Redesign of an obscure tester - design critique

[metertech58761]

So, when the lockdowns began rolling across the country last year (just as I was set to head down to a collectors' swap meet and spend a nice spring weekend at my friend's acreage in the Ozarks), I ended up dusting off my long-dormant electronics hobby.

I had stumbled across a 'field configuration terminal' on the E place that I could use to play around with some load management devices I'd picked up at some point - it's complete and works, but is missing the power / interface cable (I checked and rechecked the wiring several times before finalizing the construction of the temporary plug I have for it).

I'd planned to work with the original unit as-is, but after reverse engineering the all the boards and recreating the schematics, I realized the internal wiring of the original unit is just plain dangerous (the DC ground has a healthy dose of AC riding on it, even when when the main switch is in its OFF position)

I'm sure my USB logic analyzer took a fatal jolt at one point, as it doesn't seem to work properly since.

So... I decided to see if I could replicate this unit, with three design goals:

1) replace obsolete / inefficient parts with modern equivalents
2) eliminate that $%&?!@#?!! hot-chassis wiring
3) replace the transmit stage with one that works at a lower, SAFER, DC voltage

So, I'll present the boards as they are currently, explaining what I started with and my line of thinking towards my present design.

First up - the all-important power supply.

Pretty straightforward... I started with a transformer-type supply that, as I said, somehow introduces an unhealthy amount of AC to the DC side (how the unit even works like that, I don't know!!!).

The 5V supply used a LM340K-5, and the two 12V supplies (one for the power / display side, the other for the analog side) were nothing more than bulk supplies with discrete bridge rectifier and filter caps.

My sticker shock over how much the LM340K-5 / LM309K commands today led me to look for alternates, and I found some attractively priced DC-DC modules, and which in turn led to the discovery of and the complete replacement of the power supply circuit with three AC/DC SMPS modules (+5, +12, and +24).

The capacitors on the output were chosen based on the data sheet and are near the output terminals, and I did test this board and I have good, steady outputs on all three supplies.

The second jack on the AC side goes to the Analog board for signal coupling, and I'll get to the Analog board a bit later.

[mariush]

The device may need clean DC output of a linear regulator to work properly,  switching regulators add noise.

I'd probably go with a toroidal transformer that has two 12v secondaries and use a couple linear regulators to regulate the rectified 12v  (which should be 12v x 1.414 - 1.5v = 15.5-16v peak) ... a regulator like let's say LM1084 (max 5A) or LM1085 (max 3A) in a to-220 or TO-252/DPAK should be fine.  You can get adjustable or fixed 12v / 5v versions of it.

For the 5v, you could probably use a switching regulator to get 5v from one of the 12v windings and do some pi filtering on the output for cleaner output.  BUT, you should first check how much current it's used on 5v ... if it's something like 50-100mA then there's no point, just use a linear regulator ... 12v to 5v at 100mA is basically 0.7w of dissipated power, something that can be handled with a regular to-220 heatsink or just dissipated in circuit board.
 

[metertech58761]

I will check the amount of ripple generated by the supplies, but the data sheets for these modules claim 100 or 120 mV p-p noise / ripple on the output.

Perhaps I was a bit hasty going all-in on these modules.

[metertech58761]

The next board I redid was the display, and of the various boards, this was the most straightforward.

Other than relocation of the 8279 keyboard / display driver to another board, most of what I changed involved replacing obsolete / hard to find parts with better ones.

(MAN74 display -> HDSP-7803, 14547 -> 74LS47, UDN2981-> TBD62783A).

The other changes were to consolidate the discrete red / green LED pairs into single LEDs and swapping three partially utilized ULN2003As for two ULN2803As.

The keyboard alluded to is a simple 4x4 matrix unit driven by the scan line outputs from the 8279 and read row by row from the same IC.

Likewise, the LED cathodes are driven in turn by the scan line outputs from the 8279 (through the 14514 4-to-16 decoder and the ULN2803s).

The LED anodes are driven by one of the 8279's display ports through the 74LS47 and the TBD62783.

The discrete LEDs are driven from the A and G segments, by writing a 4 (red / fail), 7 (green / pass), 8 (lamp test / active), or F (off) to those locations respectively.

[metertech58761]

The interface board - this one had some significant changes.

The first was moving the 8279 over from the display board, as I didn't like how far the data lines were being stretched (despite the buffers on the incoming lines from the Logic board).

The next was swapping the old NVRAM ICs (HM-6561 x 2) for a 6116 SRAM.

The last was swapping a 74LS375 that only carried the clock signal to the 8279 with the otherwise unused section of the 74LS241. I can upload a snippet of how the '375 was wired.

[metertech58761]

Logic board - the only changes were cleaning off some unused options to more easily port the design to a new board, and replacing the comparator-based Init Enable circuit with an opto-based one.

(one more after this)

[metertech58761]

And the last board in the batch - the analog board.

There were two major changes:

One was replacing the original comparator-based watchdog with a MAX 706

The other, more important, change was swapping the original (and dangerous) transmit circuit for a lower-voltage design off a similar, but later module.