High Side darlington array

[deperkin]

I know that the ULN2803 is a low-side driver, but I am looking for a high-side darlington if anyone knows of any.

It seems that the UDN2981 cannot be found currently (checking mouser and digikey).

I am currently programming a PIC18f4550 with 8 MCP23017's, which I have working well. Now I would like to drive some outputs and take in some inputs but the ULN2803's don't seem to fit my application.


Thanks for any direction here.

[Ttelmah]

Generally, most people use FET's these days. PNP arrays like the 2981, and 2982, have become relatively specialist items.
The 2981, can still be 'found', for several years though it has been a 'non stock' item, with suppliers like Digi-Key ordering them in as needed, not holding stock.
Look at FET's instead.
The TPS2085, 6 & 7, are typical devices.

Best Wishes

[deperkin]

thanks for the quick reply.

The only problem is that I have 128 I/Os...
compared to the 2981 (with 8 pins) these only have up to 4 outputs.

I like the TPS2085 but do you know of a similar with 8 outputs?

I also noticed that the TPS2085 only seems to be sold as surface mount which would make it very difficult for this proto considering the number of devices.

[John P]

Jameco has them, but are you willing to pay $2.49 each?

http://octopart.com/partsearch#search/requestData&q=udn2981

[deperkin]

Thanks I ordered them... yes they are expensive but it would cost me more to use the surface mount for now.

I will look at final design using the others however.

[deperkin]

I recieved the UDN2981's finally and tried them out.

Everything seems okay now (I power with +5V and inputs are +24V).

except.....

from the output of the darlington arrays I have wired the MCP23017 port expanders...

When I removed the 24v input from one of the UDN2981 pins (only testing a single chip at the moment) it seems to remain latched high on that pin. If i check with a voltmeter it drives the output of the UDN low.

I would like to have the output driven low as soon as I remove the input (or shortly after) but I do not want to use 128 pulldown resistors here...

anyone know any other options?

I had considered using a seperate output from the uC as a power source (with a single pulldown) to power the UDN's and after each read I could reset the UDN.

Any thoughts?

[asmboy]

post a schematic of what you have wired up
and you might get more insight
as well as help && suggestions from those of us who
do circuit design for a living .

s

[deperkin]

https://picasaweb.google.com/lh/photo/7t5Ul1bFHpA9MR92CAvLRLuM1AnbSS1d_whfZsoNDaM?feat=directlink

the only difference is that the 2803's are replaced with these highside drivers (2981's) now. (and of course the power and grounds are reversed).

I will try the TPS2085 as Ttelmah had suggested next, however I am not ready to solder this many surface mount devices for this proto build.

[Ttelmah]

Er....

"from the output of the darlington arrays I have wired the MCP23017 port expanders... "

Doesn't make any sense. The layout you have shown is:

Processor -> Port expander -> Driver.

The _outputs_ of the driver arrays, must not go anywhere near the port expanders. Instant release of the magic brown smoke.

Best Wishes

[Mike Walne]

You've only just got round to declaring that you're high side driver's working from 24V.

Begs the question, what else are you not telling us?

Solution(s) / analysis has (have) been posted on the forum before.

Please give us more detail on EXACTLY what you are doing.

Then, as asmboy suggested, you may (probably will) get some guidance.

Mike

EDIT

Tell us how many high side and low side loads you've got.
Give us a clue about current/load levels and what you're driving.

[deperkin]

Ttelmah,

You are correct. I should have proofread before submit.

i have uC --> MCP --> UDN.

as far as application I can see the use of this information in general but right now I do not have anything hooked up to the UDN's.
I have changed the schematic on one of the MCPs like this:
a) I have one UDN connected to port A of the MCP that I am using as an output channel. It is powered with 24V (giving me a 24V output)... this works well.
b) I have another UDN connected to port B of the MCP (e.g. outputs of UDN connected to port B) and this is connected to 5V, giving me 5V inputs.
my input voltages are 24V going to this UDN.

my problem currently is that when I apply 24V as an input to this second UDN I do get the appropriate input to the MCP and seen on the uC. If I remove this 24V input and place it onto another pin the first pin still remains high.

As an example: If I apply 24V to pin 1 of this second UDN I see at the micro a value of 0x01 (correct). If I move this 24V to pin 2 now I see a value of 0x03 (incorrect; should see 0x02), and if I next move the 24V input to pin 3 i now see a value of 0x07 (also incorrect; should see 0x04).

you see my point? The pin remains latched high for lack of a better word.

For my application the I/Os will vary (some connected to 24V LEDs, some connected to other controllers, etc) I dont expect near 0.5A on any I/O and more like 25-50 mA.

I would like to get this working without it being connected to anything right now. The outputs work well (since I reset the MCP's to essentially turn off the output). The inputs work, but remain high after the input is removed.

Also another point, if I removed the 24V (pin1) and move to pin2 I see the value 0x03 as stated above. now if I check the voltage with a multimeter between the UDN and MCP on the first output (e.g. at pin 18 of the UDN which is connected to pin1 of the MCP) I now get a value of 0x02 at the uC, which is the value I wanted. So checking this voltage with the multimeter seems to pull the output of the UDN from a float to a low. Am I explaining the problem clearly enough?

[deperkin]

Mike Walne,

To explain exactly what I want to accomplish with this:

I have a machine that drives alot solenoids via relays. There is a Box that has 4 controllers that are used for various functions on the machine. What I want to do is connect to this box to simulate machine functionality for validating controller and wiring continutity. This is for diagnostics purposes really. So there are no solenoids or relays connected at all here. The outputs of this test unit (inputs to the box) will simulate switches on an actual machine. The inputs of this test unit (outputs from the box) will simulate loads and be used solely for checking that the box has the proper pinout and there is continuity from the boxes controllers to the appropriate pin.

Is this enough information? (never!)

The I/O's are very low current, but are 24V. I am only really using the 2981's as buffers and drivers here since the MCP cannot take the voltages I am using.

[Mike Walne]

[deperkin]

I am trying to build a test bench for a control box that drives solenoids and uses approximately 90 switches.

Control Box <-------> Test Bench

There are 4 controllers in the Control box (i cannot give you schematics or drawings for these since that is propriaterly information). These controllers have been used for more than 5 years. The I/Os for these controllers are low current.

The testbench I am trying to build will simply simulate switch input (24V) and expect certain outputs from the controllers (24V).

Maybe this is my problem however.... I do not have a level conversion. I am applying 24V directly to the UDN i am using as input to the MCP...

Are there any good recommendations for a 24V to 5V level converter that has 8 channels? (i.e. buck regulator, etc)

I am not trying to be vague here, but my point is that I currently do not have this hooked to the Control box at all, I am using a 24V supply to simulate I/O's to the testbench I am building.

[Ttelmah]

'Low current', means different things to different people. Under 1A?. Under 100mA?. Under 10mA?.
If the last, then standard opto's are the way to go. Just drive the LED from your logic outputs (with a suitable load resistor), and wire the output transistor with the collector to 24v, and the emitter to your load.
Something like a TLP504, can deliver up to a maximum of 50mA, with about 16mA 'recommended'.

Best Wishes