[time-nuts] Voltage Standards
Brian Kirby
kirbybq at bellsouth.net
Sat Nov 29 09:32:23 UTC 2008
We'll, I did not mean to hijack the previous thread on the subject.
In my case, I own one LM199AH/883 based voltage reference that I bought
several years ago. I have traceability on the unit from when it was
made and all the factory test data when they qualified it. It cost me
around $60 to get it because of the certifications.
It uses a constant current generator to set to 2 ma. That feeds two
LM399s in series that draw 1 ma, and then I feed a set of precision
resistors off of it to the LM199AH/883, which draws the other 1ma. The
constant current generator is a LT1000, with a buffer transistor. The
heaters are regulated by a LM317 regulator, that is referenced to
another LM399 regulator. The system runs off of a Lamba 24 volt linear
power supply, I have a set of power diodes "or'ed" where I can hook up a
couple of gell cells for backup power and so I can transport them to the
cal lab. This system holds 6.862429 volts and with no filtering on the
output it will wander around +/- 3 to 4 uV. When filtered, we have
watched this unit give around 0.3 PPM over 30 minutes to an hour. When
I built this unit, I burned it in for 6 months and then I carried it to
work where we had a HP3458A. I was able to use the voltmeter for about
2 weeks and we let it run and recorded 1 second data to characterize it.
This unit is my primary reference. I have since left that place of
employment and now I carry it to the cal lab once a year and they
measure it for me.
I also have another unit using a selected LM399. I bought 10 LM399s and
burned them in for a year and then we characterized them for noise. In
the end you keep the quietest one which has the lowest drift over time.
This one used OP07s for the constant current generators, and it
appears to perform around 2 PPM during the day. On this circuit, I also
bring out several points, so it has a 10V output, and a divider to bring
out 1.0183 volts (it was used for standard cell comparison), the 6V
reference point and 16V out of the current generator. I use this one to
checks voltmeters for quick test. I also hook it up to an external
bridge and use the bridge as a divider to generate precision voltage steps.
I built another super power supply, its based on several LM399s and its
basically a variable unit and you can set the output from 1 volt to 10.5
volts. It uses 2 each ten turn pots and you can set it down to the 10
microvolt level. I used it for working with synthesizers to set VCO's,
etc, and some precision temp sensor work, etc.
I had mention standard cells - Now that I think about it, its probably
not very useful for folks nowadays - they require extreme care and
handling and can be very easily ruined. I still use one in the
precision variable power supply in the above paragraph. I swap it in
place of the reference, because they are very quiet and its a quick
problem solver in my case. I have several standard cells. What I used
to do, was when the arrived, I measured them the next day and recorded
their values. Then I did not disturbs them for a month and checked
again. After 90 days, if they looked good, then I hand carried them to
the cal lab and I let them keep them for a week and they would watch
them and give me back the results.
Some folks mention the Motorola zener diodes and I believe that would
also make a good source for a project. I bought a differential
voltmeter at a hamfest one time and it did not work correctly. WHile
troubleshooting, I noticed that Fluke had wrote down the zener diodes
voltages along with the zeners current. I pulled these units and used a
constant current generator set for the values it needed and the voltage
on the zeners where what Fluke had called for. I have set them aside
for a future project. Also, I was able to salvage a lot of excellent
precision resistors from the differential voltmeter - and I picked it up
for $15. So thats another source of good parts to look at , at your
next hamfest.
The LTZ1000 use to be able to be purchased from DigiKey for $75-80 each.
I noticed they do not have any current stock and they ask you to call
if you want a single unit.
This is the problem with our hobby, it becomes serious...next thing you
know, we all end up owning part of a calibration lab...the best part is
there is a lot of excellent surplus test equipment out there that can be
picked up cheap. My method is I decide what I want, and I watch
auctions that are not too high priced to start with. I try to go into a
dialog with the seller and I have found that most who know what the
device is can tell you about the shape and condition of the unit. I buy
cheap and then I attempt to check it out. It it appears OK, then I
spend the money to get the unit calibrated. If the units fail, I hang
onto them a while, and then I will find another someday. If you find
another unit cheap, you may be able to mix and match parts and get a
unit that will play - and then I carry it off to be calibrated.
The cal lab I use, gives me a price break. They know I am not a
business, I let them know I am a ham who just likes to use good
equipment that gives repeatable performance. In my case they usually
give me before data, they cal it, and then give me after data, on my
precision multimeters. They take their time calibrating it, they
usually perform when their business is slow and that helps keeps the
cost down. I call before I go, just to see how there schedule is. On
other devices, I only have it cal to make sure its meeting specs, and I
do not ask for before/after data. All I want to know is it passed - and
that saves lots of money. And it will surprise you, some items do not
cost much for calibration - there are a lot of equipment that is closed
case calibration - they hook up to the calibrator and a controller and
it puts it thru the paces without much human intervention.
Maybe we should consider coming up with a standard voltage reference as
a TAPR project. We have a lot of good brainpower out here and it seems
a lot of experience available.
Brian - KD4FM
More information about the Time-nuts_lists.febo.com
mailing list