4066 switcher design issues

[Athin]

big edit:
real sorry for the long and messed up post which was here before, but you know how it is - when you ponder something for a longer time
you start to talk detail (ie complete gibberish) but to you it makes sense...
1) what is the circuit:
 the cross-coupled nor gates form a RS flip-flop (not an oscillator)
 the OR gates provide separation (so the signal doesn't go into another flipflop's set)
 it's a sort of radio-button thing - if we press S2 I want to get a high on the 2nd flipflop and low on the other two, then we press S1 and ff1 is high - others are low etc.)
 then the signal goes to one of the two dip-switches (channel A and channel B) where we select which of the 4 effects will be on channel A and which on channel B, the third flipflop is a "clean" chanel (direct link from guitar in to the "power amp simp")
2) what went wrong during the 1rst build:
   I think that the circuit was a real bastard to build and that I just wired something wrong. It wasn't stable and made an alien noise.
3) my questions:
a) the amp sim will be a part of the SAGT2 so where exactly should I connect PAD12 IN THE SANSAMP. I was thinking of injecting the signal after pin 7 of IC2 (IC2b 2b exact) before the 10k resistor (on the tonepad.com SA schematic). I don't think a buffor would be needed.
If you don't get me then I'll put it in a different way - at which point in tha SansAmp can we say that the preampish(distorion) section ends and the PA sim section starts?
b) can the OR gates be subbed with diodes? Diodes will generate spikes when they get turned off, but will this cause the circuit to malfunction?
c) would it work OK if I powered the digital chips with a +-9V supply - then the signal would be in the middle of the power supp. so no need to bias the 4066 switcher right?
d) if I used the same PS for this switcher and the ROG preamps and the SAGT2 would I need to seperate the switcher from the rest via a 2 22u cap and a 100R resistor bridge (caps to ground and resistor between them)?
e) Is the 2u2 C1 cap doing something eles apart from setting the 1rst chanel on? something like delaying the 1rst flipflop's reaction time? If so - how else to set the chan. on startup? (or would it be easyer to set it manually after startup by pressing one switch?)
f) pads 4,5,8,9 aren't grounded because the input of rog preamps and the SAGT2 are. Because of this - should I also throw away R2? I mean there will be always some switch open thus making a connection to ground.
g) any suggestions? obvious errors - I don't think I want to make another throw-away board

This is the schematic which I'm thinking of building now. (copy paste into new window)
http://kosi.website.pl/3chan_III.gif
The old one was way complex, and I got it wrong. But here it is for reference.
http://kosi.website.pl/3chan.gif
And the noise it made from time to time (dunno why - maybe because of real bad layout)
http://kosi.website.pl/3channoise2.mp3

[Athin]

yeath I know the post is on the long side, anyone?

[Athin]

how about I power the 40xx's with +-7V. Would thjis give the same result as biasing to 4.5V and blocking with caps (as on geofex)?

[gez]

I was thinking of injecting the signal after pin 7 of IC2 (IC2b 2b exact) before the 10k resistor . Is this right?

On my dinky little screen I can't find this bit of the schematic, roughly where is it?  What signal are you talking about, the effect?

I don't have the time at the moment to check through everything with data sheets so I'm probably not much help, but it sounds as though something somewhere is oscillating. It's obviously not what you wanted, but it sounds cool!  

Edit:  OK, bear with me as I feel like crap this morning.  If I've read correctly you're feeding the circuit from an oscillator and that's what we're hearing?  I don't understand what your problem is (too much info in your post).  What doesn't work? (please keep it simple!)  

[Athin]

thanx for your reply gez. hold on till I put up the schematic/
pS - anyone know a good free website wher I could put up pics and stuff and just give link to those (not like 4 ex. tripod)
again real sorry for the messed up post :oops: hope this time it's ok (I mean clear, not short )

[David]

thanx for your reply gez. hold on till I put up the schematic/
pS - anyone know a good free website wher I could put up pics and stuff and just give link to those (not like 4 ex. tripod)
again real sorry for the messed up post :oops: hope this time it's ok (I mean clear, not short )

Athin:

Photobucket has been used for displaying pictures with great success.  Just do a search and you should quickly find directions on how to use it.

