Poll
Hello Andreas,
I would like to emphasize, what Bill wrote, i.e. use a 2nd inner shield,
completely around the LTZ circuitry.
This also helps to thermally stabilize the circuitry, just put some
foamed polstyrene inside the inner box.
Use separate grounds for supply and analogue signal.
I also could not implement that inner shield, because my layout with new
and very old ww resistors grew too big.
More compact metal foil resistors, (<1ppm/K typ.) as in the HP 3458A
design, would be preferred.
I replaced the diode at the output of the OpAmp (OP07) of the Linear
Device schematic by an ordinary npn, to reduce the load on the OpAmp and
to have a low impedance output stage.
If you did not yet order the LTZ, I suggest to get the LTZ1000, not the
A type.
This allows to use the lowest stabilization temperature, around 45..50°C
for lowest drift, see Pickering patent and description of the Fluke 7000
reference.
Most important is proper design and shielding of the supply circuitry,
stated also somewhere in the 7000 manual.
So, please design a separate PCB for this, so that this could be
shielded separately, and can be updated later, if you encounter
instabilities also.
As I operate the circuitry from the mains, there is definitely a lot of
potential to improve my very simple supply circuit.
The LTZ circuitry seems to be prone to multiplexer spikes, caused by
DMMs. (Although the 3458A should be very quiet in that aspect). So I
thought about an RC at the output to block these.
Anyhow, after 3/4 year of permanent operation of both LTZs, if I run
them against the 5442 and the 3458A (pimped to 65°C operation), all 4
standards are within 1ppm or less. That gives me confidence that the
LTZs really have a drift of around 1ppm/year as stated in the data
sheet. So I ignore the glitches at the moment.
Frank
Hello Frank,
I would like to emphasize, what Bill wrote, i.e. use a 2nd inner shield,
completely around the LTZ circuitry.
This also helps to thermally stabilize the circuitry, just put some foamed
polstyrene inside the inner box
Thats what I have already planned but with the polystyrene around the inner
Box.
within the box will be difficult since there isnt much room left.
Use separate grounds for supply and analogue signal.
ok.
More compact metal foil resistors, (<1ppm/K typ.) as in the HP 3458A
design, would be preferred.
Thats what I have planned for the 2nd build. I have already the 120R and the
1K resistor as Z201. But its difficult for me to get some Z201 resistors
with 70K or 12K4.
If I can't get them, I will have to build them from 25K and 20K resistors.
I replaced the diode at the output of the OpAmp (OP07) of the Linear Device
schematic by an ordinary npn, to reduce the load on the OpAmp and to have a
low impedance output stage.
I have planned a J-FET since the negative gate voltage will help to get a
better stabilisation of the Zener current even with low battery voltage.
If you did not yet order the LTZ, I suggest to get the LTZ1000, not the A
type.
Too late: it took me around 12 weeks to get the A-Type.
But since my design is battery powered I thougt the A-Type could
help me to save some energy. I want to have at least about 40 hours
without recharging. My power supply voltage is only 14V so I fear
that the non A-Type may get to some heater limit.
This allows to use the lowest stabilization temperature, around 45..50°C
for lowest drift, see Pickering patent and description of the Fluke 7000
reference.
Is this not possible with the A-type? I wanted to replace the 13K-Resistor
by a 12K4 to 12K5 resistor to get a lower temperature just shure
above 40 degrees celsius.
So, please design a separate PCB for this, so that this could be shielded
separately, and can be updated later, if you encounter instabilities also.
By the way. Is the LTZ1000 more immune to mechanical stress
than references in plastic housings. Otherwise I fear that the
weight of the batteries which I have planned to fix on the
same PCB as the reference will get some influence on the
reference voltage. I have at least the possibility to put
a slot between the battery area and the LTZ-area.
Anyhow, after 3/4 year of permanent operation of both LTZs, if I run them
against the 5442 and the 3458A (pimped to 65°C operation), all 4 standards
are within 1ppm or less. That gives me confidence that the LTZs really have
a drift of around 1ppm/year as stated in the data sheet. So I ignore the
glitches at the moment.
Frank
The most interesting question for me is wether the LTZ-based
references have a output voltage change when changing the
orientation.
(see my tilt-effect post for the LM399 based Keitley 2000 instrument)
How do your references and the LTZ1000-based HP 3458A behave?
best regards
Andreas
Hello Frank,
I have made my first measurements with the LTZ1000 cirquit.
At the moment I have replaced the LTZ by a discrete "ref-amp"
consisting of a NPN and a zener to test the start-up behaviour
of the zener-current. (I do not want to kill my LTZ at the first startup).
I found out that the LT1013 is instable with capacitive loading
when using the cirquit from the datasheet without additional buffering.
In my configuration with the FET as buffer the output oscillates
with capacitive loading of greater than about 33nF.
So I will put a additionally resistor (10K) between the 120 Ohms resistor
and the negative input of the current regulator. And a capacitor (100nF)
between output and the negative input of the regulator.
With this configuration the cirquit is stable up to around 100uF
which should be sufficient.
Maybe the glitches that you recognized have a similar root cause.
With best regards
Andreas
----- Original Message -----
From: "Dr. Frank Stellmach" drfrank.stellmach@freenet.de
To: volt-nuts@febo.com
Sent: Tuesday, September 14, 2010 7:28 AM
Subject: [volt-nuts] Poll
Anyhow, after 3/4 year of permanent operation of both LTZs, if I run
them against the 5442 and the 3458A (pimped to 65°C operation), all 4
standards are within 1ppm or less. That gives me confidence that the
LTZs really have a drift of around 1ppm/year as stated in the data
sheet. So I ignore the glitches at the moment.
Frank