom right on page 3 of:

https://www.ti.com/lit/ds/symlink/sn74ls93.pdf?ts=1673101430824&ref_url=https%3A%2F%2Fwww.ti.com%2Fproduct%2FSN74LS93

You have a connection Qc to reset RO1 - why? I thought you wanted to count in binary from 0 to 15 and then back to 0 and repeat?

I think you line of thought about fast pulses is a red herring. If you doubt that the 93 is working correctly then it is simple enough to test it.

 

[_q12x_]

I didnt know this image you show with the 4 clocked output pulses ! Very interesting indeed. Thanks for it. It didnt cross my mind to search for it either! You did good (this time). Haha. I actually save it in my 7493 folder. It is extremely important to actually see the ------- how is working by default. Im learning...

You have a connection Qc to reset RO1 - why? I thought you wanted to count in binary from 0 to 15 and then back to 0 and repeat?

Well, Like I said, and also explained in my movie as well and SHOW it there very clearly.... but I will repeat, no problem, I am using this cct as 2 parts, 1=the logic and 1=the display. So in the logic part I have this counter 0-15. But I will change it, manually, depending what number I will need to count. I used some examples, including the maximum of count to 16. But it can be 1by1 count as well. Depends what I will need when I will need it. Get me?
The display part SHOULD do its job for whatever count # I SET. Simple.

If you doubt that the 93 is working correctly then it is simple enough to test it.

I already did, using that DTFF I mentioned earlier. The 7493 is fine. The pulse from it is not fine for the type of output I want to use.
The very FAST pulse that my golden osciloscope can not read it at all, (only the 7474 DTFF did) is most probably because of the strange connections with the diodes directly to each Q output of the 7493. Thats why the fast signal, like a spike. It would have being nice to be like in your 4 clock pulses picture !!! But it isnt.