[gez]

c) would it work OK if I powered the digital chips with a +-9V supply - then the signal would be in the middle of the power supp. so no need to bias the 4066 switcher right?

I can't see the schematic any more so there's not much I can reply to (that aside, Andrew or RG are the ones to ask about switching).

Not too sure what you mean by the above (though you're post is a lot easier to understand now, thanks for making it clearer!).  It's best to bias the inputs and outputs of each switch at half the supply.  If you're saying that you can bias a switch's input directly from the output of an op-amp/whatever, then you might introduce clicks unless both ends of the switch are at exactly the same DC level.

I have a Penfold schematic where switches are DC connected and a little cap is used on the output to 'absorb' any current flow that results when the switch opens.  I tried it but it didn't work.  

[Athin]

schematic is uploaded. (http://kosi.website.pl/3chan_III.gif)
gez:
I intend to use a 9-0-9 power supply (ie Vdd would be at +9v and Vss would be at -9v, ground at 0V) by using my own dual 7660 convereter (it works - checked). So the inputs and outputs would be exactly in the middle of the power supply.

[gez]

Not too sure if I'll be able to answer all (any of!) your questions, but I should have some time tomorrow morning so I'll take a look at the schematic then.  

[R.G.]

1) what is the circuit:
the cross-coupled nor gates form a RS flip-flop (not an oscillator)
the OR gates provide separation (so the signal doesn't go into another flipflop's set)
it's a sort of radio-button thing - if we press S2 I want to get a high on the 2nd flipflop and low on the other two, then we press S1 and ff1 is high - others are low etc.)
then the signal goes to one of the two dip-switches (channel A and channel B) where we select which of the 4 effects will be on channel A and which on channel B, the third flipflop is a "clean" chanel (direct link from guitar in to the "power amp simp")

See the circuit with the 74C393 octal D latch at GEO for a single IC implementation of a 2 to 8 button radio button.

2) what went wrong during the 1rst build:
I think that the circuit was a real bastard to build and that I just wired something wrong. It wasn't stable and made an alien noise.

Unless you are quite careful in building any-to-any switchers -which I think your schemo can do - you can get direct feedback paths through the switches even if the wiring is correct. That is one possible "alien sound" scenario. The other is feedback through high impedance power and ground lines or other poor power wiring that can let signal currents cause feedback voltages elsewhere in the circuit.

b) can the OR gates be subbed with diodes? Diodes will generate spikes when they get turned off, but will this cause the circuit to malfunction?

Yes. however, you'd still need flipflops. I personally would go for real, no-fooling flipflops like CD4013 or 74C74 which have set and reset inputs or for the '393 setup I mentioned. Simpler, better specs, less chasing ghosts around the circuit board.

c) would it work OK if I powered the digital chips with a +-9V supply - then the signal would be in the middle of the power supp. so no need to bias the 4066 switcher right?

Yes, that would be OK, but +/- 5 or +/- 7 would be a better choice because 18Vdc is the absolute max for 4000 series CMOS.

d) if I used the same PS for this switcher and the ROG preamps and the SAGT2 would I need to seperate the switcher from the rest via a 2 22u cap and a 100R resistor bridge (caps to ground and resistor between them)?

You will certainly need to separate the grounds somehow, or the ground-current thing I mentioned earler will cause switching pops.

e) Is the 2u2 C1 cap doing something eles apart from setting the 1rst chanel on? something like delaying the 1rst flipflop's reaction time? If so - how else to set the chan. on startup? (or would it be easyer to set it manually after startup by pressing one switch?)

Wow - that's a clumsy way to do the pushbuttons. What I think happens is that at power up, the 2u2 cap pulls up the inputs it's connected to and then slowly relaxes back to ground by the 1M resistor. This has the effect of pulling all the inputs high on that line. Then, when a pushbutton is pressed, it lets the 63uF cap fight the 2u2 cap to pull those inputs up temporarily. It would make more sense, IMHO, to put a 10K resistor between the junction of the 2u2 cap with the 1M resistor and the inputs. That would let the amount of current exchanges be much smaller. It is possible that the designer was using the heavy capacitances to swamp out switch bounce, but there are better ways to do that, too.

f) pads 4,5,8,9 aren't grounded because the input of rog preamps and the SAGT2 are. Because of this - should I also throw away R2? I mean there will be always some switch open thus making a connection to ground.

