Diodes instead of an OR gate

[Matt Jenkins]

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

[John Larkin]

On Wed, 22 May 2013 20:44:42 -0700 (PDT), Matt Jenkins <mdj...@gmail.com>
wrote:

Should work, assuming you have a pulldown after the diodes.

Actually, you can do it with one diode and one resistor:


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


I just did that today!


--

John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

[Jim Thompson]

On Wed, 22 May 2013 21:09:09 -0700, John Larkin
<jjla...@highNOTlandTHIStechnologyPART.com> wrote:

>On Wed, 22 May 2013 20:44:42 -0700 (PDT), Matt Jenkins <mdj...@gmail.com>
>wrote:
>
>>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
>
>Should work, assuming you have a pulldown after the diodes.
>
>Actually, you can do it with one diode and one resistor:
>
>
>A--------->|--------+----- A_OR_B
> |
>B--------\\\\\------+
>
>
>I just did that today!

Look up "DTL" and do it right. Don't be a hacker.

...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

[Matt Jenkins]

On Thursday, 23 May 2013 14:09:09 UTC+10, John Larkin wrote:

> 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

[Vladimir Vassilevsky]

Use pull-up.
Only lamerz use pull-down.

VLV

[Vladimir Vassilevsky]

Wow. Incredible.
Ever heard of single gate logic series?

VLV

[John Fields]

---
Using a pull-up will result in this:

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

--
JF

[Rich Webb]

On Wed, 22 May 2013 20:44:42 -0700 (PDT), Matt Jenkins
<mdj...@gmail.com> wrote:

>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.

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.

[Tauno Voipio]

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.

--

Tauno Voipio

[John Larkin]

On Wed, 22 May 2013 21:33:38 -0700 (PDT), Matt Jenkins <mdj...@gmail.com>
wrote:

Use a SOT23 dual diode!

The logic function I needed was actually

dipswitch
>> A----o/o-----|<-----+----- Y
>> |
>> B--------\\\\\------+

where Y can be B or A*B

[Spehro Pefhany]

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.

[Tim Wescott]

Not an option if it's through-hole. 'course, the OP hasn't said.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

[Tim Wescott]

You will lose on speed, power consumption, input drive needs, and if
you're not careful, available voltage overhead.

To get your voltage overhead, make sure to use pull-ups with the 74LS, or
make sure to only use the 74HCT.

[Vladimir Vassilevsky]

On 5/23/2013 11:18 AM, Tim Wescott wrote:
> On Thu, 23 May 2013 01:00:47 -0500, Vladimir Vassilevsky wrote:
>
>> On 5/22/2013 11:33 PM, Matt Jenkins wrote:
>>> On Thursday, 23 May 2013 14:09:09 UTC+10, John Larkin wrote:
>>>
>>>> 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.
>>
>> Wow. Incredible.
>> Ever heard of single gate logic series?
>
> Not an option if it's through-hole. 'course, the OP hasn't said.

Okay. Use a dual gate FET in metal can. Still need a resistor though.

VLV

[Vladimir Vassilevsky]

On 5/23/2013 10:10 AM, John Larkin wrote:
> On Wed, 22 May 2013 21:33:38 -0700 (PDT), Matt Jenkins <mdj...@gmail.com>
> wrote:
>
>> On Thursday, 23 May 2013 14:09:09 UTC+10, John Larkin wrote:
>>
>>> 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
>
> Use a SOT23 dual diode!

He uses TH components. TO-220 dual diode is his best option.

VLV

[whit3rd]

On Wednesday, May 22, 2013 8:44:42 PM UTC-7, Matt Jenkins wrote:

> 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.

[Jon Elson]

Vladimir Vassilevsky wrote:


>
> Wow. Incredible.
> Ever heard of single gate logic series?

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.

Jon

[John Devereux]

Jon Elson <jme...@wustl.edu> writes:

> Vladimir Vassilevsky wrote:
>
>
>>
>> Wow. Incredible.
>> Ever heard of single gate logic series?
>
> 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.

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

--

John Devereux

[rickman]

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.

--

Rick

[rickman]

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

--

Rick

[Vladimir Vassilevsky]

On 5/23/2013 4:13 PM, Jon Elson wrote:
> Vladimir Vassilevsky wrote:
>
>
>>
>> Wow. Incredible.
>> Ever heard of single gate logic series?
>
> 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.

Apparent weak layout problem.
At 5V, the Rds is in tens Ohm range. The 2A couldn't happen.
Anyway, the 3ns 2A spike shouldn't be big deal.

> 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.

Static shoot through current is ~10mA ballpark worst case.
Dynamic current depends on many other things.

Vladimir Vassilevsky
DSP and Mixed Signal Designs
www.abvolt.com

[Matt Jenkins]

On Thursday, 23 May 2013 15:49:25 UTC+10, Vladimir Vassilevsky wrote:

> 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.

[Matt Jenkins]

On Thursday, 23 May 2013 21:50:02 UTC+10, Rich Webb wrote:

> 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

[Matt Jenkins]

On Friday, 24 May 2013 01:30:58 UTC+10, Spehro Pefhany wrote:

> 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 is probably 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.

[whit3rd]

On Saturday, May 25, 2013 10:28:11 AM UTC-7, Vladimir Vassilevsky wrote:
> On 5/23/2013 4:13 PM, Jon Elson wrote:
>
> > Vladimir Vassilevsky wrote:

> > 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.

> 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.