# The Human Condition

### Site Tools

wiki:user:ram:vacuum:mks_903ap

# MKS 903ap cold cathode gauge

( manual)

This is a cold cathode (inverse magnetron) ion gauge for measuring in the range $3 \times 10^{-10}$ to $5 \times 10^{-3}$ Torr. I got mine on Ebay, and it was DOA. It was drawing way too much current and otherwise showing no signs of life. So, well, I could have returned it, but what would be the fun in that?

This is the insides (with the vacuum sensor cell itself already removed).

The manual does describe how to remove the vacuum sensor and how to clean it. One of the things that led me to choose a cold cathode gauge is that it is pretty rugged. Nothing terrible happens right away if you power it up at atmospheric pressure, but running at higher pressures eventually causes buildup inside the sensor. The manual says you can clean the sensor internals with a Scotch-Brite abrasive pad!

Unfortunately there is no schematic, and also the board is multi-layer so there is very little of the wiring that you can reconstruct visually. There were no obvious scorched components, though the high voltage transformer did have some brown discoloration in the potting. There are two circuit boards, the main one, and one on the side which has the connector and functions related to the “relay”. This is all through-hole, with an early 90's manufacture date, probably late 80's design.

One of the weird things, before I even started diving into the circuit, was that there was no relay anywhere, even though the gauge has a relay feature. I really began to doubt my sanity when I began to use continuity testing to trace from the (documented) connector pinout to the innards. I found things such as the ground pin was connected to the (positive) input of a LM317 regulator. It took me a while to realize that reverse engineering something that was busted is extra difficult. The ground pin was connected to the regulator input because the positive and negative power inputs were more or less shorted. And the pins that did not connect to any pad on the circuit board, or went missing at some point, with a node not connected to any other device? It must be that the internal traces were blown open by an overload.

This is a rough picture of the functional groups on the top board. Inverter and aux is where the power comes in. There is a Unitrode switching regulator chip driving a small power transistor (hard to see inside the HV shield), and an LM317. Log amp is the main input signal conditioning. I didn't dig into this at all, but obviously it has to be a log current-to-voltage converter. High voltage is the transformer, rectifier, filter capacitor, and feedback voltage divider and current limit resistor. Split aux supply is driven off of another winding on the transformer, driving the log amp and linearization blocks. Linearization conditions the log amp output, generating the final output signal. There are various resistors, diodes and trims implementing some sort of analog lookup table. Something is slightly out of whack on mine because there is at least one place where there is a step in the output. The red lead is HV to the sensor, and the black lead is the current sense from the sensor.

It became clear that the side board had the implementation of some approximation of a relay, as well as the setpoint comparator (U8) and various protection filter and protection components (CR11, C35-38, etc). Whatever was supposed to be going on on the side board, the core feature of actually sensing pressure was on the top board, and had to come over those two 6-pin right angle headers that tie the two boards together. In particular, we need power, ground and signal out. Also, some of these pins were obviously doubled up (connected to copper pours).

What I ended up doing was getting rid of the side circuit board. I replaced this with handwired surge protection on the supply and output lines. Insofar as it actually existed, the side board also contained the high voltage power enable feature. As far as I can tell this was kind of a misnomer. When you grounded the high voltage enable pin, you were just grounding the power return for the top board. The input power was always connected on the high side. Then the top board was powered. This generated the split aux supply, part of which was routed to the side board. I'm not sure why they through the song-and-dance of having a “power” light and a distinct “high voltage” light. I'm thinking maybe it was just that this is what the user interface of a high-vacuum gauge is supposed to be? Anyway, if you apply power to the top board, it powers up. So I implemented offboard power switching in my process controller interface electronics. This is what the result looks like:

I do have some scribbled partial schematics, if anyone is interested (but nothing like a full schematic).