Good design practice says to never rely on what you *think* might be connected to an input or output. Use a high value (1M-10M) resistor to pull the pads to ground.

g) any suggestions? obvious errors - I don't think I want to make another throw-away board  

Mother Nature is teaching you something fundamental. Never, ever make a PCB for a circuit you have not successfully breadboarded.

[Athin]

thanx RG, wouldn't have thought 'bout the 74C393
but B4 I let you go a few questions about your posts

"You will certainly need to separate the grounds somehow, or the ground-current thing I mentioned earlier will cause switching pops."

yes I've been thinking of that... I suppose the most straightfroward approach would be to use two PS's - how do the pro's do it?

"Wow - that's a clumsy way to do the pushbuttons. What I think happens is that at power up, the 2u2 cap pulls up the inputs it's connected to and then slowly relaxes back to ground by the 1M resistor. This has the effect of pulling all the inputs high on that line. Then, when a pushbutton is pressed, it lets the 63uF cap fight the 2u2 cap to pull those inputs up temporarily. It would make more sense, IMHO, to put a 10K resistor between the junction of the 2u2 cap with the 1M resistor and the inputs. That would let the amount of current exchanges be much smaller. It is possible that the designer was using the heavy capacitances to swamp out switch bounce, but there are better ways to do that, too."

ummm.... I'm the designer :oops:
and there is no such a thing as a 63u cap, I wanted a 22u @ 63V there, but the 10K res. idea is ok... this is also why I had to press the S1 switch longer than the other switches.

Switch bounce isn't an issue - the nor gate ff does the debouncing for itself (when you think of it - it doesn't matter if you "set" or "reset" the ff once or 10 times)

maybe using some sort of buffer would be better (it'd set the 1rst switch high but would keep the capacitance out of the circuit)

"Good design practice says to never rely on what you *think* might be connected to an input or output."

Yes, but i *know* what will be connected - those effects will be there permanent, it's not a separate switching system but a part of a bigger effect (the ultimate preamp/amp sim you might say.... well... for me and for now at least)

"Mother Nature is teaching you something fundamental. Never, ever make a PCB for a circuit you have not successfully breadboarded."

Mother Nature, isn't always right - I did a lot of circuits like that... and they worked.. some required a few more/less components, but they worked. Maybe this time I'll listen.

Oh, and what about a)? anyone (not just you RG)

[Athin]

I haven't thought of the 74C393 'cause you can't get those (not from anywhere I know of) in Poland. I know this is going the easy way, but would a 74HCT573 or 74HCT373 work too?

And what's this thing with transparent/non transparent?

edit : and one more thing - shouldn't the logic IC 1Meg pulldown resistors go to Vss (ie -9V) and not to ground? With a dual PS logic high and low would be +9 and -9, so 0V would be right in between which = real bad stuff going on, right?

[Peter Snowberg]

I'm sure the 393 is a typo as the 393 is a dual 4 bit counter. 373 is the one you want.

There is also the 74HC273 which is like a 373, but it has a clear input too.

[R.G.]

I haven't thought of the 74C393 'cause you can't get those (not from anywhere I know of) in Poland. I know this is going the easy way, but would a 74HCT573 or 74HCT373 work too?

Sorry - my few remaining brain cells misfired. The '373 and '573 are the right chip numbers. The '393 is a counter.

Yes, you can make the HCT parts work, but they are limited to 5Vdc on the power supply, not the 18V of 74C parts or CD4000b parts.
If you use them, you'll have to use some kind of power supply to get 5V for the chips and then repower the outputs up to the +/- power supplies you're using on the switches.

And what's this thing with transparent/non transparent?

