Hello Frank,

I want to build my first home made LTZ1000A-Reference the next days and
found an elder post:

Did you find the root cause of the spikes? Eventually your experiences
would help me to avoid some errors.

My planned cirquit uses the datasheet cirquit as a base with
pull-up-resistor
for the zener voltage to avoid starting issues. Further to reduce the
heating
of the LT1013 I have an (optional) BF245C J-FET as power-stage
for the zener current.

The power-supply is based on 12 NiMh-Cells which are regulated
by a LTC1763 down to 14.0 V. The whole is packaged in a EURO-sized
aluminium metal case together with a L200-based simple charger
and a undervoltage detection cirquit which switches off in case of
deep discharge.
The area around the LTZ1000A can be shielded with a tin can
metal shield.

For the first step I did no mechanical stress relief like in the
datron reference. With my UPW50-resistors (3ppm) there
is not enough place within the inner shield to do this.

best regards

Andreas

Hi Andreas,

If I may suggest, in your mechanical construction I would consider making the
circuit and its metallic shielding be an inner housing that is electrically
separate from the external metallic housing.  This would permit having a guard
terminal(s) to isolate common mode and leakage issues.

Basically, the circuit, battery and/or AC power supplies, and all connections to
such would be floating (except "earth" grounding) from the outside container.
The outside container would be tied only to the "earth" connection of the AC
power cord and/or any separate external "earth" connection that you may provide.

It is most instructive to study the Fluke 335D construction and circuitry.  This
instrument demonstrates the proper way regarding "guarding" and "leakage."  Other
"guarding" worth reviewing is the Fluke 731 and 732 reference supplies as they
also provided such processes.

Bill....WB6BNQ

Andreas Jahn wrote:

Hello Bill,

Thanks a lot for the informative hints:

I had a look at the Fluke 335D manual. So I think that against the Fluke
I am missing at least one shield (or even two since the inner shield does
not cover all cirquitry) in my planned construction.

I have only a outer metal case around all cirquitry including batteries
and voltage stabilisation
http://www.reichelt.de/?ACTION=3;ARTICLE=50424;PROVID=2402

and a inner metal shield around the LTZ1000 cirquit including resistors.
http://www.reichelt.de/?ACTION=3;ARTICLE=34042;PROVID=2402

The only thing that I eventually wanted to do is to connect some
filtering capacitors (1 to 5 nF) from the output terminals
(and the stabilized voltagte) to the inner shield.
I hoped this would work for a battery supplied reference when
removing 24V charger connection. But for shure I have some
improvement capabilities for the next build.

Something which I cannot get clear from the manual is the exact position
of the "Guard" and "Chassis" shield against the "Case".

The "Case" seems to be the outer tube around the Instrument.
But how is the front-panel constructed. If the  3 shieldings had to be
complete
I would need 3 isolated front panels. The outer connected to "Case".
The next inner connected to "Guard" and the 3rd as part of the "Chassis".

Best regards

Andreas

Hi Andreas,

You do need to be careful regarding the 335D.  What people miss is the inner
chassis is tied to the positive terminal.  The guard is floating from the outer
case and wrapped completely around the inner chassis.  When you open up the
instrument, you need to be mindful of that or you get the living hell shocked out
of you.

The concept of the guard is to shunt "ground" (earth) currents away from the
primary current path.  The best drawing in the manual to study is figure 2-4,
Guard Connection, on page 2-7 in the manual I was looking at.  From the Fluke
site it is marked as {  335_D__imeng0000.pdf  }.

I know the pictures are hard to see in the manual.  The front panel is part of
the outer, grounded case and the knobs are connected to the innards via a plastic
shaft so that keeps the guard system floating.

The Fluke 731 voltage standard uses the same basic ideal, but they went a bit
further in the design.  For one there are two different faraday shields inside
the power transformer.  One is connected to the "earth" ground and is wrapped
around the primary input coil.  The other is wrapped around the secondary coil
and is tied to the guard system.  Another important thing they did was to use a
"star" system for both of the internal positive and negative circuit paths as
indicated on the schematic.

While the 731 was an excellent device, it had a few issues.  The biggest being
the selector switch for the desired output voltage.  As you may have observed, it
was done away with in the 732 model.  The 732 has a similar guard system but it
is not as clearly demonstrated in its schematic as compared to the 731.

One of the tricks Fluke used with their resistor systems is to wind a resistor
using different wire with different temperature coefficients.  The value was
picked to be very close but NOT quite on.  Then they use a good wire wound single
turn variable resistor to finish the task.  The value of the variable is kept
quite small; either 10, 50, or 100 Ohms depending upon the size of the fixed
resistor.  This way the temperature coefficient of the variable is a very small
percentage of the total resistance and has almost no effect in the big scheme of
things.

So you could try selecting a resistor combination that is somewhat lower then
needed and do the same as Fluke and use a good wire wound single turn pot to make
that final tweak.  If the pot is contained inside the same heated space, then all
should be well.  Just provide a means of some holes for the pots that you can
cover up to keep from having the heat escape.

This same resistor process is used with the Fluke 720A Kelvin Varley Divider
(KVD).  There is no better KVD available.

As for temperature shielding, and aside from the thermal shielding of the
reference itself, having an isolated metal guard shield inside the outer cabinet
provides a given amount temperature shielding from air currents within a given
lab environment.  I suppose, from a transportable view point, some small amount
of thermal padding on the inside of the outer case would not hurt.

One other consideration when making a primary voltage source.  Fluke heavily uses
pre regulation before the reference.  You should do the same.  Something decent
like the UA-723; good for .01% of line variation.  A better choice if using line
voltage is the LT3080 if you pay attention to detail.  The LT3080 is like a LM337
regulator and must have a minimum lord current of 1/2 milliamp.  It claims low
noise and low dropout voltage.  But the more important specs are the claimed less
than mV load regulation and less than .001% line regulation.

With attention to design detail, the LT3080 should provide an excellent pre
regulation stage.  Borrowing from the Fluke ideas you should be able to make a
nice complete package.  Then we all can sit around and debate the quality of your
efforts.

73....Bill....WB6BNQ

Andreas Jahn wrote: