Where is the limit of the zener?

[CopperCone]

Has development just slowed down because of the JJ becomming economical? Or is there a theoretical limit?

I just realized the ltz1000 is 30 years old. I am supriszed no one did anything better, even as a publicity stunt.

[kj7e]

How can you do better than "God's own reference"?  Fun read: http://www.electronicdesign.com/analog/interview-analog-guru-carl-nelson

[branadic]

It's not only the reference itself but also the necessary AD converter. To be honest, there are people working on it, but nothing that can be talked about by now.

-branadic-

[cellularmitosis]

Has development just slowed down because of the JJ becomming economical? Or is there a theoretical limit?

I just realized the ltz1000 is 30 years old. I am supriszed no one did anything better, even as a publicity stunt.

Hmm, is the LTZ the best?  Or is that the LTFLU-1?

I'm actually not sure.  The LTFLU-1 is used in fluke's best standards, but is that simply an artifact of the 10V design requirement?  If the design requirement were simply a 7V output, the LTZ might very well be the better choice.

LTZ (7V circuit):

• Pro: built-in oven
• Pro: 7V output is low impedance
• Con: external ovenization required
• Con: resistor attenuation not as good (~100 minimum)

LTFLU-1 (7V circuit):

• Pro: better resistor attenuation (~200 minimum?) for the 7V circuit
• Con: 7V output is high impedance -- output buffer is mandatory
• Con: external ovenization required

[Kleinstein]

Wether the 7 V output of the LTFLU circuit is high impedance depends on the circuit chosen. If just a 7 V reference the output can be low impedance too. Another downside of the LTFLU reference is, that for best performance they need an individually selected resistor (or resistors) to set the currents and thus the raw TC without heater. The upside from adjusting the TC before the temperature regulation is that the temperature stabilization is less critical.

This difference is not directly related to the circuit topology, but more connected to the slightly lower voltage of the LTFLU. So I would consider a slightly modified  LTZ1000 with a slightly lower voltage and TC before the heater loop better than the current LTZ1000 - this would reduce the sensitivity to the temperature setting resistors. However not many factory rejected 6.9 V LTZ1000 have turned up.

[cellularmitosis]

Kleinstein, can you describe the low-impedance output?  My understanding was that the 7V output must be taken from the base of the transistor, which is fed by a resistor, and I can't figure out how to make that a low-impedance output (without adding a buffer).

[Kleinstein]

There is no need to feed the base of the transistor through a divider. If feed from the output of an OP the base node is low impedance. It is actually not that high impedance in the other circuits too because of the feedback loop - but loading the divider will shift the "10 V" output and thus effect the reference.

The downside of having the main output at 7 V is that one usually needs an extra auxiliary higher voltage (e.g. 10 or 14 V) to get enough gain from the transistor. So it needs an extra amplifier (e.g. x 2). Offset errors of the extra amplifier has about 100 times less effect than the offset of a simple buffer.

[AG7CK]

... My understanding was that the 7V output must be taken from the base of the transistor ...

There is no need to feed the base of the transistor through a divider. If feed from the output of an OP the base node is low impedance. It is actually not that high impedance in the other circuits too because of the feedback loop - but loading the divider will shift the "10 V" output and thus effect the reference.

The downside of having the main output at 7 V is that one usually needs an extra auxiliary higher voltage (e.g. 10 or 14 V) to get enough gain from the transistor. So it needs an extra amplifier (e.g. x 2). Offset errors of the extra amplifier has about 100 times less effect than the offset of a simple buffer.

As far as I know there is no Fluke circuit where a the nominal 6.5-7 voltage reference output is taken from the base of the transistor. If the needed reference voltage is the generic 7 volt or thereabout, a suitable topology different from the standard 10 volt refamp circuits used in 731/732 is chosen.

The circuit of the 8505A is an example of such a circuit. It differs from 731/732 topology in that it is more a "reference with an ERRORamplifier" than a "referenceAMPLIFIER". According to Fluke the 8505A is good enough for some 10 ppm or less / 7.5 digits in 10v averaging mode.



Note that the low impedance reference voltage output is taken between common (ground) and the output of op amp U2. The voltage arises in that the collector of the transistor in U1 is forced (by feedback) to the (virtually) same potential as the base. So the output is the sum of the zener diode and the BE junction with no loading of the base at all (ignoring transistor base current and imperfect op amp virtual potential / current cancellation).

[RoGeorge]

My profane question would be:
Why is not everybody already using voltage references based on Josephson junction?

Is this because it is too expensive for a personal lab to keep the very low temperatures required, or it is just because a Josephson chip is very expensive because of its rarity? Or is it that the results between LTZ1000 and Josephson does not justify the price difference? Or maybe it's just the availability?

What is the main reason that most volt-nuts are using LTZ1000 instead of a Josepshon junction based voltage reference?

AFAIK a Josephson junction voltage depends only by universal physics constants (which physics constants are, of course, always constant), so it seems a no brainer to go for the Josephson junction, yet it is not so common. Why?

[CopperCone]

you need a 75ghz source, cryo cooler, etc

and the people making JJ charge like 50g for one

[RoGeorge]

I was asking because I've seen this video linked somewhere else in this forum: https://archive.org/details/ucberkeley_webcast_Z9vBPQrG0U4

If it was possible to demonstrate the Josephson effect in a class, why not having such a reference at home?
Precise frequencies generating and measurement are very easy when compared with other analog measurements.

[cellularmitosis]

I was asking because I've seen this video linked somewhere else in this forum: https://archive.org/details/ucberkeley_webcast_Z9vBPQrG0U4

If it was possible to demonstrate the Josephson effect in a class, why not having such a reference at home?
Precise frequencies generating and measurement are very easy when compared with other analog measurements.

Yes, but you need to stack thousands of them (20,208 to be exact) to get to 1V.

However, I'm sure some volt nut will find a way within the next couple of decades

[cellularmitosis]

There is no need to feed the base of the transistor through a divider. If feed from the output of an OP the base node is low impedance. It is actually not that high impedance in the other circuits too because of the feedback loop - but loading the divider will shift the "10 V" output and thus effect the reference.

Hmm, so current goes from the op, to the base, then through the zener.  Does this mean all of the zener current is flowing through the base?  (or does a resistor inject additional current into the zener?).  Unfortunately I don't have a datasheet for the LTFLU-1, and have no idea what the limits are for base and collector current of the transistor.  Do you think this is safe?

[Kleinstein]

Most of the zener current is flowing through an extra resistor, parallel to the transistor of in a similar path.  There is only a small current (e.g. 100 µA range) flowing through the transistors collector and thus maybe 1 µA of base current.  It is the feedback loop that uses the transistor base as an input to cause the low impedance.

There are limits to how much current the transistor can handle - this may be less than the zener current, though short peaks might be allowed (e.g. during turn on).

User AG7CK inder post #8 has shown the circuit to get a 7 V reference. If needed the divider can go even more extreme to just a one sided resistor to get 6.9x V.