Initializing a Flip-Flop on Power-up

[Dan Charette]

Hi All!

I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a
CD4093 Schmitt trigger NAND configured as a debouncer for my mute
switch input. The output of the 4093 is fed into a CD4013 fo provide
toggling cabability. I've got S and R tied directly to ground. But,
I'd like to explore the possiblity of providing a predictable power-up
scenario. Currently, the circuit upon power up will default to one
state or the other. What are some methods that you fellas have used
to achieve a known state consistantly with a flip-flop like this when
the power supply is from a single source. First gut instinct that I'm
thinking is to make some sorta RC timing circuit on the power supply
pin so that I can provide the set or reset pins with something known
before the power supply pin comes up. Any thoughts? Also, I don't
want to add too much additional circuitry... if I can solve this
purely with passive components or a few transistors, that'd be great.

Thanks!

 

[Jim Thompson]

Hi All!

I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a
CD4093 Schmitt trigger NAND configured as a debouncer for my mute
switch input. The output of the 4093 is fed into a CD4013 fo provide
toggling cabability. I've got S and R tied directly to ground. But,
I'd like to explore the possiblity of providing a predictable power-up
scenario. Currently, the circuit upon power up will default to one
state or the other. What are some methods that you fellas have used
to achieve a known state consistantly with a flip-flop like this when
the power supply is from a single source. First gut instinct that I'm
thinking is to make some sorta RC timing circuit on the power supply
pin so that I can provide the set or reset pins with something known
before the power supply pin comes up. Any thoughts? Also, I don't
want to add too much additional circuitry... if I can solve this
purely with passive components or a few transistors, that'd be great.

Thanks!

You are on the right track. Surf on POR or "power on reset".

...Jim Thompson

 

[Frank Bemelman]

Hi All!

I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a
CD4093 Schmitt trigger NAND configured as a debouncer for my mute
switch input. The output of the 4093 is fed into a CD4013 fo provide
toggling cabability. I've got S and R tied directly to ground. But,
I'd like to explore the possiblity of providing a predictable power-up
scenario. Currently, the circuit upon power up will default to one
state or the other. What are some methods that you fellas have used
to achieve a known state consistantly with a flip-flop like this when
the power supply is from a single source. First gut instinct that I'm
thinking is to make some sorta RC timing circuit on the power supply
pin so that I can provide the set or reset pins with something known
before the power supply pin comes up. Any thoughts? Also, I don't
want to add too much additional circuitry... if I can solve this
purely with passive components or a few transistors, that'd be great.

Disconnect R or S from ground, depending on what you want, reset or
set. Connect a 10uF cap between +V and this input. Put a 47K or 100K
resistor between input and ground.

 

[Spehro Pefhany]

Hi All!

I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a
CD4093 Schmitt trigger NAND configured as a debouncer for my mute
switch input. The output of the 4093 is fed into a CD4013 fo provide
toggling cabability. I've got S and R tied directly to ground. But,
I'd like to explore the possiblity of providing a predictable power-up
scenario. Currently, the circuit upon power up will default to one
state or the other. What are some methods that you fellas have used
to achieve a known state consistantly with a flip-flop like this when
the power supply is from a single source. First gut instinct that I'm
thinking is to make some sorta RC timing circuit on the power supply
pin so that I can provide the set or reset pins with something known
before the power supply pin comes up. Any thoughts? Also, I don't
want to add too much additional circuitry... if I can solve this
purely with passive components or a few transistors, that'd be great.

Thanks!

Use something like this to generate a reset pulse:
http://ww1.microchip.com/downloads/en/devicedoc/11187f.pdf

Only one part required. BTW, don't let the "microcontroller" in the
description throw you, they are suitable for POR of any 3-5V CMOS
logic circuit.

While you can use an RC (typically about 0.1s time constant with a
diode to help out the protection diode to Vdd, a series resistor to
limit the current in case of a Vdd-to-ground short, and perhaps a
drain resistor from Vdd to Vss to make sure that Vdd falls reasonably
quickly), it's far from foolproof, more complex and not much cheaper.
Another low-cost solution is to use a zener, a BJT and a few
resistors, with or without the RC, but that consumes more power, and
is even more complex.

