[time-nuts] Replacement fan in SR620

Magnus Danielson magnus at rubidium.dyndns.org
Sun Feb 2 09:46:31 EST 2014

On 02/02/14 03:43, Charles Steinmetz wrote:
> Jarl wrote:
>> In my SR620 the fan is a Delta DBF0624H.  It is a 60x60x25 mm fan, 24V
>> /0.11A.
> Mine, too.  Does anyone know the airflow rating of the Delta fan?  It
> does not appear to move as much air as it should (at least not if the
> idea is to hold the interior at a fixed temperature) -- every SR620 I've
> used has the fan running at full speed by the time it's been on for 10
> minutes, unless the ambient temperature is less than 17-18C.
> On the one hand, it would be nice to move more air so the internal
> temperature is more tightly regulated; but on the other hand, it would
> also be nice if the fan were quieter.  It is unlikely we can have both.
> The fact that the original fan seems marginal counsels against replacing
> it with one that moves less air.

Well, considering the conditions it can't do much else.

> The thermal design of the 620 is far from optimal.  The thermistor is
> located in a "tunnel" between the interior of the instrument and the
> exterior, with the fan blowing directly across it through the tunnel.
> So, the fan startup is an ugly process of fits and starts as the
> instrument warms up.  Also, the air inlets seem to be too small and the
> internal airflow was not properly designed to circulate cooling air
> where it needs to go.

If the heat sources where well coupled to the air-flow, which they are 
not, and the flow-path as low air-flow resistance, which it also doesn't 
have, requires the fan to work at high rate to get any air move, and to 
get the thermistor happy.

> I have toyed with the idea of cutting a slot maybe 4" long and 1/8" tall
> in the right wall of the chassis at the rear, above the four TO-220
> devices mounted there -- and perhaps another slot about the same size in
> the top cover above those devices.  Also, maybe attaching some internal
> baffles to the top cover to channel airflow where it is needed.

You can mount a fan to create a steady flow on the right side, and that 
way cool off much of the heat there.

> I installed a terminal strip on the GPIB connector mounting screw and
> relocated the thermistor there in my SR620s (in the general vicinity of
> the oscillators).  (However, note that stabilizing the internal
> oscillator temperature is not really very important for most of us,
> because time nuts generally use an external time base.  In that case, it
> is probably more important that the temperature of the triggers and
> interpolators is held constant.)

Well, we *do* care about a stable internal temperature, since it will 
also shift calibration factors, so the stabler the internal temperature 
is and hence various shifts which is being compensate, the more accurate 
it becomes.

> With the thermistors relocated as I have described, the fans start up as
> they should (monotonically, speeding up smoothly from stopped to full
> speed without any fits and starts).  They still reach full speed in 10
> minutes or so, so at the end of the day I don't think I've really
> changed anything except the aesthetics of the fan startup.  (IMO, the
> change is worth it just for that, but there does not seem to be any
> operational improvement.)

If you try to achieve the goal I think it will require a different approach.

> Perhaps SRS did not intend to regulate the interior temperature of the
> SR620 -- maybe they just wanted it to warm up faster (if you did away
> with the thermistor and had the fan run full speed whenever the counter
> was on, it would presumably take longer to warm up).

Maybe, would make kind of sense, on the other hand, they could have 
achieve both quick heat-up and stable but lower temperature and quieter 
if they wanted.

> Anyway -- does anybody have an old Delta catalog or datasheet that
> specifies the airflow rating of the original Delta fan?

It would indeed be interesting. The Papst 624 seems quite capable little 
critter and there seem to be some magic to the 624 number matching, 
which doesn't seem to be accidental.


More information about the time-nuts mailing list