Noise problems with "Jeopardy" design... HELP!

Hi,

Bit of a problem - I've got to get a "fastest finger first" quiz setup for 8 people finished by thursday evening (quiz night at my kids school)... and my EE background ended 20 years ago...

I've been using the circuit from EDN design Ideas, August 16th 2001 "Circuit improves on first event detection" - the CMOS (4013) version.

Just as I come to build the wretched device, which worked fine on breadboard, I find that when built using multi-core cables that the other button sense lines seem to fire when one button is pressed, resulting in loads of lights coming on even though the inhibit line drops correctly.

Using a 'scope show loads of spikes on the other button lines when another button is pressed which is obviously causing the problem. I should also mention that the multi-core cable also carries 300mA @12V to light some MES bulbs in the button-pusher's box - I suspect that this maybe where the cross-talk is coming from.

The PSU is clean, and each chip is decoupled by a 100n right underneath it. Its built on VeroBoard, very neatly and cleanly. The veroBoard version worked fine before I connected the long wires with the buttons and lights in boxes (to sit in front of each contestant).

Now, I suspect the multi-core cable is capacitively coupling the button or light lines, causing mayhem. WHAT CAN I DO? I need to either improve the noise immunity of the inputs (how?), or would switching to the TTL (74F74) version of the circuit help? Would 74LS74 be any different to using the

74F74 for input rejection? Would stopping using multi-core and using point-to-point 2 core be the answer?

I have so little time (I have a day job) that I really need solid advice here...

My fragile credibility as a father is at stake here ;-) Sigh ;-(

Many thanks,

-- Mike

Reply to
Mike Deblis
Loading thread data ...

8

my

"Circuit

MES

it.

in

Hi Mike,

Since the circuit worked with short wires, maybe the challenge is your long wires in combination with the very high impedance inputs on the CMOS device.

A common design flaw with CMOS devices is leaving a input "open" (floating) with a long wire lead connected to them. They will pick up any stray signal! If that is the case, try a resistor at the 4013 CMOS end of the wire from the input:

1) to ground if the circuit is low and pulled high by the button, or 2) to V+ if the circuit is high and pulled low by the button.

I have found that for 5 volt supply, a few hundred ohms per input does the trick. For 9 to 12 volts, maybe a few hundred more.

Hope that helps,

Tim

Reply to
Tim Dicus

Mike, Is your cable composed of twisted pairs? Or just multiple wires? If it just multiple wires, then you can think of each of the wires as an antenna waiting to pick up any noise around them. Tim's idea of terminating the wires will help a lot to keep the inputs to a known state. If you are going to be using these buttons in a studio or auditorium, you might want to invest in using wire with twisted pairs so that you can at least ground the other wire in each pair going to your switch boxes. This will cut down noise.

I once had a PFH (Project From Hell) that had a problem with wiring. It was a control system for a large auditorium. I was the third and fifth engineer assigned to the project (it kept being moved around as no one really wanted to do it!) When I finally tried to implement it, I found that the wiring that had been used was all single conductor, and had about 200' runs of it all around a VERY EM noisy auditorium. Due to a design constraint, the signalling was down with DTMF (Don't Ask!) but there was so much noise on the cable in one of the four rooms, you could not get reliable signalling, even filtering the heck out of the signal! If it had been twisted pair, it would have been no problem.

Charlie Edmondson Engineering Unique Solutions to Unusual Problems

Reply to
Charles Edmondson

Hi - Thanks for the replies...

The CMOS inputs (S input of a 4013) are all tied to ground with 10K at the chip - they are pulled high by pressing a button.

I agree that having the power to the indicator lights in the same bundle is a BAD PLAN as there is a 300mA pulse going up to the light in the selected box, and a corresponding 300mA whack down the ground wire in the same bundle - this means that all the boxes in the chain will see the pulse.

I suspect that I'll either have to star-wire or use a seperate bundle for the light power, and try to keep the two cables apart (else I'll just have the same problem).

It may be that daisy-chaining the boxes, whilst neat, is not such a good idea...

Thanks

Mike

Reply to
Mike Deblis

A 'star' configuration would probably be better.

