# ULN2803 ok to be driven by 3.3V logic?

J

#### Joerg

Jan 1, 1970
0
Gents,

Have to drive a bucket full of 12V relays but the uC is only 3.3V. The
ULN2803 is touted as a driver for 5V TTL logic by most companies (like
TI) but datasheets hint that it can happily live down to around 3V drive
level. Is that kosher?

http://www.toshiba.com/taec/components2/Datasheet_Sync//393/22738.pdf

App note SLAA148 suggests it is kosher but ... I figure I'd better ask
the famous gurus in this here group

http://focus.ti.com/lit/an/slaa148/slaa148.pdf

A FET version would be nice as it gets the outputs lower but that's only
available from the more boutiquish suppliers. Demasiados Dolares, as usual.

J

#### Joerg

Jan 1, 1970
0
Jim said:
Should work just fine.

I'm quite surprised there isn't a MOS version.

So am I. There are some but with boutique pricing. Not surprisingly many
of them didn't make it in the marketplace because guys like me wouldn't
use them. Here is a decoder type which can be used in "memory mode", and
it also went to lalaland:

http://www.allegromicro.com/en/Products/Part_Numbers/6259/6259.pdf

I don't keep track of off-the-shelf stuff. Have you looked at
open-drain buffers/inverters?

Yeah, they are either too wimpy or way more expensive than two dozen
discrete parts. This time I need to drive relays with up to 60mA coil
current each.

J

#### Joerg

Jan 1, 1970
0
Jim said:
A devious loaded question... What would you consider a reasonable
price for such a device? Max output voltage? Max output load
current?

30c/1k, about 15c/100k, 30V, 300mA but that will be bond wire limited
for the total. _Absolutely_ crucial: Industry standard pinout so it can
be a drop-in for the ULN2803. Non-standard might work but only if your
entrepreneur owns a company the size of Texas Instruments ;-)

I'm working with an entrepreneur... sold his last company and is
looking at new ventures, new parts.

We just spun a specialty part aimed at surveillance camera controls.

Oooh, big brother ... Should be a good market in the UK, they seem to
have every single pine cone under 24/7 surveillance.

We'll take a look at the relay/low-side/high-side market and see what
looks realistic.

Don't be surprised if you get a call from Kiwi Semiconductors ;-)

Anytime. But I am usually more into fast stuff when it comes to power
drivers. Injectors, transducers, pulsers et cetera.

J

#### Joerg

Jan 1, 1970
0
Jim said:
A devious loaded question... What would you consider a reasonable
price for such a device? Max output voltage? Max output load
current?

P.S.: If you guys are eyeing the automotive and aerospace markets as
well it needs to be at least 60V. I prefer 100V parts there, but you
know automotive quite well. Avoids having to place a surge cutter. And
don't forget them thar catch diodes.

[...]

J

#### Joerg

Jan 1, 1970
0
John said:
TPIC6595 is cool, serial in and eight fet drains out, but it needs 5
volts. We're using them to drive a lot of relays.

A buck fifty wouldn't fly in this case, serial would work well though.
However, I have (almost) been burned by TPIC parts in the past, by part
obsolescence. Somehow I had a hunch and chickened out. Now I am glad I
did. But this one is cool if you have the budget and it's been around
for almost two decades. If anyone is interested:

J

#### Joe G $$Home$$

Jan 1, 1970
0
Joerg said:
Gents,

Have to drive a bucket full of 12V relays but the uC is only 3.3V. The
ULN2803 is touted as a driver for 5V TTL logic by most companies (like TI)
but datasheets hint that it can happily live down to around 3V drive
level. Is that kosher?

http://www.toshiba.com/taec/components2/Datasheet_Sync//393/22738.pdf

App note SLAA148 suggests it is kosher but ... I figure I'd better ask the
famous gurus in this here group

http://focus.ti.com/lit/an/slaa148/slaa148.pdf

A FET version would be nice as it gets the outputs lower but that's only
available from the more boutiquish suppliers. Demasiados Dolares, as
usual.
Infineon SPIDER Driver IC's - Multi channel SPI Driver?

