Maker Pro
Maker Pro

Analog mux poor rise and fall time

Dear All,

I've been bitten by a slower rise and fall time than I was expecting
at the multiplexer ouput O in the following circuit, what have I
missed, please? You'll need to copy and paste into a constant width
font to see my ASCII, dare I say it, art :)

------
| U1 |
-------|A |
| | U2
-------1k--------|I1 | |\
| | | O|--------|+\
| ---1k--------|I2 | | >-----
| | | | | | |--|-/ |
1.1V0.1V 15n 15n ------ | |/ |
| | | | | |
--- --- --- --- |---91k--|
| |
|---27p--|
39k
|
---

U1 is a simplified representation of an MC14067 analogue multiplexer.
Two DC inputs have low pass filters of 1k/15nF.
The selection of these inputs is controlled by logic input A.
The multiplexer output drives an MCP601 CMOS opamp with a gain of 3.3'
and a low pass characteristic with 27pF in the feedback path.

When switching between the inputs I am finding a larger rise and fall
time at the multiplexer output, ~250us, than I am expecting.
I would have expected small rise and fall times (us at most) due to
the multiplexer's on resistance (say 1k) and output capacitance
(100pF).
Note I say at the multiplexer output. The inputs I1 & I2 are
unaffected during switching so I don't believe I'm seeing a charge
transfer effect. The effect is symmetrical so I don't believe I'm
seeing crosstalk from the control input.

....and what can I do about it?

regards & thanks for looking,
Colin Smith
 
J

John Devereux

Jan 1, 1970
0
Dear All,

I've been bitten by a slower rise and fall time than I was expecting
at the multiplexer ouput O in the following circuit, what have I
missed, please? You'll need to copy and paste into a constant width
font to see my ASCII, dare I say it, art :)

------
| U1 |
-------|A |
| | U2
-------1k--------|I1 | |\
| | | O|--------|+\
| ---1k--------|I2 | | >-----
| | | | | | |--|-/ |
1.1V0.1V 15n 15n ------ | |/ |
| | | | | |
--- --- --- --- |---91k--|
| |
|---27p--|
39k
|
---

U1 is a simplified representation of an MC14067 analogue multiplexer.
Two DC inputs have low pass filters of 1k/15nF.
The selection of these inputs is controlled by logic input A.
The multiplexer output drives an MCP601 CMOS opamp with a gain of 3.3'
and a low pass characteristic with 27pF in the feedback path.

When switching between the inputs I am finding a larger rise and fall
time at the multiplexer output, ~250us, than I am expecting.
I would have expected small rise and fall times (us at most) due to
the multiplexer's on resistance (say 1k) and output capacitance
(100pF).
Note I say at the multiplexer output. The inputs I1 & I2 are
unaffected during switching so I don't believe I'm seeing a charge
transfer effect. The effect is symmetrical so I don't believe I'm
seeing crosstalk from the control input.

I can't see anything wrong with it. Are the inputs all within the
supply range of the 4067? Other than that, I would suspect you've got
the pins wrong somehow. The real inputs are just floating?
 
J

John Devereux

Jan 1, 1970
0
Jan Panteltje said:
Use a decent mux, like the 74HC4053.
Switching times is measured in nanosconds (a lot of those),
and Ron is about 150 Ohm.
And there are 3 in one package.

There's something else wrong I think.
 
Use a decent mux, like the 74HC4053.
Switching times is measured in nanosconds (a lot of those),
and Ron is about 150 Ohm.
And there are 3 in one package.

Dear John & Jan,

thanks for the quick responses.
The inputs are all within the supply range. Vdd and inputs are 5V max.
The pins are right as far as I can tell - mature board, signals vary
at the output when varied at the input as expected.
I am increasing the performance in a mature product which has run a
lot slower in the past. From the datasheet I hadn't expected I had
already reached its limit :)
I'll try another board on Monday just in case there's something dodgy
with this one e.g. ESD.

In what way is the HC4053 a "decent mux" compared to the 14067? I am
aware the HC part is better comparing datasheets but the plain vanilla
CMOS part ought to be good enough by a long way already - I'm only
looking for rise and fall times of the order of 20us.
I need 16 channels so I can't use an HC405x. I've looked in Farnell
for an 74HC4067 but they only stock stupidly small outline (SSOP) and
my board has small outline (SO). I'd really, really like to avoid a
PCB relayout! When I next have a large order I'll buy some from
Digikey (I'm in the UK) who wack a hefty handling charge for orders
under £50.

One other thing I tried without effect was to disable the mux using
the INH input for ~5us either side of switching the logic input
forcing break before make and to reduce any charge transfer,

regards,
Colin
 
S

Spehro Pefhany

Jan 1, 1970
0
Dear All,

I've been bitten by a slower rise and fall time than I was expecting
at the multiplexer ouput O in the following circuit, what have I
missed, please? You'll need to copy and paste into a constant width
font to see my ASCII, dare I say it, art :)

------
| U1 |
-------|A |
| | U2
-------1k--------|I1 | |\
| | | O|--------|+\
| ---1k--------|I2 | | >-----
| | | | | | |--|-/ |
1.1V0.1V 15n 15n ------ | |/ |
| | | | | |
--- --- --- --- |---91k--|
| |
|---27p--|
39k
|
---

U1 is a simplified representation of an MC14067 analogue multiplexer.
Two DC inputs have low pass filters of 1k/15nF.
The selection of these inputs is controlled by logic input A.
The multiplexer output drives an MCP601 CMOS opamp with a gain of 3.3'
and a low pass characteristic with 27pF in the feedback path.

