EEVblog® Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: wpietri on September 30, 2026, 11:43:07 pm
-
I have a pretty unusual problem that I'd like to solve with mostly cheap off-the-shelf gear and I'm looking for advice.
The context: A pinball museum with a lot of volunteers of varying skill levels trying to repair electrical/electronic gear across nearly a century of technology.
The problem: For a given failure, like a flipper not working, there is a chain of possible causes. An expert repair person has a good mental model of that chain and knows how to test at various points to isolate the problem. But it's hard for our volunteers to build that mental model up by diving in with a meter, especially for intermittent problems.
The experience I want: Someone can take a device to a misbehaving machine, hook up probes to a number of points (which can range from TTL-level signals up to say 75VDC and 120VAC, with spikes caused by the fat solenoid coils), and see a reasonable graph of what each thing is doing with something like 10 Hz resolution. Then they can trigger various pinball machine behaviors (turn on the machine, start a game, press the flipper button) and watch for correlations. I'd like this to be useful both for mid-level techs on their own and for me to give in-class demonstrations.
As an example, I was just trying to run down an intermittent flipper problem on an 1992 Williams Addams Family. the causal chain runs from the flipper button to the CPU to some TTL-level logic to a TIP102 transistor that connects 75 VDC to ground and then to a couple sets of windings on the flipper coil, and also involving another physical switch called the end-of-stroke switch. Plus assorted connectors and wiring runs, any of which could have gotten flaky. I personally can figure this out by getting in their with a good multimeter and checking various points while somebody else presses the flipper button a lot. But it's hard even for whomever's assisting me to learn what's going on.
There are a lot of cool and relatively inexpensive ways to have multichannel probing for TTL or low-voltage DC stuff. But early pinball was relatively high voltage AC (24-120), and once they started converting to solid-state tech, it becomes a mishmash of voltages. Even in a machine working correctly it's hard to know what a particular point will have on it; given decades of wear and tear fixed by random technicians, you just have to assume any point could be anything from 0 to line current.
My ideal here would be some sort of 4-8 channel box that can be safely hooked up to those voltages and provide 10Hz sampling so that i can connect it to a cheap laptop and write my own software. A decent alternative is something somewhere between a high-end handheld multimeter and a low-end oscilloscope, but with more channels and that's pretty robust. The closest I can figure out so far is getting a few OWON-grade handheld multimeter/oscilloscope devices and mounting them all on a board. Which I will do if I have to, but I'm hoping you folks have some better ideas. Thanks in advance!
-
... hook up probes to a number of points (which can range from TTL-level signals up to say 75VDC and 120VAC, with spikes caused by the fat solenoid coils), and see a reasonable graph of what each thing is doing with something like 10 Hz resolution. ..
.. early pinball was relatively high voltage AC (24-120), and once they started converting to solid-state tech, it becomes a mishmash of voltages. Even in a machine working correctly it's hard to know what a particular point will have on it; given decades of wear and tear fixed by random technicians, you just have to assume any point could be anything from 0 to line current.
.. My ideal here would be some sort of 4-8 channel box that can be safely hooked up to those voltages and provide 10Hz sampling so that i can connect it to a cheap laptop and write my own software.
That's quite a range.
It might be do-able with a simple chain of parts :
Diode Bridge to accept AC/DC - cheapest eg https://www.lcsc.com/product-detail/C42433400.html (https://www.lcsc.com/product-detail/C42433400.html) 200mA 1kV
Current limiter appx 0.5mA - 600V Depletion mode mosfet with current-set source resistor eg ARKmicro DMZ6001E with ~1k Rgs
Decent Optocoupler, spec'd for 0.5mA operation eg TOSHIBA_TLP293-BLL
That combination would tolerate 120VAC but still work at 12VDC or below.
The output is isolated OC, so easy to connect to a 16 channel logic capture module, or a 16 chan ADC module.
The TLP293-BLL is tightly gain-banded CTR, 200~400, so you could linear calibrate each probe with a trim of the collector resistor.
Somewhere just under 3V would be the analog sense baseline.
Addit: Since your scan rate is low, an alternate approach would be to get a standard Aliexpress 8 channel ADC module, and modify the input pad area for 200V+ max span.
ADS8688 modules look to have 16 resistors, but that part has a fixed 1M +/-15% input impedance.
Some ADS1256 modules also have 16 resistors, but that part has slower scan, and more samples needed for settling at max sampling speed, but it should still be above 10sps
ADS131M08 modules have 8ch simultaneous sampling and more input-pad choices, for a little higher price.
-
8x isolated channels to prevent a brain fart from extending a bad day handshake.
-
Why not just a big resistive voltage divider. Something like 100K and 1K resistor. So the line voltage signals would be well within the data acquisition system input range. And 5V inputs would still be 50 mV, still plenty to see with appropriate gain settings on the channel. Two resistors per channel isn't too tedious to assemble and also not expensive, and the loading should be light enough that it would change the pinball side of the problem.
-
For a ready made instrument, a low cost 4 channel scope is likely the best option. More channels may be good, but hardly available as low cost.
Ease of use can make up for fewer channels. To avoid beginner typical ground loops have only 1 probe with ground connector. This should be OK for relatively large signals.
There may be multi-channel ADC cards for a PC or via USB. They may need added dividers and protection. The software support could be an issue.
For a DIY solution, one would likely still want sampling fast enough to resolve normal mains frequency and also 120 Hz ripple at least. Quite some issues can be from ripple and half wave signals. This would want something like 1 KSPS sampling rate. The final display could still be slower. So no need for extra retifier or AC/DC converters. It should be simple for the softwas to show min and max for slower scales.