 

[Joerg]

Hello Spehro,

While you can use an RC (typically about 0.1s time constant with a
diode to help out the protection diode to Vdd, a series resistor to
limit the current in case of a Vdd-to-ground short, and perhaps a
drain resistor from Vdd to Vss to make sure that Vdd falls reasonably
quickly), it's far from foolproof, more complex and not much cheaper.
Another low-cost solution is to use a zener, a BJT and a few
resistors, with or without the RC, but that consumes more power, and
is even more complex.

My circuit du jour is a BSS123. Drain tugs on input that needs reset
low. Gate has cap to VCC and resistor to GND. Plus discharge path if
some BOR is required as well. If the 4-5c are too much I'll sometimes
use a BJT. There must be a discharge path from VCC to GND if power is
turned off. If there isn't that needs some extra parts.

Regards, Joerg

 

[Richard Henry]

You are on the right track. Surf on POR or "power on reset".

Here's a good start:

http://www.maxim-ic.com/products/supervisors/reset_ics/

 

[Tam/WB2TT]

Disconnect R or S from ground, depending on what you want, reset or
set. Connect a 10uF cap between +V and this input. Put a 47K or 100K
resistor between input and ground.

That will work, but I unless your power supply is very slow to ramp up, a
0.1 uF will work. If that is too many parts, just connect a small capacitor,
22 - 100 PF, from the Q or Q^ output to ground on the side of the FF that
you want to come up in the 0 state. Note that both of these schemes may be
unsuitable for production,and may not work during a brown out; for that you
would need a power reset chip with a defined threshhold.

Tam

 

[Rich Grise]

Hello Spehro,

My circuit du jour is a BSS123. Drain tugs on input that needs reset
low. Gate has cap to VCC and resistor to GND. Plus discharge path if
some BOR is required as well. If the 4-5c are too much I'll sometimes
use a BJT. There must be a discharge path from VCC to GND if power is
turned off. If there isn't that needs some extra parts.

I'm throwing my hat in with Frank on this one. This is way overkill
for what the OP wants.

I wouldn't use 10 uF, however; 1 uF is way more than enough - heck,
what's the time constant of 1 nF and 100K?

And an "R" or an "S" input doesn't need a lot of noise immunity -
once it's reset, it's reset! ;-)

Cheers!
Rich

 

[Joerg]

Hello Rich,

And an "R" or an "S" input doesn't need a lot of noise immunity -
once it's reset, it's reset! ;-)

Ahem, if it's not a Schmitt input then rather bizarre things can happen
when applying the ramp of a slowly charging cap. Stuff like both Q and
/Q being high at the same time and occasionally that is followed by
something that can't take that and croaks. Like an H-bridge.

Regards, Joerg

 

[Frank Bemelman]

Hello Rich,


Ahem, if it's not a Schmitt input then rather bizarre things can happen
when applying the ramp of a slowly charging cap. Stuff like both Q and
/Q being high at the same time and occasionally that is followed by
something that can't take that and croaks. Like an H-bridge.

Yeah, and the H-bridge was in charge of keeping the sattelite in orbit.

 

[Fred Bloggs]

Hi All!

I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a
CD4093 Schmitt trigger NAND configured as a debouncer for my mute
switch input. The output of the 4093 is fed into a CD4013 fo provide
toggling cabability. I've got S and R tied directly to ground. But,
I'd like to explore the possiblity of providing a predictable power-up
scenario. Currently, the circuit upon power up will default to one
state or the other. What are some methods that you fellas have used
to achieve a known state consistantly with a flip-flop like this when
the power supply is from a single source. First gut instinct that I'm
thinking is to make some sorta RC timing circuit on the power supply
pin so that I can provide the set or reset pins with something known
before the power supply pin comes up. Any thoughts? Also, I don't
want to add too much additional circuitry... if I can solve this
purely with passive components or a few transistors, that'd be great.

Thanks!

For an audio amplifier you can justify using a pot and spare 4093
adjusted to change state at Vcc/2; the 4093 holds the FF RST at low
voltage and pulls 4013 R low for Vin>Vcc/2. This assumes Vcc/2>=2V or
maybe 3V. FF S input is tied to GND.

