Hi,

I have made many air baths over the years.
One very effective technique is to instal a fan or fans to cause the
air to swirl around
the inside of the box in a cyclonic pattern. One fan, near the side,
mounted tangentially,
is usually enough. Random air stirring is not nearly so effective.
To sample the temperature, a very small thermistor is mounted off the
opposite side
of the box. (half way rround the air circulation)
The air mass, about 1 kilogram per cubic metre, is now mixed with a
time constant of
1-5 seconds. The thermistor, with a good air velocity across it has a
time constant
of a second or two. Using light bulbs as heaters (shielded to contain
the radiation)
the unit has a bang bang controller without any hysteresis. Unlike
most electric ovens,
the device behaves as a fast response thermal mass, heating at say
1/100 degrees per second.
When the set point is crossed it cools at a similar rate. Control
continues cycling on and off
every few seconds with a triangle wave of better than 1/10 degree.
For it to work you need:
1 well stirred air mass
2 fast response temperature sensor
3 fast response (low mass) heater.
4 no time delay or hysteresis in the power control.
In cases where switching is too noisy, I insert a mono to keep the
heater off for 2 secs each time it turns off.
The unit then develops a 2-10 second switching cycle but with no
delay to the heater
being turned on when the set point is crossed.
This is only good for projects were the ambient 1-2m/s air flow does
not hurt the project.
Often it helps to keep the temperature gradients down.

cheers, Neville Michie

On 22/01/2012, at 1:20 AM, Will wrote:

Thanks for the practical advice. Exactly what I was looking for.

There are a few heater related things I would like to ask before
testing in real life

What is the optimal location of the heater, close proximity with the
blower or some distance?

High power heater and low duty cycle or vise versa?

Is there any specific reason to use a light bulb, except low price and
easy availability? I have some resistors which have bare wire on
ceramic body. I think they would be even faster than a light bulb. And
smaller size too.

Will

2012/1/22, Neville Michie namichie@gmail.com:

The blower/fan is only to keep the air swirling around the box and
mixing
although it may be better if it is just upstream of the heater, so
there is
maximum velocity over the heater.
If the box has volume of a cubic metre is contains 1.2Kg of air.
That has a specific heat of 1000 J.Kg/degC. In other words,
one watt will heat it one degree C in 1000 seconds,
100 watts in 10 seconds. This does not take heat losses into account.
If it is losing 20 watts then a 30 watt heater will heat it at a rate
of one degree in
100 seconds, so a one second response time from the thermistor will
control the temperature to a 1/100 degree overshoot. The overshoot is
what degrades control, so if your heater takes 10 seconds to cool you
have a 1/10 degree overshoot.
Incandescent light bulbs are good because they do not store much heat
and if you need more heater power you just plug in a bigger bulb.
Best control is a 50% duty cycle, there is power up your sleeve and the
overshoot is not too great.
The best heater would be a frame with nichrome wire making  a grid
with the air blowing through it, but for convenience and safety bulbs
are
hard to beat.
Use an opto-isolated solid state relay with zero voltage switching
and there will be less QRM.
Put 2 bulbs in series gives more surface area and longer life in the
rapid switching regime although bulb failure was not a great problem.

cheers, Neville Michie

On 23/01/2012, at 3:14 AM, Will wrote:

Neville wrote:

Interesting.  I would have thought that a fan placement that
generates lots of random turbulence would be superior to a
"patterned" flow such as cyclonic.  I'd be concerned that a patterned
flow could sustain temperature zones because the air mass might not
mix thoroughly or as fast.

Best regards,

Charles

I think it is because momentum is better conserved in a big swirl
whereas if you break the stirring into smaller cells they decay much
faster leaving
dead spots or areas that are under mixed.
The shear of the large swirl draws in the peripheral zones and the
density differentials
tend to aid the mixing.
What is certain is that quite a small amount of power in a grand
swirl does as much good as much more power in smaller
distributed cells.

I have also used the technique with a window air condtioner. A
special thermistor controlled circuit reduced the cycling of the
conditioner to less than 0.5 degrees. The compressor requires a
special algorithm where when the compressor cuts out it is inhibited
from starting until it has had 120 seconds to lose its back pressure.
Otherwise it starts or stops within a second of the setpoint crossing.
This algorithm increased the service life over other units that
relied on the standard hysteresis based mechanical thermal switches.
Two small computer fans were enough to maintain the swirl in a 4 by 6
metre laboratory.
I built more than 5 such installations.

cheers, Neville Michie

On 23/01/2012, at 4:13 PM, Charles P. Steinmetz wrote: