"Wild_Bill" wrote in message
...
A friend asked me to attempt to find a fault in a short-lived dehumidifier
that functioned well for 1 season, but failed to operate after being unused
until needed again.

The sensor for the relative humidity is a thin ceramic square with
interlaced finger patterns deposited on the ceramic (looks like a
VCR/camcorder dew sensor). This sensor is placed inside the front grille
in path of the incoming air flow (but inside a small vented plastic box).
As the humidity rises, the very high resistance decreases.

Searching the model number revealed that many owners have had the same
experience with this model, that the unit worked great for 1 season, then
no worky.

This Goldstar DH305 is a normal looking small dehumidifier with no digital
readout or other controls that would give the impression that it's a
complicated appliance.. only an Off/1-9/Max pot (no integral switch) and a
high-low fan speed rocker switch.

BTW, the wrap-around metal cover needs to come off to access the
controller board.

When turned on, the fan runs for maybe a couple minutes, then stops. The
unit will not do anything else unless turned off then back on, after which
the same action repeats.

The capture container float switch isn't a problem.. it works normally and
turns on the red panel LED to show that the bucket needs emptied,
indicating that the unit isn't going to operate until it's emptied.

As stated earlier, the Off/On is just a pot, and the varying resistance
turns the controller circuit on, and the 1-to-Max setting will determine
the amount of run time.

The controller circuit looks fairly simple until the solder side of the
board is seen.. yep, another example of unnecessarily complicated design,
IMO.

The underside of the board has about 30% of the area populated with
surface mount components, including a 87C809 8-bit microcontroller.

After determining that the pot, switch and other sensors are apparently
working (PTC, NTC components).. and the ciruit board soldering looks
reasonable, I forced the unit into operation by huffing moist breath onto
the humidity sensor, the way one does the Haaaaah on their eyeglass lenses
to clean them.

So, the unit is capable of operating normally, and I checked that the
controller relay, not the temperature protector, was cutting off the
compressor.
This isn't an expensive appliance (maybe $140 new), but since it's still
functional, I'll attempt to return it to "fairly normal" operation (gold
star, better than it was/is now).

The voltage across the humidity sensor is approximately 4V AC when dry
(well, that's relative to the ambient humidity), and the unit goes into
operation when the moisture causes the voltage to drop to approx 0.085V
(85mV), which may not be accurate if there is a delay involved for some
(unknown to me) reason.

I'm contemplating cleaning the sensor a couple of times (although it
appears to be perfectly clean), checking the cicuit for a resistor that's
drifted high in value.. or adding a parallel resistor of maybe 2-5 Megohm
across the sensor (to get the voltage level down to a reasonable range.

Information I have, regarding cleaning and calibration of humidity sensors
cautions against using certain chemicals such as ammonia, alcohol,
formaldehyde.. and suggests only using distilled water and drying in a
clean area (or perchloroethylene for contaminated sensors).

This effort could be somewhat simplified if I were using a calibrated RH
meter.. I have these, but they haven't been calibrated for over 6 years,
so I'm not confident of their accuracy. The Solomat manual includes a
saturated salt reference method but I haven't tried it.

An alternative option could be to install a mechanical strip-type sensor
with an integral microswitch.. I've seen old versions which have set-point
adjustments, but I don't think I have any.

--
Cheers,
WB
.............





I recently tried to troubleshoot a Zenith dehumidifier that turned out ot be
a LG underneath. The humidy sensor was made by these guys

http://www.samyoungsnc.com/products/SYH-2R.pdf

and is capacitive in design. The board also has a thermistor with a 2wire
plus common wiring harness and sells for 27USD in Ohio.
--
They can have my command prompt when they pry it from my cold dead fingers.