 

[Frank Bemelman]

Hello Spehro,


My circuit du jour is a BSS123. Drain tugs on input that needs reset
low. Gate has cap to VCC and resistor to GND. Plus discharge path if
some BOR is required as well. If the 4-5c are too much I'll sometimes
use a BJT. There must be a discharge path from VCC to GND if power is
turned off. If there isn't that needs some extra parts.

If there is no discharge path of importance, the 4013 stays in position
for weeks. If there is a load on the power supply, the cap gets discharged
through the internal protection diode.

Spending an extra 4-5 cents doesn't sound like you, what have you been
drinking this afternoon?

 

[Joerg]

Hello Frank,

If there is no discharge path of importance, the 4013 stays in position
for weeks. If there is a load on the power supply, the cap gets discharged
through the internal protection diode.

Yes, with CMOS this cane become a concern. I have had designs that could
have properly idled for hours on a bunch of bypass caps.

Spending an extra 4-5 cents doesn't sound like you, what have you been
drinking this afternoon?

Oh, just coffee. But not the Douwe Egberts brand :-(

It's a toss-up between cost and battery life. The BJT needs some base
current. Only for a second or two but it can matter. Also, you need one
additional resistor to avoid frying the base on power-up and this adds
almost a cent back into it. So, if you can haggle the FET down to under
3c it'll break even. A lot of times I use the same FET for boost
conversion and stuff like that and when there are no BJTs anywhere else
it can be cheaper to use a FET for POR even if it's over 3c, by lower
inventory and rigging costs.

So far the discrete solution was always more economical than one of the
ritzy POR chips. Until one comes out that does it for around 5c.

Regards, Joerg

 

[Jim Thompson]

On Fri, 17 Feb 2006 21:26:39 GMT, Joerg

[snip]

Oh, just coffee. But not the Douwe Egberts brand :-(

[snip]

Douwe Egbert makes coffee too? I used to smoke their Cavendish until
the management decreed no smoking in the new house ;-)

Personal opinion: Anyone who relies on an edge detector for POR is a
bit of a fool. In recent months it's been an almost weekly event to
patch some chip that won't POR properly.

I prefer a level detect with hysteresis AND delay... easy to do if the
chip design includes my bandgap reference ;-)

...Jim Thompson

 

[Rich Grise]

Yeah, and the H-bridge was in charge of keeping the sattelite in orbit.

Did any of you guys even read the original question?
-----<quote>-----
From: Dan Charette <[email protected]>
....
I've got a small toggle circuit that I'm using to drive a shutdown
circuit on an audio power amplifier. The circuit is made from a CD4093
Schmitt trigger NAND configured as a debouncer for my mute switch input.
The output of the 4093 is fed into a CD4013 fo provide toggling
cabability. I've got S and R tied directly to ground. But, I'd like to
explore the possiblity of providing a predictable power-up scenario.
Currently, the circuit upon power up will default to one state or the
other. What are some methods that you fellas have used to achieve a
known state consistantly with a flip-flop like this when the power
supply is from a single source. First gut instinct that I'm thinking is
to make some sorta RC timing circuit on the power supply pin so that I
can provide the set or reset pins with something known before the power
supply pin comes up. Any thoughts? Also, I don't want to add too much
additional circuitry... if I can solve this purely with passive
components or a few transistors, that'd be great.
....
----</quote>-----

i.e., a small cap from Vdd to R, and a 47K-100K from there to Vss.

If you want to be persnickety about it, then Dan could see if there's a
spare gate in his 4093

Or, better yet, put the R/C on one of the 4093 inputs - I'd have to see
the existing circuit to see if that's doable.

Thanks!
Rich

 

[Frank Bemelman]

Did any of you guys even read the original question?

No, what was it?

 

[Joerg]

Hello Jim,

Douwe Egbert makes coffee too? I used to smoke their Cavendish until
the management decreed no smoking in the new house ;-)

Ah, a pipe smoker. Did that in my 20's for a couple years, on and off.
Usually Skandinavik Mixture or Irish tobaccoes. That, and a nice glass
of single malt.

