Cascading two 74LS90's doesn't work (needs deglitching)

Hi Fred, Thanks for that cct. After doing a lot of reading on this topic last night your cct. makes sense to me. I will try it out later on tonight. BTW what about the idea previously suggested by Arie and cbarn24050 to include a series termination resistor (50 to 100 ohms) on the driving(source) side? Detailed here:

Which option would be preferable (series termination vs. AC trace termination (i.e. your cct.)) and why ? Power consumption via the

BTW, another problem I found with my cct is that of cross talk between the wires in the cable (resulting in unwanted CLK triggering). To reduce that effect I will probably have to replace the round multicore VGA cable with a flat ribbbon cable and ground every second conductor.

regards Peter

That's a good reference...

Series termination is close but not really a match for your situation for a few reasons: 1) your line length is much shorter than the transition time for LS logic and does not act like a true transmission line, and 2) the output stage of the LS is not symmetric, over 100 hundred ohms when driving high and maybe a few hundred milliohms when pulling low.

You don't have a choice because it is not possible for you to series terminate and your logic does not have enough drive to develop valid logic levels across a resistor parallel termination.

Right- crosstalk results from the capacitive coupling between the lines and high impedances at the source and receive ends. You told us you were using bundled VGA cable, and those lines should be miniature 75 ohm coaxial. If you ground the shields then crosstalk is greatly attenuated, and by maintaining a low impedance at the receiver as I suggested, the crosstalk is further attenuated. I went with 82 ohms because it a common value and is less than 5% mismatch reflection for 75 ohm lines, which is more than tolerable. The 82 ohm in series with the drivers does provide a series match on the logic low transition which your destination triggers on but does not help much on the logic high transition. It does help reduce crosstalk though. That ac-termination should be right up between the center conductor and shield of the cable.

Rocky wrote:

I would skip routing the clocks through the selector switch altogether and use a data selector 74LS153 like so, it can handle a dual rotary but only one switch because the A B inputs are common to both halves: View in a fixed-width font such as Courier.

. . . . translation to 2-bit binary . . . . . ROTARY SWITCH . --------- B | A . | | ----|---- . | o-|-> P0 __ | __ . | | P0 | PO . | o-|-> P1 __ | . VCC---|-o---> | ====> P1 | P1 . | o-|-> P2 | __ . 5V | | P2 | P2 . | o-|-> P3 | . | | P3 | P3 . --------- | . . . circuit for LS switch tranlation . . . P0 >--N/C . . . . P1 >---------+------------------+------->A . | | . | | . | | . P2 >------------+--------+------|------->B . | | | | . .-|>|-' | | | . | | | | . P3 >---+ | [470] [470] . | | | | . '-|>|----' | | . ----+------+------ . 2x 1N4148 GND . . . . . . . inputs to dual 4:1 selector . . . . . 75LS153 . ----------- . from >----| A | . translator | | . ckt | | . >----| B | . 4 | | . from >-----/------. | | . LS90 |\\| C0 | . Q outputs | | | . |\\| C1 Y|---> clk out . | | | . |\\| C2 | . | | | . \\| C3 | . | _ | . .--| G | . | ----------- . --- . gnd . . . . . . . . ------ --------- . | | 4 | 74LS153 | . |74LS90|---/----|SELECTOR |---> CLK OUT . | CTR | | | . | | | A B | . ------ --------- . ^ . | . '---------. . ------- ------------- | . |rotary | 3 | switch | 2 | . | sw |-/-| translation |-/-' . | | | | . ------- ------------- . . . .

You could do a 74LS09 quad AND OC, but it's not really all that common: View in a fixed-width font such as Courier.

. . +5V . | . | . | . ROTARY SWITCH 74LS09 | . --------- --------- [1.2k] . | | | __ | | . | o-|-> P0-----------+--------|--| \\ | | . | | [470] | | >--|-----+-->clk . | o-|-> P1-------. | .--|--|__/ | | out . VCC---|-o---> | | | | | | | . | o-|-> P2----. | | | | | | . 5V | | | | | | | __ | | . | o-|-> P3-. | +------------|--| \\ | | . | | | |[470] | | | | >--|-----+ . --------- | | | | |/-|--|__/ | | . | | | | | | | | . | | | | | | | | . | | | | | | __ | | . | +---------------|--| \\ | | . | | | | | | | >--|-----+ . |[470]| | |/-|--|__/ | | . | | | | | | | | . | | | | | | | | . | | | | | | __ | | . +------------------|--| \\ | | . | | | | | | | >--|-----+ . [470]| | | |/-|--|__/ | . | | | | | | | . | | | | | --------- . '--+--+---' | . | | . gnd | . | . 4 | . LS90 OUPUTS>---------/--------------- . . . .

Make that a 74LS38: View in a fixed-width font such as Courier.

. . +5V . | . | . | . ROTARY SWITCH 74LS38 | . --------- --------- [1.2k] . | | | __ | | . | o-|-> P0-----------+--------|--| \\ | | . | | [470] | | o--|-----+-->clk . | o-|-> P1-------. | .--|--|__/ | | out . VCC---|-o---> | | | | | | | . | o-|-> P2----. | | | | | | . 5V | | | | | | | __ | | . | o-|-> P3-. | +------------|--| \\ | | . | | | |[470] | | | | o--|-----+ . --------- | | | | |/-|--|__/ | | . | | | | | | | | . | | | | | | | | . | | | | | | __ | | . | +---------------|--| \\ | | . | | | | | | | o--|-----+ . |[470]| | |/-|--|__/ | | . | | | | | | | | . | | | | | | | | . | | | | | | __ | | . +------------------|--| \\ | | . | | | | | | | o--|-----+ . [470]| | | |/-|--|__/ | . | | | | | | | . | | | | | --------- . '--+--+---' | . | | . gnd | . | . 4 | . LS90 OUPUTS>---------/--------------- . . . .

