unl2803 probably won't be able to drive incandescents without overheating.

If you're making your own board I'd decide up front whether you can do all LEDs or not. It'll make your board much cheaper if you can avoid the current draw of incandescents

I've never been able to find one. Transistor arrays are hard to find

I'm not super good with electronic design though, maybe I'm missing something. I think at one point I calculated that the unl2803 couldn't handle that much power dissipation going through it though.

There are combined shift registers and led drivers. They're more expensive then just having a shift register and a driver IC, but if you do go with LEDs they can save board space

take a peak at what bally did. the used 4 to 16 decoders to latch SCRs.

Why strobed? Aren't you not doing PWM?

Won't the decoders not work if you're not using SCRs? Latches are nice though. If you are doing any quick lamp manipulation, I'd prefer latches to shift registers (especially chained shift registers) for speed

Now you see why most approaches are with a matrix. The minute you need more than 8 or 16 outputs, if you plan to driver current hungry incandescents, it is most cost effective to have 2X 8 high current drives.

The most sensible design for your 2 wire interfacing would be to use an inexpensive PIC or 8051 core microcontroller to do the background matrix processing. They can be had for under $2.

However, I don't know if you have the programming skills set to create that background Firmware??

Firmware wouldn't be very complicated, but you can't exactly convert a non-matrix game to use a matrix without major requiring, can you?

Additionally, the power requirements of having 64 lamps all on at the same time is also a consideration if directly driving all 64.

#44 lamp uses 0.250 amps.

64 X 0.25 = 16 amps.

The main benefit to direct drive is if ( like others have already done), you make smaller addressable driver boards that can be mounted close to the end user to reduce wiring requirements.

Correct. I wasn't aware if this was a re-work or a brand new project. Sometimes, I miss key words in a thread.

It would therefore almost make more sense to design an interface to re-send the existing driver tech of the System 1 driver board or an aftermarket remake.

Interface the 2 wire comm. To the 4 bit latches of the Gottlieb driver. But I think that was already discussed in your other thread.

Perhaps the LISYS project PCB would be a good approach here instead of re- inventing the wheel?

Okay, so, question: Exactly what brand and part number did you use for the connector (and/or crimp pins) you used for the ground wire(s)?

And the gauge of wire?

Not all crimp connectors are not created equal. All have current limitations.

I am not criticising here, just trying to help on your journey.

And, I would not even consider TIPs at this point in time. MOSFETS are far more efficient and logic level ones are simple to interface to. The main consideration will be how much current the ground return traces (and wires through connectors) will need to handle. You may need to spread the current load over multiple conductors. Just like the Gottlieb engineers did when using multiple ground paths.

I found that 2n7000 blew up under the strain of 44 lamps. The BS170 at 500ma is a much better choice and can be driven directly by TTL levels.

Yes, the Sys1 driver board used 4 bit latches with individual latch pulses for the lamp drives. And completely discrete Solenoid output drives from the MPU. So, you would need some kind of a slave processor in between to repurpose the original driver board.

Is it your plan to hack all the old Gottlieb edge connectors off and replace them with the 0.100" headers? If so, you might want to use 9 or 10 pins (utilize one pin as a key in a different place in each connector) and let each 8 pin group have it's own ground wire back to the power supply. But still common-ize all the grounds together on the driver board so as to spread the ground loading out over all connected wires. Basically what the System 1 Ground mods do.

Yes, the LISY1 is designed to interface with the cabinet wiring of a System 1 game. Sorry about butchering the name. It is a great creation for people that want to dive in to reprogramming their games.

Bally Theory of Operation describes how they do it.

http://arcarc.xmission.com/Pinball/PDF%20Pinball%20Misc/Bally%20Theory%20of%20Operation.pdf

near bottom of page 8

I did a bit of research and if you want to stick with the TO-92 package, the only other cost effective one I could find was the VN2222LL which can handle a bit more current - 0.750A (500ma on a 250ma draw has me a bit worried). But I am not sure how easily it will interface to the logic outputs. I am used to using only Logic Level Mosfets just to make things easier.

Personally, I would switch over to a TO-262 (I2PAK) package as it can still be used as a through hole and it is smaller than a TO-220. It will cost a bit more each (around $35 at 100 pricing) but will be plenty strong as I think the lowest rated one I could find was 5.5 amps.

I am not sure I understand your question.

Total Amperage would only come into play when it concerns the power supply source, and the conductors and connectors that have to pass the total amperage of ALL the lights.