In a transparent latch, any change in the data when the clock is high is transmitted directly to the Q outputs. When the clock drops, whatever the logic state was at the setup time before the clock dropped remains in the latch. The latch lets data "shine through" as long as the clock is high. The 74C373 is odd in that the setup time is *negative* so that data that was on the D input a few nanoseconds after the clock drops remains in the latch. That lets you run the clock from the "or" of all the data inputs.

yes I've been thinking of that... I suppose the most straightfroward approach would be to use two PS's - how do the pro's do it?

That's one way. Another is to decouple the power to logic and analog sections through two entirely separate current carrying paths with enough fast capacitance to ground on each one to shunt any power supply noise to ground. And the grounds don't necessarily have to be separate, but the conductors that the currents flow in have to be.

Imagine that copper wire is not 0.0014 ohm per inch, but 100 ohms per inch. You have two sections, a noisy logic one with current spikes that have to go down that ground wire, and a high gain analog one with signal riding on it. If both sets of currents share the same wire, then the current spikes on the ground wire move the analog ground around, and the analog stuff thinks that's signal.

If instead, you had the logic grounded through one wire and the analog through a separate wire, the two of which only connect at the power supply, then the logic ground can bounce around all it wants and the analog ground never sees it. This is a two-way version of "star grounding".

Think about where the *currents* are traveling, and imagine that that path is a resistor. Then you'll see the ground feedback paths.

ummm.... I'm the designer

Sorry - no insult intended. It just struck me that way. I apologize for my big mouth.

Switch bounce isn't an issue - the nor gate ff does the debouncing for itself (when you think of it - it doesn't matter if you "set" or "reset" the ff once or 10 times)

You would think so, wouldn't you? All I can say is, I've done this before and Mother Nature occasionally surprises me. I always worry about switch bounce. You may be luckier.

Yes, but i *know* what will be connected - those effects will be there permanent, it's not a separate switching system but a part of a bigger effect (the ultimate preamp/amp sim you might say.... well... for me and for now at least)

That's fine. I will occasionally do things like that. But I have wound up debugging things when I reused the innards for something else and had forgotten the short cuts I took. You're probably OK. I'm just conservative (about electronics anyway).

Mother Nature, isn't always right - I did a lot of circuits like that... and they worked.. some required a few more/less components, but they worked. Maybe this time I'll listen.

Mother Nature *is* always right. We sometimes hear her incorrectly. I do anyway. You probably have better ears than I do.

Oh, and what about a)?

It'll have to be someone else - I don't know nothing about the tonepad stuff.

[puretube]

...if you think mother nature isn`t always right,
Murphy will teach you a new trick...

[Athin]

Sorry - no insult intended. It just struck me that way. I apologize for my big mouth.

no insult taken. Actually it's good to have a big mouth when what you say is valuable

Mother Nature *is* always right. We sometimes hear her incorrectly. I do anyway. You probably have better ears than I do.

Oh come on! Little ol me better ears than you?! Stubborn, yes, ignorant, yes, but not better ears

.if you think mother nature isn`t always right,
Murphy will teach you a new trick...

can't teach an old dog new tricks  :wink: [but then again - I'm not that old....]

[David]

can't teach an old dog new tricks  :wink: [but then again - I'm not that old....]

Ummm...  I guess I am.  Bow-wow!

[donald stringer]

Yes, you can make the HCT parts work, but they are limited to 5Vdc on the power supply, not the 18V of 74C parts or CD4000b parts.
If you use them, you'll have to use some kind of power supply to get 5V for the chips and then repower the outputs up to the +/- power supplies you're using on the switches.

So if I have this straight I can supply the hct type chips with 5 volts and run the output through a 4049 to bring them up to the proper voltage in order to turn the 4066 control pins on or off. All I want to do is use your 1-8 selector schem. to choose between one of four possible circuits wired into a 4066. And power up leds for indicators. I dont care if I have to use another 4049 to do that as long as it functions well.

[Athin]

well... >>I<< think it'd work - just as long as the voltage between Vss and Vdd won't exceed 10V, because then even a high signal from the 5V supplied latch would be still in the low range (Vss to 1/2*Vdd). Maybe add a 2 resistor biasing network [this is my idea so it probably won't work!]

[R.G.]

You can use discrete transistors or MOSFETs or darlington arrays at the output for voltage level translation.