Fred, I think this latest suggestion of yours is a great one. I'm going to order a few 74LS38's tomorrow. For the record I tried out both the series termination and the AC termination ccts. mentioned above. I had mostly encouraging results. The AC termination (220pF+82R) at the load end of the cable turned out to be the most effective just by itself. When I tried a series termination of 82R at the source end (i.e. the Q outputs of the

I had to increase the series termination resistor from 82R to 180R before it eliminated spurious triggers. Anyway once I placed BOTH series and AC termination components in place, the two 74LS90 counters seemed OK until I introduced a third counter in the cascade (74LS92). I'd rather not go into details as I've decided it is not worth refining the different termination methods given that I'm going to use the 74LS38 now.

Oh, and before all the PIC fans continue telling me how I should have used a PIC to solve this problem, well the CLK selection part of the cct. (that we've been discussing) is actually independent of the CLK generation part (i.e. PIC vs. 74LS90's). Additionally I actually have

Fred, what package do you use to generate the ASCII schematics ? The are certainly very helpful. My schematics are currently drawn up using PSPICE Schematics evaluation version (v9.1). I also have LTSpice installed, but my circuit currently isn't laid out in that editor.

It will be some time on the weekend before I can report back on my success with the 74LS38's as per Fred's cct.

thanks again

regards Peter

On a sunny day (Wed, 30 Apr 2008 06:28:44 -0700 (PDT)) it happened Rocky wrote in :

Now that is vague is it not?

One more reason to use a PIC.

Always state your real requirements.

Divide and conquer. Subsystems that are accurately specified can be designed independently to other subsystems as long as someone with the complete picture (i.e. a full system perspective) is ensuring on paper and in practice that the subsystems integrate properly.

regards Peter

BTW Jan, since you were the first to suggest this solution I extend the credit to you too and then to Fred for elucidating the idea and drawing up the schematic.

regards Peter

On a sunny day (Wed, 30 Apr 2008 07:06:00 -0700 (PDT)) it happened Rocky wrote in :

That works sometimes.

Yes, like NASA mars lander landing gear - engine off microswitches- activating already when gear wen down. BOOM. I have seen so many times several teams on a project, and you really need very good inter-team communications to make the thing work. Just writing down some specs is not sufficient, specs can be interpreted in many ways.

And politics can come into play too, sometimes people in control do not even care if the end result works...

On a sunny day (Wed, 30 Apr 2008 07:10:43 -0700 (PDT)) it happened Rocky wrote in :

Thank you, my admiration goes to the nice 3D PIC board somebody to you presented here.

So it turns out the 74LS38 is indeed a rare beast and might take me a bit longer than I hoped to get a hold of some. So I am after some alternative suggestions for the mean time.

Briefly here is why the 74LS38 is so useful in this application: The four 74LS90 outputs are OR'd together (in a negative logic sense) by the 74LS38, this is achieved because the Open Collector outputs will happily remain connected together irrespective of their logic state. The 1K2 resistor pulls up the output to +5V as necessary. CLK_OUT in the above schematic actually connects to the CLK input of a D-Flip Flop (74HCT74) (not shown). Now what can I substitute for the

Any ideas ?

regards Peter

BTW is it possible to export my PSPice schematics into ASCII art ?

On a sunny day (Thu, 1 May 2008 04:45:16 -0700 (PDT)) it happened Rocky wrote in :

CMOS switches could work just fine, 74HC4016?

The HC4053 / 52/ 51

If you use the 1 of 8 decoder, perhaps you can do your switches to ABC with a diode matrix? Perhaps all you need is a diode matrix? Or does LS TTL not work with 4016 (Rd on is about 150 Ohms IIRC?

PIC

Just make a gif image screenshot and use one of those free sites to publish it.

Digikey has some 1,319 parts of it in stock:

If component count is important for you, you should really use a microcontroller. What do you think about my suggestion at

PS: For larger PIC projects, like with the PIC18F252, I'm using the commercial MCC18 C-compiler, which works very well, because programming a PIC in assembler needs some time.

The fact that you had to increase the series R to 180 leads me to believe your main problem was crosstalk, the 180 and 82 ac-divider is about 0.3 which puts the crosstalk edge at less than 1V at the victim. But you're right, it's not worth the trouble if you can find a low cost solution. Beware that the '38 will invert your clock, so you will want to invert the wire-AND junction, either with a spare gate or a little transistor ckt, that is if phase is important to the application: View in a fixed-width font such as Courier.

. . . . . . . +5V . | . -----+----- . | | . | [1k] . | | . [2.7k] +----->CLK . | | . | 1N914 |/ . from>---+--|>|-+--| . LS38 | |>

. [1k] | . | | . ----------+----+----- . . . .

The '38 is not rare at all, a very common part number...

I live in Sydney, Australia. Digikey has no outlet there, so shipping time will be longer. Local Farnell and RS don't stock the part. So unless I get it locally I'll just order one from say Futurlec and wait

at

The learning curve and setup time/cost was not warranted for this project I'm doing. I agree if I had it all setup and was familiar with a particular PIC then I'd use it in a flash.

regards Peter

Thanks Jan. I'll sleep on that idea (It's nearly 12:30am in Sydney.

Good idea

regards Peter

quite possibly. You wrote before that the VGA cable would have mini

Yeah phase is, but I have a few spare inverters to use. Thanks for the cct anyway. Tell me Fred, which application do you use to convert your schematics to ASCII ?

regards Peter