IIRC coffee is Douwe Egbert's mainstream product. But Frank would know
better, it's been two decades since I lived in the Netherlands. And I
always liked the other brand better, Kanis&Gunninck or something like that.

Personal opinion: Anyone who relies on an edge detector for POR is a
bit of a fool. In recent months it's been an almost weekly event to
patch some chip that won't POR properly.

Agree. It's pathetic what young engineers consider a POR these days.
It's actually sad that they have to make a distiction between POR and
BOR. Every reset should be good enough on brown-outs. What's really sad
is that even our (rather expensive) Motorola cordless phone appears to
perform dirty resets on occasion. Got enough honey-do projects so I am
not going to fix their design.

When VCC is low or the circuits are picky I use a BJT because it makes
for a better threshold. Or a TLV431 but that's expensive. If there is a
regulator I take it off before that and after, then issue the reset
before it is too close to dropout.

I prefer a level detect with hysteresis AND delay... easy to do if the
chip design includes my bandgap reference ;-)

Us circuit designers have to pay extra for that bandgap. But a
hysteresis is a good thing though I have to confess that I sometimes
forego it, but only if there is enough margin between threshold and the
circuit actually conking out. Plus enough decoupling capacitance to tide
it over the reaction time.

Regards, Joerg

 

[Jim Thompson]

Hello Jim,


Ah, a pipe smoker. Did that in my 20's for a couple years, on and off.
Usually Skandinavik Mixture or Irish tobaccoes. That, and a nice glass
of single malt.

I smoked primarily Dutch Cavendish from age 18 to age 54.

IIRC coffee is Douwe Egbert's mainstream product. But Frank would know
better, it's been two decades since I lived in the Netherlands. And I
always liked the other brand better, Kanis&Gunninck or something like that.



Agree. It's pathetic what young engineers consider a POR these days.
It's actually sad that they have to make a distiction between POR and
BOR. Every reset should be good enough on brown-outs. What's really sad
is that even our (rather expensive) Motorola cordless phone appears to
perform dirty resets on occasion. Got enough honey-do projects so I am
not going to fix their design.

When VCC is low or the circuits are picky I use a BJT because it makes
for a better threshold. Or a TLV431 but that's expensive. If there is a
regulator I take it off before that and after, then issue the reset
before it is too close to dropout.



Us circuit designers have to pay extra for that bandgap. But a
hysteresis is a good thing though I have to confess that I sometimes
forego it, but only if there is enough margin between threshold and the
circuit actually conking out. Plus enough decoupling capacitance to tide
it over the reaction time.

Regards, Joerg

I live the fun life... if I need a bandgap I just roll one... designed
one last week that consumes only 5uA and comes fully alive at 1.5V
input. Includes a POR.

...Jim Thompson

 

[Spehro Pefhany]

On Fri, 17 Feb 2006 21:26:39 GMT, Joerg

[snip]

Oh, just coffee. But not the Douwe Egberts brand :-(

[snip]

Douwe Egbert makes coffee too? I used to smoke their Cavendish until
the management decreed no smoking in the new house ;-)

Personal opinion: Anyone who relies on an edge detector for POR is a
bit of a fool. In recent months it's been an almost weekly event to
patch some chip that won't POR properly.

I bet. This kind of thing can cause annoyance and much more serious
things in the field.

I prefer a level detect with hysteresis AND delay... easy to do if the
chip design includes my bandgap reference ;-)

...Jim Thompson

How does your bandgap compare with published single/double bandgap
references? Curvature correction PTAT term added?

 

[Joerg]

Hello Jim,

I live the fun life... if I need a bandgap I just roll one... designed
one last week that consumes only 5uA and comes fully alive at 1.5V
input. Includes a POR.

Makes me drool. Often I envy you guys, maybe I should get into chip
design some day. The few that I was involved in were lots of fun. But
the SW tools we used were freaking expensive and had a steep learning
curve even though someone else did the layout in all cases.

There was always that adrenaline rush during fab time. Will it work?
What if the data input turns out to be deaf? What if we pushed the
design rules too far this time? Tossing and turning at night. Luckily
they did work. Whew.

Regards, Joerg