Maker Pro
Maker Pro

Darlington array in parallel.

J

J3nn1f3r L0p32

Jan 1, 1970
0
Hi guys.

I have a a couple of ULN2003a darlington arrays at my disposal.
The big bummer is that they can only cream out around 500mA
per channel. And i need abit more (around 800mA).

I was thinking... would there be any problem if i just
hooked them up in paralell so i can get 1A per channel.
I just want to know if its possible so i dont fry any
more of these babys in vain. =)

todilo
//L0p32
 
P

Pieter Hoeben

Jan 1, 1970
0
I have a a couple of ULN2003a darlington arrays at my disposal.
The big bummer is that they can only cream out around 500mA
per channel. And i need abit more (around 800mA).
I was thinking... would there be any problem if i just
hooked them up in paralell so i can get 1A per channel.
I just want to know if its possible so i dont fry any
more of these babys in vain. =)
What will happen is:
one of the two conducts a little more than the other. It therefor gets
most of the current, and gets even hooter. Because of the temparature
rise it will conduct better, and get even hotter. Ends with smoke.

To use them parallel you could use small series resistors.

Pieter
 
I

Iman Habib

Jan 1, 1970
0
Pieter Hoeben said:
What will happen is:
one of the two conducts a little more than the other. It therefor gets
most of the current, and gets even hooter. Because of the temparature
rise it will conduct better, and get even hotter. Ends with smoke.

To use them parallel you could use small series resistors.

Pieter

This is not true. Atleast not for the the ULN2003A.

As i had to drive a stepper that wanted to a bit more current
than the driver could handle i chose to set up two of them in paralell.
Inddeed it did not get at all as hot as when only one driver. And the
whole thing ran for several hours.

But there was a strange thing on the other hand.

I tried to mesure the current running trough the "second" driver
and my multimeter would not give me any values. As if no current
was going through it. I kind of suspect that this was due to small
differences in the swithing time of the darlingtons (the cable length
might have been a factor to maybe?) and the slow update rate
on my multimeter.


cheers

//iman
 
C

Costas Vlachos

Jan 1, 1970
0
Iman Habib said:
This is not true. Atleast not for the the ULN2003A.



I've used power Darlingtons in the past, and what Pieter says is true. When
connecting BJTs in parallel you must use current sharing resistors. Another
very important thing that many designers fail to realise is that the BJTs
must be mounted on the SAME heatsink, so that they are thermally coupled.
Otherwise (even with the series resistors) one of the two will dominate and
pass most of the current (and pop eventually).

Now, the ULN2003A has 7 Darlingtons inside it. This means that they are
thermally coupled (being in the same IC), so there should be some current
sharing because of that. I've never used the ULN2003A, but have used several
power BJTs and the above are definitely true.

cheers,
Costas
 
K

Kevin Aylward

Jan 1, 1970
0
Costas said:
I've used power Darlingtons in the past, and what Pieter says is
true. When connecting BJTs in parallel you must use current sharing
resistors. Another very important thing that many designers fail to
realise is that the BJTs must be mounted on the SAME heatsink, so
that they are thermally coupled. Otherwise (even with the series
resistors) one of the two will dominate and pass most of the current
(and pop eventually).

Now, the ULN2003A has 7 Darlingtons inside it. This means that they
are thermally coupled (being in the same IC), so there should be some
current sharing because of that.

What is much more significant, is that the devices in one package will
have vbe's that match!, Typically within a few mv.

I've never used the ULN2003A, but
have used several power BJTs and the above are definitely true.

Because device to device variations in Vbe migt be say, 50mv. A 60mv
differance is a 10:1 ratio in current.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
 
C

cpemma

Jan 1, 1970
0
J3nn1f3r said:
Hi guys.

I have a a couple of ULN2003a darlington arrays at my disposal.
The big bummer is that they can only cream out around 500mA
per channel. And i need abit more (around 800mA).

I was thinking... would there be any problem if i just
hooked them up in paralell so i can get 1A per channel.
I just want to know if its possible so i dont fry any
more of these babys in vain. =)

todilo
//L0p32

Read the datasheet.

"The ULN2801A-ULN2805A each contain eight darlington transistors with common
emitters and integral suppression diodes for inductive loads. Each
darlington features a peak load current rating of 600mA (500mA continuous)
and can withstand at least 50V in the off state.

*Outputs may be paralleled for higher current capability.*"
 
G

Guy Macon

Jan 1, 1970
0
Costas Vlachos said:
I've used power Darlingtons in the past, and what Pieter says is true. When
connecting BJTs in parallel you must use current sharing resistors. Another
very important thing that many designers fail to realise is that the BJTs
must be mounted on the SAME heatsink, so that they are thermally coupled.
Otherwise (even with the series resistors) one of the two will dominate and
pass most of the current (and pop eventually).

Now, the ULN2003A has 7 Darlingtons inside it. This means that they are
thermally coupled (being in the same IC), so there should be some current
sharing because of that. I've never used the ULN2003A, but have used several
power BJTs and the above are definitely true.

The fact that the 7 Darlingtons are on the same die not only works
like having them on the same heatsink only better, but also makes
it likely that they will be closer matches electrically. You might
even luck out and find that the individual device you are using
does do some sharing without the resistors, but it would be a house
of cards ready to topple, and might not work at all with another
device. The resistors are the robust solution.
 
G

Guy Macon

Jan 1, 1970
0
cpemma said:
Read the datasheet.
http://www.w-r-e.de/robotik/data/treiber/uln2003an.pdf

"The ULN2801A-ULN2805A each contain eight darlington transistors with common
emitters and integral suppression diodes for inductive loads. Each
darlington features a peak load current rating of 600mA (500mA continuous)
and can withstand at least 50V in the off state.

*Outputs may be paralleled for higher current capability.*"

Point well taken.

Figures 11 and 12 tell you how well you can expect it to work.
Note that it's better than you would get with two seperate devices
with seperate heatsinks (which work about the same as one device)
but not as good as two perfectly matched devices would be (they
would be twice as good as a single device).

Maybe I am just old fashioned, but I would put in the resistors.
You could get away with smaller resistor values, though.
 
Top