For the Transistor/MOSFET choice, it only depends on each individual load (lamp socket). But for me, while a single lamp may be only 250ma, I am concerned about how the driver semiconductor will hold up if, perhaps, the bulb burns out and the filament shorts out due to vibration. I see this happen when SSR ( Solid State Relays) are used to control 120VAC incandescent bulbs. Even though the SSR is rated highly enough to drive the normal lamp (55-100 Watts), they get destroyed when the bulb is burnt out and someone goes to replace it with the power on. The blown filament, as it dangles, flips around and shorts across the insides and takes out the SSR.

Obviously, the 2N7000 at 300 ma was easily eaten up by a #44 lamp. Of course, since you are dealing with VDC, there really isn't anything wrong with using an MPS-A14 Darlington transistor like Gottlieb did which is also rated at 500 ma. So, I suppose, the BS170 is a good choice too.

As to your Math: A #44 lamp uses 0.250 amps (A #47, uses only 0.150).

32 Lamps X 0.250 amps = 8 amps when all lamps are on (not 5 to 6).

Are any lamps in your game doubled up? I'm pretty sure gottlieb does this sometimes

I made a lamp driver to drive 64 individual RGB ws2803's (im using the fast pinball ones). It has its own mcu + ram that you dump patterns onto and just send like 2 byte messages over serial (rs485) to do like, PLAY PATTERN1, which might be a fan etc.

I'm using individual instead of serial so I dont need to worry about chained LED's breaking down in the chain.

my driver board only sends data to the WS2803, voltage is external so I dont have to worry about current draw through the board itself.
LED driver

Good point. That's what those MPS-U45s were for.

Why not sidestep the whole driver problem, and start using addressable LED's?
The ws2812 leds are RGB leds what are individually programmable. They needs one line to drive a whole chain of 100 or more leds. The protocol is available on the arduino as libraries.
And to save you soldering, they are available pre-soldered with 10cm of wire between them. The presoldered ones are available as small round pcb great for inserts, and as 10mm leds for GI operation.
See here for what i am using: http://3dwingmaster.blogspot.com.au/2015/06/lights.html

And did I mention they are cheap as chips: 50 LED presoldered with wire for usd$10

Got a link on where they sell them?

I feel like that math is wrong in some way. Most games I've seen are rated for 8 amps and I have yet to see that (maybe if everything is running) for example this is the case with both my bk2k and trident and surely they both have more than 100 incandescent bulbs each and both can be running full blast without tripping a 20 amp breaker... not sure if I missed something hahaha

Most of that is GI, which is usually on a separate 15-20A fuse. And frequently melts connectors anyway

Yup, you missed a thing called a matrix!

With a 64 lamp matrix, the maximum bulbs being driven at any instant in time is 8. They are pulsed at a rate of 1 or 2 ms every 8 Ms at 18vdc.

And, as was mentioned, GI is on a different circuit with up to 15 to 20 amps of constant current.

Also, a game like a Classic Bally with direct drive feature lamps, if you leave it in lamp test (all lamps on, all lamps off) it can burn up the bridge.

It was accepted at the time, that not all lamps would ever be turned on at the same time (10 bonus lamps, only one at a time etc.). Some lights start lit. Getting the rollover or Target extinguishes lamp and lights another one.

Attract modes can be hard on drivers and power supplies but again, there is a lot of sharing of on and off times between lamps.

Yeah, I'm running something like 80 44 bulbs in my custom game - all GI is controlled - and that giant bridge gets too hot to touch even with the giant heatsink I have on it when I run it in test mode!

The BS170 has been perfect for me, haven't blown one since I switched from the 2N7000.

This was it, derp. hahahha

For inserts:
https://www.aliexpress.com/item/50x-Pre-soldered-WS2812B-LED-Heatsink-5V-5050-RGB-WS2811-IC-Built-in-12cm-Wire/1941066393.html
or
https://www.aliexpress.com/item/Wholesale-DC5V-WS2812B-10mm-3mm-rgb-led-strip-5050-individually-addressable-led-pixel-module-50-pixels/1969164799.htm

GI:
https://www.aliexpress.com/item/50pcs-Non-waterproof-12mm-WS2811-2811-Pixel-node-Module-Light-String-Addressable-LED-Module-DC5V/1547858503.html

Since you are living in a world of DC when using the MOSFETs (not 6.3vac), you could always switch to a high wattage switching power supply. You could even split the load and run two less costly units to do the job.

I just used the existing EM transformer and made my own power supply. Works fine, saved money.

I'm waiting until I finish to do the write up, but I'm running Open Pinball Project boards and wrote the documentation for them on Pinball Makers. Most of the photos there are my machine.

http://pinballmakers.com/wiki/index.php/OPP

Yeah, they usually ship same day. The $8 flat rate shipping blows though so I try to build up an order.

https://pinside.com/pinball/forum/topic/gottlieb-hot-rod-a-tribute-to-classic-em-pinball

Yes, I switched from OPP since I was having a lot of serial issues with it. P-ROC working much better.