Industrial timer/temp. controller teardown (Ametek TMM-0999M-461, Omega CNi1653)

[D Straney]

Here's part of a batch of a few industrial temperature PID controllers or timers that I ran across at work and took apart on a slow day.  The electronics in most of these usually aren't all that interesting compared to the test equipment or high-power supplies I normally get to see, but these two had some mechanical-construction details that were worth a post.

Ametek timer
On the outside, this is just a multi-function (one-shot, free-running, etc.) timer, with some switches to set the function and time interval.

Like many other industrial controls, though, it fits into a relay socket, which makes it easier to swap out individual units:

Those 4 screws at the corners are all that holds on the cover: let's pull that off, prop it up on a conveniently-at-hand ferrite toroid, and take a look inside:



Well, that's some interesting construction.  They sure had to Tetris things a bit to fit them in there.  It looks like the relay socket is a discrete piece with flying leads, which are soldered to the 3 boards.  On top, there's a control board, with a Freescale microcontroller and the switches:

With all the dual diodes ("CR*") in the top-right corner, there may be either some diode-ROM-style binary encoding of the rotary switch state going on (I think thumbwheel switches normally have a binary output already?), or muxing of the many switches, either way meant to get away with a low-GPIO-count microcontroller.

The control board is propped up on one side with a metal bracket.  On one of the other sides is what seems like an AC input board:

It's got an SMT MOV (the bright blue thing), a bridge rectifier, a 400V MOSFET (IRFR310), and an MOC256 AC-input optocoupler.  3 connections (MOSFET gate, optocoupler outputs) go directly to the control board.  My vague guess is that the big 33K resistors (R14, R15) and optocoupler are used to sense zero-crossings in the AC line, for either timing reference or turning on the relay, while the MOSFET is used as a linear regulator for the control power or switching the relay (if it has a high-voltage coil).

Finally, there's the relay board, which doesn't have much besides a bunch of wires coming in, a couple power resistors, and the output relay itself:



Omega CNi1653 temperature controller
This one's a little more complicated, as it has things like temperature sensing (with many types of sensor supported, see page 6), PID control, and multiple outputs (this model has 0-10V, and a relay).

The relay's busted on this one, which is part of why I got to take it apart.

Inside, there's an interesting multi-part construction, with a 3-board sandwich:



This sandwich plugs into the rear of the back-panel screw terminals at one end, and into the front-panel board (with display and buttons) at the other end:

There's nothing interesting going on with the front panel board, so let's move on.

The "PCB sandwich" consists of a main board (bottom), power board (top right), and output board (top left).


Main board has a microcontroller on the other side (or so I assume, there's a version sticker on it) and some misc. passives.

There's also a custom "Newport"-labeled IC (subsidiary of Omega, from what I can tell with silkscreen labels and online) which is probably the analog front-end and ADC: I can imagine it being difficult to support variable-gain and cold-junction compensation for multiple thermocouple types + RTDs + voltage/current sensing, in such a small form factor without doing something custom (or adding a couple more boards to the vertical stack).  There's also a TLC2201 precision CMOS op-amp, 74HC04 hex inverter, and an LM2931 LDO.

The other side of the main board has a temperature sensor glued to the terminals (at top-right) for thermocouple cold-junction compensation, as well as...
* TLC27M2 dual precision op-amp
* REF192 2.5V voltage reference
* LM311 comparator
* 74HC4051 8:1 analog switch
* 93C66 serial EEPROM (prob. saved settings or calibration data for the microcontroller)
* An oscillator crystal for the microcontroller
* Unpopulated optocoupler footprints, maybe for digital input options?
* What seems to be a 220nF SMT film capacitor (the white rectangle), probably for something to do with the analog front-end (sample-and-hold? dual-slope ADC?)


The power board is pretty standard except for its weird shape: seems to be a simple isolated flyback supply based on a Power Integrations all-in-one offline chip.  You can see the EMI filter/MOV, bridge rectifier, and bulk cap (covered with red insulating sleeve) at left, with the Power Integrations chip and feedback optocoupler in the middle, and the transformer at and output caps at the right (cross-isolation Y caps are at the top, 2 in series).


The female header shows where it plugs into the main board.

The output board is where things get a bit weird though:


Ok, looks like the same PCB gets used for all the output options, and the relay at top and its coil-driver transistor on the opposite site seem pretty straightforward, but what's that weirdness below it for the 0-10V output?

Huh.  It's two daughterboards, soldered to each other at right angles, with two connections between them (the right-angle solder joints at each end) plus maybe some on the opposite faces at the middle overlapping parts.  There's two LM358 op-amps, plus a shitload of passives.  Both of these boards have passives on the opposite faces as well (although that was impossible to photograph well): the layout people must've really run out of room on this one!  Interesting 3D assembly style there that smells a bit of desperation.