Xytronic 988D desoldering unit not working

[jrseattle]

Hi all,

I have a Xytronix 988D solder/desolder machine but the desoldering side is malfunctioning. It starts ok but then suddenly the temperature display (set to READ) shows an abnormal high temperature (>600 C) and, as a result, the desoldering head is not heated. After 10 to 20 seconds it goes back to normal (and starts heating) for a few seconds before showing the high temperature again. This behavior oscillates continuously and the head never gets hot enough to use.

It turns out that exactly the same thing happened after I purchased the unit in 2005 and Xytronic USA repaired it under warranty. I sent in the desoldering handle first but was told I had to send in the main unit for repair; the DIA60A desoldering handle was not the problem; Since it happened again, this may be a more common problem with these units.

I have been troubleshooting the PCB but can't find the fault. Attached are the schematics for this unit. The PCB at issue is labelled 66-213118. The DIA60A handle uses a thermocouple to send temperature information to the 66-213118 PCB (see top left “SENSOR”). It generates a voltage ranging from 4 mV to 21 mV. I've tested this separately from the main unit (using an external 18V AC transformer to heat the handle) and it works fine. On top of that, I have tested 3 different heater/thermocouple units and they all show the same behavior. I've tested the power supply voltages on the PCB (+8V, -8V, +5V used by the display); all are ok all the time.

This PCB has several ground levels which is a little confusing. There is chassis GND and PCB GND. The negative of the thermocouple is connected to the GND for the +8 and -8 V supplies but also to the 24V AC heater circuit on one side. The chassis GND goes to the desoldering head where it is more or less connected to the thermocouple (5R resistance). The separate display PCB has its own floating ground. A scope shows no odd signals on the PCB ground line when the cycling happens: it stays at close to 0V (this GND measures 7 mV from chassis GND and changes a little when the DIA60A is heated)

The + side of the thermocouple is connected to OpAmp LM358 pin 5 (plus), conditioned with several capacitors, diodes and a pull-up resistor to +8V. What I'm seeing is that the 5-20 mV signal from the positive side of the thermocouple suddenly goes to >400 mV for a while, usually when the heater circuit is de-energized (but not always) then it goes down again to 10-20 mV (takes about 20 seconds) and continues cycling (heat on, heat off (usually) -> jump to >400 mV, slowly goes back to 10-20 mV, heat on etc.). The read temperature display goes to >600 degrees Celcius with the jump to >400 mV happens, which is not normal, so this form of cycling is not normal, and the solder head doesn't get hot enough to be useful.

I tested all of the components connected to the thermocouple signal, replaced the capacitors, diodes, 1M resistor and OpAmp, but the problem doesn't go away. The supply voltages +8V and -8V stay stable during the cycling. This behavior continues, even if I remove the OpAmp (by manually cycling the heater by pulling the base of Q1 High). This cycling also happens when I disconnect the 'E' line (which goes to the display). All connections have been checked and double checked.

When the thermocouple is disconnected from the PCB, the input to the OpAmp pin 5 is always 400 mV, presumably because no current is going into the thermocouple, so there is no voltage drop across R4 (1M to +8V) so the OpAmp pin 5 line is now at +8V and diode D4 start conducting, leaving a forward voltage drop of about 400 mV.

I measured the current going from the + side of the thermocouple to the PCB at -7.9 uA (+ connected to thermocouple, - connected to circuit), so current is going into the thermocouple. At around 400 C (this value is not necessarily accurate and it changes anyway), the current drops to 0 and the voltage on the PCB jumps to 400 mV. After cooling down, the current resumes etc. The actual current value may be inaccurate due to the meter I'm using but the fact that it drops to zero is clear. I also tested the voltage across the thermocouple in the desoldering handle while using a patch cable to be able to disconnect the + side of the thermocouple to the PCB. When the unit cycled (showing +400 mV across the thermocouple, I disconnected the patch cable and the thermocouple quickly goes back to the 10-20 mV range while the PCB circuit shows 400 mV . This tells me it's not the thermocouple suddenly shutting down.

I am at a loss why the thermocouple voltage suddenly jumps to >400 mV. Since the GND level doesn't change (as far as I can measure) and this also happens when the OpAmp is removed, the voltage must come from the 1 M pull-up resistor to 8V . I measured the current going through this resistor at 7.8 uA normally, 7.3 uA when the thermocouple voltage jumps, implying that the +8V doesn't change. I replaced the 1M resistor R4 but there is no difference. I also replaced C11, C12, D4, D5.