See here http://bit.ly/5xVmH8

Parallel and Serial Isolated ISOFACE Driver?
http://bit.ly/5GrCFF

Joe

J

#### Joerg

Jan 1, 1970
0
Joe said:
Infineon SPIDER Driver IC's - Multi channel SPI Driver?

See here http://bit.ly/5xVmH8

Thanks. However, it costs almost two bucks in qties and Digikey has no
stock (which is always a red flag for me). Yikes. That's almost ten
times the cost of the ULN parts, not a chance. What were they thinking?

Parallel and Serial Isolated ISOFACE Driver?
http://bit.ly/5GrCFF

Those are very nice indeed when you need isolation (I often do). But
also non-stock and they'd have to compete with the usual optocoupler
plus driver solution. Considering that they want $6 for them that'll be a tall order for their sales guys. I wish Infineon would come off their key account thinking and stuff the channels, such as Digikey. And adjust their pricing. Their R&D develops top notch products but I often shun them because of past purchasing problems. N #### Nico Coesel Jan 1, 1970 0 Joerg said: Gents, Have to drive a bucket full of 12V relays but the uC is only 3.3V. The ULN2803 is touted as a driver for 5V TTL logic by most companies (like TI) but datasheets hint that it can happily live down to around 3V drive level. Is that kosher? I never had problems driving it with 3.3V. Its a darlington array after all. Not mosfets. N #### Nico Coesel Jan 1, 1970 0 Joerg said: Thanks. However, it costs almost two bucks in qties and Digikey has no stock (which is always a red flag for me). Yikes. That's almost ten times the cost of the ULN parts, not a chance. What were they thinking? Those are very nice indeed when you need isolation (I often do). But also non-stock and they'd have to compete with the usual optocoupler plus driver solution. Considering that they want$6 for them that'll be
a tall order for their sales guys.

I wish Infineon would come off their key account thinking and stuff the
channels, such as Digikey. And adjust their pricing. Their R&D develops
top notch products but I often shun them because of past purchasing
problems.

The first thing Infineon needs to do is learn how to write datasheets.
Back in my design-stuff-for-telecom days I noticed Zarlink (formerly
Mitel) could explain the same in 20 pages where Infineon needed 200
pages to explain nothing at all. Guess what made it to the PCB.

N

#### Nico Coesel

Jan 1, 1970
0
John Larkin said:
TPIC6595 is cool, serial in and eight fet drains out, but it needs 5
volts. We're using them to drive a lot of relays.

Those also work at 3.3V. There is an appnote from TI. I don't know
about the second source version from (IIRC) ST though.

S

#### Spehro Pefhany

Jan 1, 1970
0
I never had problems driving it with 3.3V. Its a darlington array
after all. Not mosfets.

It should be fine-- "5V" TTL has output voltage in the 3.3V range when
sourcing current, but I'd get nervous if driving a lot of output
current and the nominal supply is less than 3.3V.

I have not been able to use them in some cases because the output
voltage drop is too high (Darlington). A CMOS version with low Rds(on)
would be nice, but it has to be really cheap (the bipolar 7-output
ULN2003A is under a dime).

J

#### Joerg

Jan 1, 1970
0
Nico said:
I never had problems driving it with 3.3V. Its a darlington array
after all. Not mosfets.

Thanks. I was just wondering about resistor tolerances in there, can be
quite high in chips. But the datasheet sez ok and you plus others also
said it, so that gives me the warm and fuzzy about it.

J

#### Joerg

Jan 1, 1970
0
Spehro said:
It should be fine-- "5V" TTL has output voltage in the 3.3V range when
sourcing current, but I'd get nervous if driving a lot of output
current and the nominal supply is less than 3.3V.

Under 100mA in my case and we can live with the 1V or so of drop because
it's driving 12V relays.

I have not been able to use them in some cases because the output
voltage drop is too high (Darlington). A CMOS version with low Rds(on)
would be nice, but it has to be really cheap (the bipolar 7-output
ULN2003A is under a dime).