At the low end one might get away with a µC internal ADC (e.g. 12 bit) and suitable switchable dividers to adapt. Higher resolution ADC (e.g. the ADS131M0x) got surprisingly cheap and just software zoom can be a real alternative. So a better ADC, but little or no range switching.
At the low end one can likely still get away with resistors in the protection. A divider at the input would already provide this. Those depl. FETs are a thing for super low noise for really small signals.
I would consider a limited number of channels (e.g. 4) and if needed use 2 identical units to also allow 2 isolated parts for the rate cases one has 2 grounds. One would want isolation from the PC (maybe raspberry ?) ground anyway.
-
Yokogawa made some oddball modular 8-channel scope / data recorders with high voltage isolated inputs that mostly meet your description except for the "cheap" part.
-
It is quite easy to write a bit of software for a microcontroller (with built in ADC) that samples a bunch of channels and sends them to a PC (Either serial with serial to USB converter or USB directly). there are lots of uC's with 10 or even 12 bit resolution, and it's also easy (and low cost) to make a frontend (some voltage dividers, opamp buffers) for such a system.
PC software to visualize the results (numbers, graph, data logging?) is more involved, but these newly fangled AI things can probably do 90% of such an application in a few hours of massaging them.
And there are probably already such (open sourced) applications "out there".
Edit / Addition:
Starting @ 14:02 "Applied Science" describes his circuit of a "Teensy" with an AI designed "web app" that draws 6 graphs at once. The rest of the video is also pretty interesting.
https://www.youtube.com/watch?v=l0qqUE1pgXU (https://www.youtube.com/watch?v=l0qqUE1pgXU)
-
For a ready made instrument, a low cost 4 channel scope is likely the best option. More channels may be good, but hardly available as low cost.
Ease of use can make up for fewer channels. To avoid beginner typical ground loops have only 1 probe with ground connector. This should be OK for relatively large signals.
There may be multi-channel ADC cards for a PC or via USB. They may need added dividers and protection. The software support could be an issue.
For a similar application, we used a Hantek 1008 [1] some years ago (for university students). With a single ground connector and some voltage dividers, this might also be a possible solution here. The protocol on the USB interface has been reverse engineered [2], so even vibe coding a custom software should be not that difficult.
[1] https://sigrok.org/wiki/Hantek_1008C
[2] https://github.com/mfg92/hantek1008py
-
Since how easy novice pinball repairers experience your solution is all in the software.
Eg: you can probably vibe code some python frontend for an DAQ970A or similar to do completely independent multichannel measurements. Closest to "multichannel multimerer" you can get. But 10hz on each channel is fast for relays.
Or what I can see working is a custom frontend box for a Saleae logic analyzer, hard to beat that on simplicity.
Safest off-the-shelf way is one of those cheap 4-channel scopes and some diff probes. Then you can't blow up your scope, but you can still measure TTL up to 120vac. Some even have in-probe RMS to DC conversion (eg: hioki) to get a nice straight line when "120v present".
-
I have a pretty unusual problem that I'd like to solve with mostly cheap off-the-shelf gear and I'm looking for advice.
My ideal here would be some sort of 4-8 channel box that can be safely hooked up to those voltages and provide 10Hz sampling so that i can connect it to a cheap laptop and write my own software. A decent alternative is something somewhere between a high-end handheld multimeter and a low-end oscilloscope, but with more channels and that's pretty robust. The closest I can figure out so far is getting a few OWON-grade handheld multimeter/oscilloscope devices and mounting them all on a board. Which I will do if I have to, but I'm hoping you folks have some better ideas. Thanks in advance!
https://www.youtube.com/watch?v=2WgXzfIqftc (https://www.youtube.com/watch?v=2WgXzfIqftc)
Since you mentioned a "laptop," standard DAQ cards are out of the question. However, there are cards that work with laptops. Unfortunately, most available ones are older models compatible only with 32-bit Windows 7 drivers. You can buy newer versions that work with 64-bit systems, but they are quite expensive. Look for this type on eBay; if you add input dividers, it’ll run up to 1 MHz. It offers 8 channels, expandable to 32. You can look up the rest yourself if you're interested. If you only find them for over $500 on eBay, let me know—I have one for sale that’s been sitting unused for a few years.
-
As an example, I was just trying to run down an intermittent flipper problem on an 1992 Williams Addams Family. the causal chain runs from the flipper button to the CPU to some TTL-level logic to a TIP102 transistor that connects 75 VDC to ground and then to a couple sets of windings on the flipper coil, and also involving another physical switch called the end-of-stroke switch. Plus assorted connectors and wiring runs, any of which could have gotten flaky. I personally can figure this out by getting in their with a good multimeter and checking various points while somebody else presses the flipper button a lot. But it's hard even for whomever's assisting me to learn what's going on.
Remembering from way back in the eighties when repairing pinball machines was my job I would argue that it is far easier to educate your volunteers (which makes them more happy) and buy them a good DVM and ditto headlamp. Nine out of ten problems with these machines can be diagnosed with a good look (wire or solder fail, molten coil insert if plastic, blackened or broken end-of-stroke switch, connector fried on the PCB) and few simple measurements (TIP102 shorted, flyback diode fried). The tenth problem has to do with the hardware (not) driving it all and needs more diagnostic skills than a black box would offer. Just my 2c