I tested Q1 (2SC945) separately and it works properly. I also disconnected the "E" line to the display PCB with no change.

What else can I test?

Jrseattle

[Kean]

If you are measuring >400mV at the input to the opamp, then that will be coming from the 8V via R4 pullup and clamped by D4/D4.  That implies the thermocouple input is going open circuit.  If the thermocouple input wasn't open circuit you should only see something like 22mV max (depending on thermocouple type and temperature).

If you are reasonably certain that the thermocouple is OK (seems like it as you tried multiple heater units), then I would concentrate on the wiring and connections between the desoldering handpiece input connector and the PCB traces.  Maybe a cold solder joint at the PCB.

[jrseattle]

Thanks Kean for your reply. I agree completely that the thermocouple must be the problem but had multiple reasons to believe that it was ok: 3 different units with the exact same behavior and the earlier repair requiring the main 988D unit to be sent in.

However I now have proof that the main 998D unit is working just fine: I found a HAP60 hot air pencil in the original box. I didn't remember that this pencil came with the system and never used it. Since it replaces the DIA60 desoldering handle (same connector) it is a good test for the 988D unit. I plugged it in and the system now works perfectly, showing consistent temperature readings and heating as necessary. This proves that the problem is with the DIA60 desoldering handle. This handle houses the heater/thermocouple unit plus plumbing for the sucked up solder, nothing else.

Attached is a picture of the heater/thermocouple unit. It is connected to the 988D with 7 wires: 1 GND, 2 for the vacuum switch (SW 4), 2 for the thermocouple ('S'), 2 for the heater ('H'). The 'S' connections go to the black and white wires near the top, to the thermocouple, the 'H' wires go to the 2 read wires on the top, the heater. The heater/thermocouple itself is sealed with some kind of cement but is really small. I don't see anything besides traces on this PCB (ignoring the vacuum switch).

I have been reading about thermocouple circuits and they require a reference junction (generating a compensating voltage) which must somehow compensate for the thermocouple voltage. I don't see such a reference junction anywhere in the 988D design but it looks like the system compensates by subtracting the reference junction voltage at room temperature from the thermocouple voltage. SVR1 sets the baseline voltage if the thermocouple pins are shorted. See https://instrumentationtools.com/problems-of-reference-junction-compensation for a more detailed description.

None of this explains with three different heater/thermocouples (2 brand new) go open circuit when the unit heats up. Could it be the long shelf life? This particular design was discontinued 15 years ago (about the time I bought it for a lot of money) so the parts are probably just as old.

By the way, I checked and double checked all soldering joints, everywhere.

Thanks for any other suggestions

jrseattle

[Kean]

Yeah, I don't see any proper cold juntion compensation in the schematic.  Not a huge deal when measuring soldering temperatures, it just won't compensate properly for any large ambient temperature changes.  Be aware the readout won't be very accurate, and lots of soldering stations can be of by +/-20°C in my experience.

I would hazard a guess that the construction of the thermocouples in the heater units are just poor quality.  As the thermocouple depends on a pair of dissimilar metal wires being welded at the sensing tip, if that is put under stress (i.e. cycled heating & cooling) and the welding was not performed correctly, then you'll get a (potentially intermittant) open circuit.  Could be one of the reasons it was discontinued!

[jrseattle]

Thanks again Kean for your assistance. I also want to thank Howard Electronic Instruments (New York) for assisting me in this frustrating experience. They provided excellent customer service by sending a replacement thermocouple/heater unit (when the first one I purchased failed to work) and when that again didn't work (and we conclusively decided the main 988D unit was not the cause of the problem) sent me a replacement desoldering handle DIA90A to test. That new handle works perfectly! I was able to desolder a 40 pin IC from a PCB easily. I'll be paying them for the price difference. Thanks again, Howard Electronic Instruments.

On the other hand, I'm not sure what is going on with Xytronic from Taiwan. They refused to provide any customer service/advice (both US and Taiwan office) and their warranty service in the US closed down. That really is too bad because their products are quite nice.

jrseattle

[Kean]

Glad to hear you have it sorted out.