When switching between the inputs I am finding a larger rise and fall
time at the multiplexer output, ~250us, than I am expecting.
I would have expected small rise and fall times (us at most) due to
the multiplexer's on resistance (say 1k) and output capacitance
(100pF).
Note I say at the multiplexer output. The inputs I1 & I2 are
unaffected during switching so I don't believe I'm seeing a charge
transfer effect. The effect is symmetrical so I don't believe I'm
seeing crosstalk from the control input.

...and what can I do about it?

regards & thanks for looking,
Colin Smith

What are the power supply voltages? Those antediluvian COS/MOS MUX
circuits don't work so well at low supply voltages. They work even
less well if you accidentally leave the supply pins floating.
 
R

Raveninghorde

Jan 1, 1970
0
Dear All,

I've been bitten by a slower rise and fall time than I was expecting
at the multiplexer ouput O in the following circuit, what have I
missed, please? You'll need to copy and paste into a constant width
font to see my ASCII, dare I say it, art :)

------
| U1 |
-------|A |
| | U2
-------1k--------|I1 | |\
| | | O|--------|+\
| ---1k--------|I2 | | >-----
| | | | | | |--|-/ |
1.1V0.1V 15n 15n ------ | |/ |
| | | | | |
--- --- --- --- |---91k--|
| |
|---27p--|
39k
|
---

U1 is a simplified representation of an MC14067 analogue multiplexer.
Two DC inputs have low pass filters of 1k/15nF.
The selection of these inputs is controlled by logic input A.
The multiplexer output drives an MCP601 CMOS opamp with a gain of 3.3'
and a low pass characteristic with 27pF in the feedback path.

When switching between the inputs I am finding a larger rise and fall
time at the multiplexer output, ~250us, than I am expecting.
I would have expected small rise and fall times (us at most) due to
the multiplexer's on resistance (say 1k) and output capacitance
(100pF).
Note I say at the multiplexer output. The inputs I1 & I2 are
unaffected during switching so I don't believe I'm seeing a charge
transfer effect. The effect is symmetrical so I don't believe I'm
seeing crosstalk from the control input.

...and what can I do about it?

Daft question. Have you checked the supplies and switching waveform?
 
J

Joerg

Jan 1, 1970
0
Phil said:
Wow, a 4067--I last used that in 1981, to implement a 5-input logic
block using data selector logic. From the symptoms, you might have
forgotten to connect the power supply to the 4067.

That would also be my guess, missing supply connection. Along the lines
of "is it plugged in?" ... SCNR.
 
T

TTman

Jan 1, 1970
0
Jan Panteltje said:
Use a decent mux, like the 74HC4053.
Switching times is measured in nanosconds (a lot of those),
and Ron is about 150 Ohm.
And there are 3 in one package.

Better still max 4619.... <10 R on
 
R

Raveninghorde

Jan 1, 1970
0
Better still max 4619.... <10 R on

How is that better? He's feeding into an opamp from a 1k resistance so
150R or 10R on resistance isn't going to make a difference. You just
end up paying more for no improvement.
 
R

Raveninghorde

Jan 1, 1970
0
Use a decent mux, like the 74HC4053.
Switching times is measured in nanosconds (a lot of those),
and Ron is about 150 Ohm.
And there are 3 in one package.

If you're going to change the design make U2 a virtual earth with U1
feeding the -ve input. If you use a 4053 feed the signal into the
common pin and ground the terminal not going to U2. This gives a T
configuration and can take signals outside the supply range of U1.
 
J

John Devereux

Jan 1, 1970
0
Raveninghorde said:
How is that better? He's feeding into an opamp from a 1k resistance so
150R or 10R on resistance isn't going to make a difference. You just
end up paying more for no improvement.

Yep, in fact it's *worse* because you have got from a multi-sourced
commodity part to Maxim! I always try to go the other way...

I can't see anything much wrong with the original circuit, as drawn, I
think he must have just done something silly somewhere.
 
Yep, in fact it's *worse* because you have got from a multi-sourced
commodity part to Maxim! I always try to go the other way...

I can't see anything much wrong with the original circuit, as drawn, I
think he must have just done something silly somewhere.

--

John Devereux- Hide quoted text -

- Show quoted text -

Dear All,

thanks for your help - issue solved.
The general concensus has been it's something simple and obvious,
there's nothing unusual about the part or my application.
So this morning I've checked the power - fine. The logic select inputs
- ah, eh, they've got the same rise and fall as the analogue signals.
It's the bleeping 'scope probe compensation, idiot, embarrassed
sheepish red faced grin on my part.

Good point John raised about sourcing, engineering isn't just about
electronics, it's a multidimensional discipline isn't it? My only
worry about the 4067 rather than the HC part is long term
availability, and then I expect to be able to just change over, I'm
only using 5V. This analogue input structure has been a part of many
of my products for 6 years,

regards,
Colin
 
Top