Leon

--
Leon Heller, G1HSM
Email: aqzf13@dsl.pipex.com
 Click to see the full signature
Reply to
Leon Heller

Hi Mike,

I would recommend changing those pull-down resistors to 2K, maybe even to

1K, especially if the wires are long, unshielded, and unbalanced. You may need to play with the resistance value a bit.

I will guess that switch bounce is not causing the trouble, or you would have had that problem with short wires also. You are using the same switches, correct?

Tim

Reply to
Tim Dicus

I read in sci.electronics.design that Mike Deblis wrote (in ) about 'Noise problems with "Jeopardy" design... HELP!', on Tue, 20 Jan 2004:

Very likely.

Put capacitors from input to ground; experiment to find the minimum value that works. If this causes driving problems, maybe you need to put resistors in series before the capacitors.

--
Regards, John Woodgate, OOO - Own Opinions Only. http://www.jmwa.demon.co.uk 
Interested in professional sound reinforcement and distribution? Then go to 
 Click to see the full signature
Reply to
John Woodgate

Mike, maybe it's time to dig into the capacitor junkbox and capacitively slug down the inputs to the chips. Another solution may be to condition the risetimes from your hard-contact switch closures by inserting a series resistance driving a shunt capacitor at each switch.

Nearly 40 years ago I learned the importance of not mixing hard contact closures with fast logic the hard way, just as you evidently are now doing. Since you're an EE, you should be able to appreciate the fact that an unconditioned hard contact closure results in nanosecond risetimes that freely couple inside your circuitry, particularly when cables are involved.

This leaves you with two obvious solutions. Slow down the extreme risetime created by the hard contact closure using a simple RC filter very close to the switch, or de-sensitize the chip to the reception of nanosecond impulses by slugging down the circuit at the chip end. Both methods work, and sometimes both are needed! :-)

A more elegant solution is to employ only SPDT switches that direct set or reset a flop that provides a clean transition. Still, this is what you do in design 2, but a few resistors and capacitors should be sufficient to kludge design 1 into proper operation.

Enjoy. Isn't doing stuff with your kids fun! :-)

Harry C.

Reply to
Harry Conover

You could use lower pullups/pulldowns to lower crosstalk. Maybe add an RC filter to the IC input. And separate the signal ground and supply from the power ground. You can then properly decouple the ICs, and the ground bounces will be in the other ground line and at most cause a LED flicker.

The good thing about the circuit is that the system is idle until the first button is pressed. The others should be blocked at that instant. A slicht RC delay will not make a difference here.

Thomas

A separate bundle may not be needed. Also take a look at cat5 data cable; it has very little crosstalk between pairs.

I think you have a bus that is idle until action, and at action everything should block and only prevent false triggers.

Maybe even reversing the switches (normally closed) would work... much lower impedance when it is not pressed yet...

Thomas

Reply to
Zak

I don't know what that circuit looks like. Here's another, by Daniel Simon in this group, for 8 inputs.

formatting link

--
-Reply in group, but if emailing add 2 more zeros-
-and remove the obvious-
Reply to
Tom Del Rosso

It's already Wednesday so this may be too late, but here's a simple design using cheap, low-power SCRs that I found in my files, originally I believe from a Dave Thomas:

========================= Here's a simple, cheap circuit that will let you conduct your very own quiz shows. It has a lamp and a button for each player. When a button is pressed, it lights that player's lamp and locks out the other button until the circuit is reset. SW3

+6V -o_|_o----+----------+----------->>--------+----------+----->>

| | | | LAMP1 | | LAMP2 | | SW1 | SW2 | +--A> |o | |== >----------------+---+--->>

SW1,SW2 normally open momentary pushbuttons SW3 normally closed momentary pushbutton LAMP1, LAMP2 6V incandescent lamps R1, R6 470 ohm R2,R3,R4,R5 1 K SCR1, SCR2 Small SCRs, not power type CR1, CR2 1N914 diodes

