EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: guenthert on March 24, 2023, 06:05:24 pm
-
The joy over the acquisition of a -hp- 735A didn't last long: after it performed well for two weeks or so, I left it off for a few weeks and upon powering it on again, the dreaded oven light oscillation occurred. :-BROKE
This happened to others too [1]. Working hypothesis is that the glass bead of the NTC busted, presumably due to mechanical stress during cooling. I kind of was hoping that perhaps the insulation accumulated some moisture (we had unusually cold and wet weather in California the past weeks), but a run over night didn't change the symptoms (oscillation of oven lamp between off and near-full on every ~4s), so that's probably not it.
The output voltage oscillated likewise some 1μV around the previously recorded mean, so not too bad. So what are the options now? -hp- considered the diode oven assembly a replaceable part and didn't even bother to list the parts included (the NTC is stated as 3650Ohms at 80°C, but no further information is given). Following came to my mind:
1. Ignore it, do nothing, it ain't that bad. It still can serve as reference to compare different DMMs, even if at reduced resolution.
2. Attempt to replace the NTC. Arguably the Right Thing to do, also in order to verify the hypothesis, but [1] wasn't encouraging.
3. Acquire a diode oven assembly. Well, it's not -hp- most popular product and hasn't been offered since the mid seventies or so, so that might be a challenge. Further, as a replacement will be just as old, chances are it'll die the same way sooner or later. (mine is one of the younger ones)
4. Disable the heater. That would eliminate the oscillation in exchange for - I'm afraid - considerable dependency on ambient temperature.
Option 1 didn't sit well with me, as the 735A was the most stable source I had and one of its application is to serve as comparison with other sources. Broken as it was, it couldn't do that anymore.
I pondered the (fairly simple) schematics for quite a while and an idea formed.
The glass bead around the NTC is meant to hermetically seal it. A side effect however is that it'll retard the sensing of temperature changes. With the bead broken, it'll react faster than before. Only a proportional control scheme is employed in the 735A, which happens to work with the lag and power of the heater and not with the reduced response time. I thought about reducing the heater's power, but didn't follow up on that idea. How about putting on some brakes instead? Slowing the NTC's reaction down?
Fortunately the output of the temperature sensor is exposed (TP2, the base of A3Q2). With an oscilloscope attached I could see some considerable hum of which the amplitude oscillated with aforementioned 4s period. Connecting a .560μF capacitor between TP2 and output low (the troubleshooting guide suggests to measure against that point, but that is some 146Ohms or about 146mV away from circuit and zener common) got rid of the hum, but didn't do anything noticeably to the oscillation. "Viel hilft viel" I thought and attached the largest capacitor I had at hand: 2200μF. Much better, the oscillation slowed way down, but was still observable. Thus encouraged, I substituted that capacitor with a series of the 2200μF capacitor and a 3kOhm resistor (one end attached to circuit common, which is some 3V above the base of A3Q2). Tada! No oscillation observable anymore. Now, did I trade an obvious short oscillation with a not-so-obvious long one? I think not, as - crucially - there is no overshoot anymore. Can it still reach the target temperature or did I overdampen it? A test from (near) cold reached the previously recorded output voltage in reasonable time, so I seem to have gotten it accidentally right.
Now, this is rather a hack than a proper repair, but since it seems to work and a proper repair might not be feasible, I'll call it a success. There is of course the concern about the leakage current (which might change with time and temperature) of the electrolyte capacitor, but since the current through the NTC (~3.3V / 3650Ohms) is some three orders of magnitude larger than the expected leakage current and the temperature isn't that critical (not sure whether -hp- bothered to select a zener current minimizing the TC of the zener, but even if not, it should be only a few ppm/K). More of a worry is the deterioration of the now exposed NTC, but since it operates at a fairly high temperature, I'd think moisture is not too much of a concern.
As it is, my unit performs now after the 'repair'
(http://[attach=1])
much like it did before:
(http://[attach=2])
[1]https://www.eevblog.com/forum/testgear/hp-735a-dc-transfer-standard/msg651816/#msg651816 (https://www.eevblog.com/forum/testgear/hp-735a-dc-transfer-standard/msg651816/#msg651816)