DC-16 and ULN2803/ULN2003 increasing current capacity

[jukingeo]

Hello All,

I was wondering if there was a higher current capacity ULN2803/ULN2003 available that could handle twice the current.   I am planning an application around the DC-16 as a small bulb / relay coil driver.   Using all 8 channels of a ULN2803 at the same time would result in a 125ma current capacity for each channel.   As it is, this is just shy of the #47 bulb...which is very popular general use and it consumes 150ma.  However, there are other bulbs and even relay coils that I intend to use that are over the 125ma/ chan figure.   So if there was a ULN2803 with just double the current capacity that would solve my current needs.

Sure I could build an outboard transistor circuit and in some cases I still have to do that for larger current draw items, but for general single bulb flashing, I was hoping there was an easier solution.

One thing that came to mind is stackability.   Can the ULN2803/ULN2003 be stacked? (IE, a header is put in the IC socket that would allow two ULN's to be conneced together on a single position.   Would this effectively double the current capacity per channel?

Using a Prop 2 and 4 DC-16's would allow a system that can take 16 inputs and 64 outputs, which isn't anything to sneeze at.

For those curious as to why I am looking at so many inputs and outputs, I am exploring the concept of using Basic Stamps for a control system in a pinball machine.

Thank You.

Geo

[JonnyMac]

According the the chart below, you can get about 175 mA per channel with all active.  This is the spec that I've used for a long time and have never had any problems.  Remember, too, that it's unlikely that all outputs will be on all the time, so you have a little bit of wiggle room.

[JonnyMac]

For additional safety, you might be able to find a heat sink for the ULNs.  The problem with excessive current flow is that it builds up heat in the chip; if you can dissipate this heat you can run more current through.  Maybe there's a way to put a fan that blows over the DC-16s to help with cooling as well.

[menehune]

Digikey 14-16 pin DIP heatsink
same sink at Kronos robotics

The sink looks like a smaller version of the 20 pin sink I have attached to my 80287 co-processor chip.  The sink won't fully extend past the end of the chip, so the retaining clip won't lock the sink to the DIP package.  The spring clips are plenty strong on my '287 and the sink is held very snugly to the chip body.

You may also be able to "glue" a stick on video ram heatsink onto the chip if you have any lying around.  newegg example currently out of stock.  You can also get ram heatsinks with passive cooling solutions for VGA video cards.

[JonnyMac]

The ULN2803 has 18 pins so you may be tight with that socket -- perhaps there is a similar model for 18-20 pins.

[menehune]

Digikey does list 18 pin version and 20 pin versions, but they are in "non-stock" status so you would need to purchase a minimum order (2000 pcs?) to obtain them. 

The sink looks like a smaller version of the 24 pin sink I have attached to my 80287 co-processor chip (40 pins).  The sink won't fully extend past the end of the chip, so the retaining clip won't lock the sink to the DIP package.  The spring clips are plenty strong on my '287 and the sink is held very snugly to the chip body.

[jukingeo]

According the the chart below, you can get about 175 mA per channel with all active.  This is the spec that I've used for a long time and have never had any problems.  Remember, too, that it's unlikely that all outputs will be on all the time, so you have a little bit of wiggle room.

Hello Jon,

You know that I usually have very weird applications for Stamps that do run outside of the context of just using them for haunted attractions.  Chasers for one.   While I was exploring the options of how many inputs and outputs I could get out of the Prop - 2 I realized that for control of a pinball machine, I would have to enter the DC-16.  With four of them hooked up and at 64 outputs...that isn't anything to sneeze at.  But there are many instances that I do have to take into consideration that all the outputs will be on at one time.   Plus pinballs can be set to what is called an 'attract' mode...which is something that involves chasing as well.

Getting back on track, so now even though the documentation states that the ULN chips should be able to drive 125ma, you are saying it can actually do 175 with all 8?  That is interesting as it would solve my problem with driving the #47 bulb directly (which draws 150ma).  It may be an issue with some relays though, so I still would be interested in something that has a bit higher capacity.   What I really want is something that can get 500ma per channel for 8 channels (on at the same time) in a monolithic package.  As I said in my original post, I should be even fine with a 250ma drive.

This is a very ambitious project, and I do want to keep it as simple as possible.

Thanx,
Geo

[jukingeo]

For additional safety, you might be able to find a heat sink for the ULNs.  The problem with excessive current flow is that it builds up heat in the chip; if you can dissipate this heat you can run more current through.  Maybe there's a way to put a fan that blows over the DC-16s to help with cooling as well.

Hello Jon,

That was something I was going to ask you.  Since any individual channel capacity is 500ma, I figured the limitation was thermal rather than electrical.  I noticed that some versions of the ULN chips actually have metal tops, and I did think about getting those and gluing a heat sink to the top.  But glue isn't a good heat conductor so either I would have to find some other way to get a heat sink on the ULNs or perhaps find a bigger ULN.