  • connection ^ cathode of a diode

--A>

Reply to
Terry Pinnell

Your problem comes from the fact that the multicore cable in addition to providing interconnections for all the signals is also a series of capacitors coupling each wire to its adjacent wires. This is responsible for the spikes you are seeing- press a single player switch, and a voltage is transiently coupled into some other FF SET input causing a false latch. This is why the EDN design idea requires separate cables from each input to the corresponding player switch- a star configuration. The basic circuit can be re-arranged to support a daisy-chain with multi-core cable that allows for cross-coupling between wires. The player switch inputs are routed to the FF D-inputs, the D-inputs are OR'ed to trigger a one-shot, FFOS, and at the end of the one-shot timeout, set for 100us, the FF CK inputs are strobed to capture the valid input. The one-shot is then latched RST so that it will no longer respond to additional player switch presses until the circuit is RST by the momentary. Also, the circuit will not be glitched by your large

300mA lamp current going to the player lamps. The resulting circuit consists mostly of what you already have in place with a single FF added.

Please view in a fixed-width font such as Courier.

+---------|>|---------+ '1' | | | +--- | '0' | | | | | | o| | | +---S--+ | |-RST PSW0 -+ | | | | o| +---o/o---cable-----+---+------D FF0 Q | | | | | | 4013 | | | | | \ | | | | | === / +--CK /Q | | | 10n 1k | | | | | | | / | +---R--+ | | | | | | | | | | +---+ | +-------+---------+---------+ | | | | | | | | --- | | | / | | | O | | 10K | | | | | / | | | O | | \ | | | | | | | | | O | | | | | | | | --- | | | | | | | +---------|>|---------+----------+ | | | | | | | | | | '0' | | | | |\ | | | | | | | | \ | | +---S--+ | | | | PSWN _| \ | | | | | | | +---o/o---cable-----+---+------D FFN Q | | | | | | | | 4013 | | | | | | \ | | | | | | | === / +--CK /Q | | | | 10n 1k | | | | | | | | / | +---R--+ | | | | | | | | | | | | +---+ | +-------+ | | | | | | | | --- | '0' | | | | | | | | | +---S--+ | | | | 47k | | | | | --+ +--| +-/\/\----Q FFOS D-'1' | | | | | | | | 4013 | | | | | | | | 470p | | | | | +--+ | +--||-+---/Q CK-+---+ | | | | | | | | | | | +-----|-----+ +---R--+ | \ | | | | | === / | +--cable--< VDD | | 100k | 470p 10k | | +----/\/\-----+ | / | | | | | | | | +---+ | | D1 | | | | +-----|>|-----+ --- | | | | | | '0' | | | | | | | +---S--+ | | | | | | | +--Q FFMR D-'1 | | | 4013 | | | | | | | /Q CK---+ | | | | | | | +---R--+ | | | | | | | +---------------------+ | | +-------------------+
Reply to
Fred Bloggs

...snip

Many thanks - I've decided that the EPN circuit is far too unreliable to be used, and decided to go with yours...

I was looking for a reference to the use of the 4013 as a monostable - couldn't find one on the web and coundn't make it trigger - is the circuit as posted correct?

I know I should experiment, but no time now (I've got only a couple of hours of time left to getr this working...)

Thanks again

Mike

Reply to
Mike Deblis

This type of one-shot is well known. Double check that S='0' and D='1'. The timing waveforms look like this- Vn node is essentially same as R input- your circuit uses the /Q output:

Please view in a fixed-width font such as Courier.

'0' | +---S--+ R | | +----/\/\-------Q FFOS D-'1' | | 4013 | +---- | Vc | | | Vn +------||----+--/Q CK-+---+-----< | | + - | | | | | | OUT -Vdd

Q output charges C thru R, so C charges from -Vdd to ~0.5*Vdd

at which point RST starts to activate to clear FF back to 0.

Vdd-(-Vdd) Timing is then RxCx Ln (--------- )= RxCxLn(4)=1.4 x R x C Vdd-0.5*Vdd

The waveforms look like so:

+---------------------------------- | CLK | | -------+

+----------+ | | Q | | | | --------+ +-------------------

--------+ +-------------------- | | | | /Q | | +----------+

3Vdd | \ ---- | \ 2 | \ | \ | \ | \ | \ Vdd | \ --- | \ 2 / \ Vn / \ / \ 0V--------+ / --------- | / | | | / | | | / | | / | | / | | / | -Vdd|/ | ||
Reply to
Fred Bloggs

Should read Q=0 /Q=1 and Vn=0, the FF will not trigger unless S=R='0' and D=1, so double check this before you test- it is a very reliable circuit-especially for the very short timeout shown.

