Maker Pro
Maker Pro

Diodes instead of an OR gate

M

Matt Jenkins

Jan 1, 1970
0
Hi,

I've recently designed a circuit in 74LS (that I plan to redo in 74HCT) that requires just a single OR gate. Rather than haul in an entire IC I was intending to replace this with Schottky diodes on each line.

I couldn't find a lot of information online about this being done, so I thought I would ask the good people here if there's any unwanted side-effects (beside the forward voltage drop) that I should be aware of.

TIA

Matt
 
M

Matt Jenkins

Jan 1, 1970
0
A--------->|--------+----- A_OR_B

|

B--------\\\\\------+


Thanks. I opted for:

A--->|---+---- A_OR_B
|
B--->|---+
|
\
/
\
/
|
GND

Cheaper and more convenient for board layout than using one quarter of an 'LS32 was my thinking.

Matt
 
T

Tauno Voipio

Jan 1, 1970
0
Hi,

I've recently designed a circuit in 74LS (that I plan to redo in 74HCT) that requires just a single OR gate. Rather than haul in an entire IC I was intending to replace this with Schottky diodes on each line.

I couldn't find a lot of information online about this being done, so I thought I would ask the good people here if there's any unwanted side-effects (beside the forward voltage drop) that I should be aware of.

TIA

Matt

With 74LS circuits, your diode gate is a bad idea: The TTL logic
is designed to sink current from the inputs into the outputs, and
your gate is going to do just the opposite.

With CMOS, the situation is easier, but the gate will be sloow.
 
S

Spehro Pefhany

Jan 1, 1970
0
---
Using a pull-up will result in this:

A B Y
---|---|---
0 0 1
0 1 1
1 0 1
1 1 1

On the bright side, it does guarantee a proper LS TTL or CMOS level.

Instead of using diodes, you can use a Diodes Inc. 74AHCT1G32SE-7
SOT-23-5

http://www.diodes.com/datasheets/74AHCT1G32.pdf

The diodes could work okay (assuming HCMOS), but there will be a time
constant on turn-off related to the resistor value and input+stray
capacitance, and, of course, there will be static power dissipation
due to the resistor if either input is high, and you have to make the
resistor high enough to guarantee enough noise immunity at the output
taking into account both the diode drop and the output voltage under
load.
 
W

whit3rd

Jan 1, 1970
0
I've recently designed a circuit in 74LS (that I plan to redo in 74HCT) that requires just a single OR gate. Rather than haul in an entire IC I was intending to replace this with Schottky diodes on each line.

Don't. The low threshold for TTL is VERY close to the highest LS-series output-low specification.
The best way to do an OR with LS TTL is with open-collector gates (in negative logic, two
open collectors wired together with a pullup resistor, is an OR function).

If this was CMOS and high voltage (5V or more), the diodes would be a good option.

74LS00 needs 0.8V to reliably indicate LOW, and drives only to 0.4V (loaded). There's
not any margin left if you add a diode drop.
 
J

John Devereux

Jan 1, 1970
0
Jon Elson said:
Vladimir Vassilevsky wrote:



You have to be careful with some of these.
I used the 74LVC1G00 family in something and had
horrible noise/crosstalk issues.
After some fooling around,
I discovered this family had about 2 A of shoot-through
whenever the gates switched. The shoot-through lasted
about 3 ns. It may be that parasitic inductance made
the current look worse than it really was, but I was
astounded by the magnitude of this current pulse.

Haha, yes, try the 74LVC245, 8 outputs switching simultaneously. My
introduction to shoot-through currents.
I had to redesign the whole thing, taking the time to
move from 5 V down to 3.3 V and switching to the
74AUP1G00 family. The power supply noise was not
even measurable! (Looking at the datasheets, I chose
this particular family because I assumed that the
equivalent switching capacitance must be related to
shoot-through and timing of the output drivers, and
it was the lowest value I could find.)

In my case I was able to substitute a milder LVX part.
 
R

rickman

Jan 1, 1970
0
Vladimir Vassilevsky wrote:



You have to be careful with some of these.
I used the 74LVC1G00 family in something and had
horrible noise/crosstalk issues. After some fooling around,
I discovered this family had about 2 A of shoot-through
whenever the gates switched. The shoot-through lasted
about 3 ns. It may be that parasitic inductance made
the current look worse than it really was, but I was
astounded by the magnitude of this current pulse.

I had to redesign the whole thing, taking the time to
move from 5 V down to 3.3 V and switching to the
74AUP1G00 family. The power supply noise was not
even measurable! (Looking at the datasheets, I chose
this particular family because I assumed that the
equivalent switching capacitance must be related to
shoot-through and timing of the output drivers, and
it was the lowest value I could find.)

Of course, nobody directly talks about shoot-through
in their datasheets.

I've never tried to quantify noise from digital logic. I assume a
direct decoupling cap of very low inductance didn't help? A 3 ns
shoot-through short is just the sort of thing the power distribution
system (PDS) is supposed to keep out of the other logic and analog.

Did you ever figure out exactly how it was getting into the rest of the
board? Common resistance/inductance in the power/ground traces/planes
maybe? A better topology for your PDS might have fixed this.
 
R

rickman

Jan 1, 1970
0
He uses TH components. TO-220 dual diode is his best option.

TO-220 is relatively huge. Diodes are common in DIP packages which
would be much smaller.
 
M

Matt Jenkins

Jan 1, 1970
0
Use pull-up.

Only lamerz use pull-down.

Perhaps you should consider the problem I'm trying to solve before posting. Clearly, this will not work.
 
M

Matt Jenkins

Jan 1, 1970
0
Take a look at Fairchild's NC7SZ57 / NC7SZ58 chips. One of those may

be smaller than one diode package (or at least not much larger) and

with one or the other of them you can get pretty much any two-input

logic gate. Really handy for one-offs and prototypes.

And at $0.41 in single unit quantities, they're cheaper and simpler to implement than the SOT dual schottky I was going to use instead.

Thanks very much, Rich. Very cool little chips.

Matt
 
M

Matt Jenkins

Jan 1, 1970
0
Instead of using diodes, you can use a Diodes Inc. 74AHCT1G32SE-7

SOT-23-5



http://www.diodes.com/datasheets/74AHCT1G32.pdf

This would also do the trick!
The diodes could work okay (assuming HCMOS), but there will be a time

constant on turn-off related to the resistor value and input+stray

capacitance, and, of course, there will be static power dissipation

due to the resistor if either input is high, and you have to make the

resistor high enough to guarantee enough noise immunity at the output

taking into account both the diode drop and the output voltage under

load.

Yeah. In the circuit in question, both inputs to the OR are normally high, and one occasionally strobes low, and I'm wanting to propagate the strobe only when the other input is also low. Speed isn't a terrible concern since the strobes would be at worst < 300Khz, but the static power dissipation isprobably worth avoiding since the rest of the circuit (if implemented in HCT) has tiny power consumption so even though it doesn't matter a lot it's probably aesthetically poor form to waste current to save a small space.
 
W

whit3rd

Jan 1, 1970
0
At 5V, the Rds is in tens Ohm range. The 2A couldn't happen.

That Rds is for the OUTPUT stage; there's also internal stages
in most CMOS, and they add to the shoot-through currents.
Anyway, the 3ns 2A spike shouldn't be big deal.

Look up 'exploding wires'; I've seen pass transistors feeding
CMOS circuits with entirely missing bond wires, after failing
under a load that my VOM claimed was sub-milliamp. Plastic-case
transistors work better, in this kind of surge situation.
 
Top