The drop also causes it to warm up. Hey, an anti-icing feature

Wish I could use the 2003 but it's eight relays. Darn. I never
understood why nobody came out with a CMOS equivalent. It would be ok if
it had only 30V or so, since 2.7V logic level FETs are difficult on a
cheap process if you want higher voltages.

M

#### Mark Zenier

Jan 1, 1970
0
Gents,

Have to drive a bucket full of 12V relays but the uC is only 3.3V. The
ULN2803 is touted as a driver for 5V TTL logic by most companies (like
TI) but datasheets hint that it can happily live down to around 3V drive
level. Is that kosher?

http://www.toshiba.com/taec/components2/Datasheet_Sync//393/22738.pdf

So, get some ULN-2805A. Or ULN-2801 and pick your own input resistors. ;-)

A graph of Iin vs. Vin for the '03, in the 22 year old Sprague datasheet,
shows the normal operating range is expected to be about 2 1/3 and
about 3 2/3 volts. Giving an input current of .5 to 1.25 mA.

TTL Voh was pretty limp, so CMOS at 3.3 can't be much worse.

Mark Zenier [email protected]

N

#### Nico Coesel

Jan 1, 1970
0
Joerg said:
Under 100mA in my case and we can live with the 1V or so of drop because
it's driving 12V relays.

The drop also causes it to warm up. Hey, an anti-icing feature

Wish I could use the 2003 but it's eight relays. Darn. I never
understood why nobody came out with a CMOS equivalent. It would be ok if
it had only 30V or so, since 2.7V logic level FETs are difficult on a
cheap process if you want higher voltages.

IIRC we use the TSM2302 N channel MOSFET a lot for small relays. Dirt
cheap but only up to 20V though. And beware there is also a P channel
datasheet which is wrong

N

#### Nico Coesel

Jan 1, 1970
0
Joerg said:
A buck fifty wouldn't fly in this case, serial would work well though.
However, I have (almost) been burned by TPIC parts in the past, by part
obsolescence. Somehow I had a hunch and chickened out. Now I am glad I
did. But this one is cool if you have the budget and it's been around
for almost two decades. If anyone is interested:

ST has some equivalents so thats not bad. IMHO this series of chips
have become an industry standard like the NE555, ULN2003 and the 8051.

J

#### Joerg

Jan 1, 1970
0
Nico said:
ST has some equivalents so thats not bad. IMHO this series of chips
have become an industry standard like the NE555, ULN2003 and the 8051.

But for that they would have to be << 50 cents

J

#### Joerg

Jan 1, 1970
0
Jim said:
Cheapskate ;-)

When it comes to circuit design I am, when it comes to beer, wine and
cheese then I am not

Once I had an inductor in a design and it cost something like a buck
sixty. Was quite happy that I got away with that one by upping a
switcher frequency. While talking with the client they asked whether
there would possibly be a cheaper inductor.

Remember when some guys here thought I was nuts for suggesting that
resistors other than 1% are still useful? An assembler in China once
sent a formal request via my client, to identify those that could be
increased from 10% to 20%. That is not a joke. 15 minutes of billed time
later they had the list. Don't remember exactly but I believe it saved
something like 30 bucks. Per month, and it's now 15 years and counting ...

J

#### Joerg

Jan 1, 1970
0
Mark said:
So, get some ULN-2805A. Or ULN-2801 and pick your own input resistors. ;-)

Thanks, Mark. However, Digikey has only DIP and stock on the first one
is zero, very low on the second one. That always gives me the goose bumps.

A graph of Iin vs. Vin for the '03, in the 22 year old Sprague datasheet,
shows the normal operating range is expected to be about 2 1/3 and
about 3 2/3 volts. Giving an input current of .5 to 1.25 mA.

TTL Voh was pretty limp, so CMOS at 3.3 can't be much worse.

Yep, that's right, old TTL never went much past 3V.

Replies
19
Views
5K
D
Replies
4
Views
10K
Jasen Betts
J
L
Replies
10
Views
5K
Mike Page
M
Replies
4
Views
2K
Replies
2
Views
606