Reply to
Fred Bloggs

Also, if you are testing with a square wave generator- use 50x the timeout or frequency of less than 200Hz for trigger.

Reply to
Fred Bloggs

[... snip ...]

Fred,

Thanks for your help - The evening, last Thursday, was a great success - the boxes worked fine 95% of the time (one or two mis-fires! I'll sort those later), but the whole system was very impressive. I added a small piezo sounder and a 555 timer to give a different beep for each team for about 0.5 secs. The boxes, about 25cm x 10cm x 10cm from MDF with inset acrylic fronts and the buttons on top looked the part, and were much appreciated.

We raised about USD 750 for the school in Sri Lanka, so a good result all round...

PS. fred, could you drop me a line at mdeblis at hotmail dotty com as I would like to ask you one question off line if possible...

Mike

Reply to
Mike Deblis

for 8

and my

...

...

Fred,

I built this and it worked fine... except that if there is a switch bounce as the monostable times out, i.e. a zero on the line that triggered to monostable , no light comes on (the "0" is latched through) and what's more, all buttons are now locked out. This happened a couple of times during the evening we were using the boxes.

What's required is for the monostable to reset if no capture line is active when it times out, i.e. to allow a subsquent "bounce" to re-fire the mono and latch the button.

I've tried to find a simple solution for this, with not much luck.

Ideas welcome!

Thanks again,

Mike

Reply to
Mike Deblis

Mike Deblis wrote:

I find it difficult to believe this is happening- but a simple and expedient fix is to condition the lockout on a D input being '1' like so: Please view in a fixed-width font such as Courier.

CHANGE THIS:

----|>|-------+ from D | inputs | - | ---|>|-------+ | | | to CK's | | /|\ '0' | | | | | +---S--+ | | 47k | | | --+ +--| +-/\/\----Q FFOS D-'1' | | | | | | 4013 | | | | | | 470p | | | +--+ | +--||-+---/Q CK-+---+ | | | | | | | +-----|-----+ +---R--+ | \ | | | === / | | 100k | 470p 10k | +----/\/\-----+ | / | | | | | | +---+ | D1 | | | +-----|>|-----+ --- | | | | '0' | | | | | +---S--+ | | | | | +--Q FFMR D-'1 | | 4013 | | | | | /Q CK---+ | | | | | +---R--+ | | | | | +-------------< reset sw | | +-------------------+

TO THIS:

----|>|-------+ from D | inputs | - | ---|>|-------+ | | | to CK's | | /|\ '0' | | | | | +---S--+ | | 47k | | | --+ +--| +-/\/\----Q FFOS D-'1' | | | | | | 4013 | | | | | | 470p | | | +--+ | +--||-+---/Q CK-+---+ | | | | | | | +-----|-----+ +---R--+ \ | | | | / | | | 100k | 1k | | +----/\/\-----+ / | | | | | | | | | | D1 | | | | +-----|>|-----+ --- | | | | | | '0' | | | | | | | +---S--+ | | | | | | | +--Q FFMR D------------+ | | 4013 | | | | | /Q CK---+ | | | | | +---R--+ | | | | | +-------------< reset sw | | +-------------------+

Reply to
Fred Bloggs

Try this. Not built or tested.

o V+ o V+ | | .-. .-. | | 1k | | 1k | | | | '-' '-' | | master | | reset | Pos 1 | _/ | .-----------. | .-o/ o-| | | | | S2 | GND ---+1 8+----V+ | | | ___ | | | | +|_R_|----+-----+2 7+--------------------o GND | +-----| | 555 | _/ | | | o--------+3 6+--------o-o/ o-----o | --- | | | | S1 | | C--- | V+ ---+4 5+----+ .|. | | | | | | | | | | | GND | | | --- | |100k | | - '-----------' --- '-' | | ^ LED | | | | | |---+ | | | GND | | .-. | | | | | | | | | | '-' | | | | | V+ | | | | | ' To pos 2...N To pos 2...N created by Andy´s ASCII-Circuit v1.24.140803 Beta

formatting link

Reply to
GPG

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.