[time-nuts] Racal 9475 Rubidium

Adrian Godwin artgodwin at gmail.com
Thu May 3 10:38:00 EDT 2018

What sort of heatsink does it need ? Do you need the original part or would
a lightly-machined generic module do ?

There are some lovely heatsinks in obsolete (server quality) PCs. Huge
chunks of copper with a fine fin structure and a matched fan. Not all of
them are noisy.

On Thu, May 3, 2018 at 3:13 AM, Roger Tilsley <j.r.tilsley at runbox.com>

> Greetings Paul,
> The Racal 9475 does have some inherent shortcomings, the principal one is
> due to the power supply and the lack of the purpose-designed heat sink for
> the FRK module.  The unit is built into a “heat-sink case” but to keep the
> FRK base plate within the specified temperature range Racal had to operate
> it from a 23.5 V supply.  The power supply originally had a magnificent
> C-core transformer which produced a rather too-high output voltage
> consequently the raw d.c. supply was rather high.  The main supply voltage
> regulator has a rather weird configuration to facilitate heat sinking the
> pass transistor to the case and this unfortunately precludes using a
> battery back-up.
> What Racal overlooked is that with a linear regulator the total amount of
> heat dissipated is sensibly constant so the FRK module dissipates less heat
> than if it were operated at a higher voltage and the PSU pass transistor
> dissipates more heat.  When the case is opened a well-used 9475 usually
> stinks of “hot electronics”!  The problem is compounded by the fact that
> the 5 V logic supply regulator is not adjustable, it is essentially a high
> current follower from the tap of the R62/R63 potential divider across the
> 23.5 V main supply rail.
> In my experience, FRK modules produce their best performance when
> operated  from a supply voltage between 27 V and 28 V, selected for
> individual units but 27.6 V is a good starting figure.  At this voltage the
> FRK module needs the special heat sink if operated at high ambient
> temperature.
> More recent versions of the 9475 have a good but lower-quality mains
> transformer (perfectly  adequate for its intended purpose) which has a
> lower output voltage and this has mitigated the heat dissipation by the PSU
> pass transistor.
> The first thing to do is to monitor the voltage of the 5 V logic supply
> (between TP2 and TP3) and increase the voltage of the 23.5 V main supply
> rail (R39) until the 5 V rail is as much above 5 V as you care to risk for
> the 74-series ICs used.  A safe starting point is 5.2 V but I am wary of
> going too close to 5.5 V.  I have succeeded in getting several recalcitrant
> 9475s working properly by this simple ploy of increasing the supply voltage
> to the FRK module.  In my experience an FRK module operated a with a supply
> voltage lower than 24 V is prone to failing to lock or intermittent locking
> after long use.
> The complete cure is to increase the value of R36 to produce 5 V with the
> voltage measured between TP2 and TP3 with the 23.5 V rail wound up to
> between 27 V and 28 V.  This cures the high internal dissipation problem
> and considerably improves the performance of the FRK module but the special
> heat sink is necessary for operation in high ambient temperature or with
> poor ventilation.
> There are a few incidental modifications which I like to do to 9475s.
> They were designed when one could have whatever colour of LEDs which one
> wished provided that the colour was red!  I like green LEDs for the
> “normal” situation and with the ready availability of LEDs of different
> colours it is easy to follow the IEC recommendation of orange for “power
> on” and green for “normal situation”.  The “Rb lock” indicator (LP50) is
> already green but it is worth changing the “Power on” indicator (LP51) to
> orange or yellow and I consider it essential to change the “1 MHz”
> indicators (LP52, LP53, and LP54) to green.  If 4 greens are showing the
> unit is locked and the outputs are OK which is perceived with a quick
> glance.
> These mods and adjustments are simple and easy to do.  Unfortunately it is
> not nearly so easy to arrange a battery back up facility so it is wise to
> operate the unit from a UPS.  The warm-up frequency drift of an FRK module
> is still detectable 6 weeks or 1000 hours after starting from cold.  The
> 9475 was apparently intended for supplying the 1 MHz reference frequency
> for the synthesisers in Naval communications equipment and its drift is
> sufficiently low for this application within about 2 hours of starting from
> cold, however Time-nuts are more discerning!
> At the time of the introduction of the 9475, Racal were the agents for
> Efratom but went to great lengths to save the cost of the special heat sink
> for the FRK module and in my opinion “spoiled the ship for a ha’p’orth of
> tar” (for US readers ha’p’orth is a colloquial expression for a
> “half-pennyworth” of old UK money, about 1/250th of 1 US$ at the present
> rate of exchange).  False economy is also the reason for the 5 V regulator
> in the 9475, at the time of its design 7805 regulators cost an arm and a
> leg!
> There are not really any other worthwhile mods.  The quality of the 1 MHz
> outputs is remarkably good considering how they are derived.  In my
> experience FRK modules usually work satisfactorily if the “lamp voltage” is
> above 5 V (it should really be above 8 V but rarely is in disposal units).
> It is also worthwhile to occasionally tweak the FRK trimmer to keep the
> oscillator control voltage around mid-range, i.e. 10 - 12 V.
> It is not recommended to leave the 9475 meter switch in the oscillator
> control voltage position, the meter amplifier has too low an input
> impedance.  Using a high input impedance op-amp as a voltage follower on
> both the lamp voltage and oscillator control voltage outputs of the FRK
> module and switching the meter to measure the appropriate output voltage is
> a mod which some may consider useful - the temperature coefficients of Q12
> and Q13 do not cancel and, unlike that of the Q12/Q13 meter amplifier, the
> high input impedance of the op-amp does not load the oscillator control
> voltage monitoring output enough to significantly affect the output
> frequency.
> The 9475 is a good “starter” atomic frequency standard, it is reliable and
> easy to service and the FRK module is also easy to service.  Careful
> setting up of the FRK module improves its stability.  The older FRK modules
> found in old 9475s have a high oscillator control voltage which suddenly
> falls when lock is established while in more recent FRK modules the
> oscillator control voltage “hunts” up and down until lock is obtained on
> falling voltage.  Lock is usually obtained in about 5 minutes at normal
> ambient temperature though the output frequencies still changes slowly for
> a relatively long time.
> I hope that this is helpful.
> If any Time-nuts reading this know of a source of not-too-expensive heat
> sinks for FRK modules, I would be glad to receive details.  I wish to
> obtain two but I am a pensioner and cannot afford the cost of having them
> specially manufactured - the cost of the CNC setup kills it stone dead!
> Regards,
> Roger T.
> On Tue, 1 May 2018 10:18:35 +0100, "Paul Bicknell" <paul at bicknells.f2s.com>
> wrote:
> > Hi all new member hear could any of you help with the following
> information
> >
> >
> >
> > As I have just bought a Racal 9475 Rubidium and it has problems
> >
> >
> >
> > Is there any stock faults ?
> >
> > What is the life of the rubidium standard?
> >
> >
> >
> > Regards Paul
> >
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at febo.com
> > To unsubscribe, go to https://www.febo.com/cgi-bin/
> mailman/listinfo/time-nuts
> > and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/
> mailman/listinfo/time-nuts
> and follow the instructions there.

More information about the time-nuts mailing list