Do you know if they make a higher capacity ULN that would fit on the Prop-2 (or DC-16)?

I like the ULN's because they offer the diode protection and can drive coils/relays directly too.  The trouble is the capacity.

Geo

[jukingeo]

The ULN2803 has 18 pins so you may be tight with that socket -- perhaps there is a similar model for 18-20 pins.

What about the other alternative (if a larger IC isn't available).  Can you stack the ULN's?  I know this is an old trick with the 78xx voltage regulators in that you can stack them for more current capacity.

[jukingeo]

Hello guys,

I did a bit of digging on line and I came up with a possible alternative solution, it would require a bit more work on my part in terms of wiring, but current capacity would be more than enough.

The solution, interface a prop-2 with an actual pinball driver board:

Take a look:

http://www.pinrepair.com/sys1/driver2.gif

This is a schematic for a board that goes into a Gottlieb pinball machine.

Notice that the drive lines are part of a matrix.  A single DC-16 will use 13 lines total to drive the lamp matrix.  There are 4 L-lines and 9 D-lines.  This sets up a matrix of 36 outputs and there would be another 3 lines available to drive part of the direct (off matrix) outputs on the right.   With using a DC-16, this would only require an additional 5 lines from the prop-2.  Thus, with two DC-16's and two of these driver boards, a single BS-2 will drive a total of 88 outputs from a single prop-2, with still 5 lines available on the Prop-2 for perhaps additional outs/ins or even an audio drive.   NOT BAD! 

This is slightly better than my first scenario of ganging 4 DC-16's together, but with higher current capacity.  The downside is that the first 72 outputs would have to be bulbs and cannot drive relays or solenoids.  But with two of these boards I would have a total of 16 lines for that purpose (not show in the schematic, but common practice with pinball machine wiring is that the protection diode (for back EMF) is directly soldered to the coil/solenoid).

So I think I may go this route.  The board is very convenient in that it uses card edge connectors and those connectors I can hard wire to the Prop-2 and DC-16's

The way I look at it with this new discovery, I think I can do this with 3 Prop-2's.  One for the output matrix (as outlined above), one for the input matrix, and finally one for the score display (which is the brain behind the whole thing...I may need a Prop-SX in this spot...we will see).

[jukingeo]

Hello all,

I been doing a bit of thinking and I decided that I should use an existing pinball machine output driver board.  Many of these boards are directly TTL driven and it would be easier to interface a Prop-2 or DC-16 directly to one of these boards.  So that pretty much would wrap this up in terms of increasing the output capacity of the Prop-2 or DC-16 without resorting to separate components.

One thing, many of these boards do requiring a strobe system for demultiplexing.  While this reduces the number of lines needed for information transfer, I think not even the BS-2 would be fast enough.   Thus I think it may be time to move up to the Prop-SX.  How much faster is this board in comparison to the Prop-2?

Thanx,

Geo

[JonnyMac]

It's difficult to do a direct comparison -- let's just say the Prop-SX is loads faster.  The reason is that you're running code directly on the SX at native speed (50 MHz = 50 MIPS).  The speed gives one ability to slice up the bandwidth to do multiple processes.  For example, I'm sitting here in a hotel overlooking the Las Vegas strip with a Prop-SX connected to an articulated skull filled with servos.  There are two LED spotlights on the mounting plate.  I've written a program that can receive serial data, update eight servos, and update eight PWM outputs -- all in the backgournd.  The foreground code pulls the serial data from a buffer and parses it to move command values to the correct output.  You can find this program in the Library under the VSA section.

There is no way to do this kind of program on the Prop-2; it's fairly trivial for the Prop-SX.

[jukingeo]

It's difficult to do a direct comparison -- let's just say the Prop-SX is loads faster.  The reason is that you're running code directly on the SX at native speed (50 MHz = 50 MIPS).  The speed gives one ability to slice up the bandwidth to do multiple processes.  For example, I'm sitting here in a hotel overlooking the Las Vegas strip with a Prop-SX connected to an articulated skull filled with servos.  There are two LED spotlights on the mounting plate.  I've written a program that can receive serial data, update eight servos, and update eight PWM outputs -- all in the backgournd.  The foreground code pulls the serial data from a buffer and parses it to move command values to the correct output.  You can find this program in the Library under the VSA section.

There is no way to do this kind of program on the Prop-2; it's fairly trivial for the Prop-SX.

So the Prop-SX runs at 50mhz as opposed to how much can the Prop-2 do?

[JonnyMac]

It's not apples and apples.  The Prop-SX runs machine code; the Prop-2 runs an interpreter which has to fetch instructions from an external EEPROM.  I think it's safe to say that the Prop-SX is better than 1000 times faster than the Prop-2; using the available bandwidth wisely one can create true multi-tasking programs.

[livinlowe]

My two cents-
There isn't much differnce between the two, just in execution. The SX is harder to learn, but as Jon said you get much more in the way of POWER  ;D