Ultra Precision Reference LTZ1000

[MiDi]

Tell me a mistake in my reasoning
A typical circuit stabilizes the current through the zener diode.
If we, knowing this current, add a precise resistor in series with the Zener diode, then we can get 10 V out by applying only 1 exact resistor.
Where am I wrong?

With LM399 or other zener diodes you can exactly do that (bootstrap).
For LTZ1000 the circuitry is totally different and not applicable in this way.

[Kleinstein]

A set of 2 resistors with matching TC is often cheaper than one resistor with good absolute TC at the same level.

Anyway the current through the zener is not as constant as the voltage. The current depends on one resistor (usually 120 Ohms) and the actual die temperature.

If it is about reducing the number of resistors, one could in theory combine the 7 to 10 V step with the divider for the temperature setting. This adds another OP and a little bit (but not that much) of extra drift and noise from that OP. So it's not the usual way.

The LM399 circuit may be combined to use the 7 to 10 V step to also provide the current, but this still needs to very stable resistors as a pair. Just for the current to the LM399 less accurate resistors are sufficient.

[MegaVolt]

For LTZ1000 the circuitry is totally different and not applicable in this way.

What will hurt? What is the physical phenomenon?

[MegaVolt]

A set of 2 resistors with matching TC is often cheaper than one resistor with good absolute TC at the same level.

Thank!

If it is about reducing the number of resistors, one could in theory combine the 7 to 10 V step with the divider for the temperature setting. This adds another OP and a little bit (but not that much) of extra drift and noise from that OP. So it's not the usual way.

Maybe you have a diagram of this idea? I do not quite understand how temperature setting is related to output voltage?

[Kleinstein]

The temperature setting is not related to the voltage, but there are already 2 highly stable resistors that could be extended to a third. So 3 resistors in a row to get the  500 mV , 7 V and 10 V.  It's more like a theoretical option - not really practical.

[branadic]

We know that SZA263 was a buried zener on one die and a silicon diode on a second die, mounted on some ceramic substrate, while LTFLU are two identical transistors with one used as a more or less ordinary zener. So the question is, how could a buried zener based refamp with the transistor integrated on one die plus the onboard heater perform? So the question in more detail is:

Has someone tried to use LTZ as a LTFLU like refamp with the LTFLU circuitry aorund? This could be done by using zener diode and Q2. Therefore connect pin 3 and 7 and there is your LTFLU like refamp on pin 4, 3+7, 6 and 8. So the already shown LTFLU circuit could be used to make a references, starting with 3mA through the zener (R12) and a simplified version of R7 divider. The remaining part of Q1 could be used as a temperature sensing diode. R13 is individual selected for a specific oven temperature.



What could be the advantage? R7 divider is still critical, but no extra gain stage for 7V --> 10V is required, thus lower noise of the 10V? Demands on oven stability are decreased? However, the oven is small and onboard so could be well regulated?
What are your thoughts?

-branadic-

[Kleinstein]

One can not wire the zener and transistor the other way round in the LTZ - they are already connected through the base and there is the substrate with more parasitic "diodes".

The LTFLU / SZA circuit is somewhat related, but for a different use. It essentially needs to make a higher voltage reference, more like 10 V. There is a way a round with an extra amplifier just for the collector current, but that would be more unusual.
There was a slight advantage of this configuration in the old days before good OPs, as one could directly get a 10 V or other higher reference. However the performance of the transistor to sense the voltage divider is limited. One is not really free to chose the transistor current for the amplification, but usually has to choose it to get a low TC. To get low noise it may take relatively low values for the divider R7, as there can be quite some base current (and thus current noise).

The LTZ circuit looks odd at first sight with the transistor at the low side, but it has some good points, as the transistor current can come directly from the 7 V level and only some 0.5 V for the current setting resistor is also not that bad. The circuit may get clearer if one considers it as a kind of bridge with the zener and R1 on one side and  V_BE and R2 at the other side.

There are quite a few modern OPs with less noise than the LTZ1000, so an extra OP for the 7 to 10 V stage is not a problem.
So the LTZ1000 internal circuit is good as it is - with the exception that the TC without the oven is relatively high and can thus not be trimmed to 0 by the transistor current. This is where the LTFLU is a little better.

[dietert1]

As far as i understood the LTZ1000 datasheet page 6 shows how to trim the TC for unstabilized applications.

On the other hand, running a LTZ1000 or LTFLU reference at a temperature where the TC curve exhibits a maximum or minimum may not be very desirable. It results in a "noise rectifier", since minor variations of temperature will not average out any more in the reference output. Better leave a small linear TC.

Regards, Dieter

[Kleinstein]

Adding a linear contribution does not help the rectifying action of a square law curve. So the nonlinear shape may be less visible if there is a linear background, the nonlinear part is still there.

The bigger "noise rectifying" effect is likely with the temperature regulation, due to the square law heater with a resistor. If well shielded it should not matter, but in a noisy environment and close to the drop out limit (low set temperature) there may be some effect of noise / variations in the temperature to shift the actual temperature a little lower. One could at least approximately correct the heater curve and this way get slightly better regulation close to the edge. However it's only s small improvement - though at moderate effort.

[Dr. Frank]

We know that SZA263 was a buried zener on one die and a silicon diode on a second die, mounted on some ceramic substrate, while LTFLU are two identical transistors with one used as a more or less ordinary zener. So the question is, how could a buried zener based refamp with the transistor integrated on one die plus the onboard heater perform? So the question in more detail is:

Has someone tried to use LTZ as a LTFLU like refamp with the LTFLU circuitry aorund? This could be done by using zener diode and Q2. Therefore connect pin 3 and 7 and there is your LTFLU like refamp on pin 4, 3+7, 6 and 8. So the already shown LTFLU circuit could be used to make a references, starting with 3mA through the zener (R12) and a simplified version of R7 divider. The remaining part of Q1 could be used as a temperature sensing diode. R13 is individual selected for a specific oven temperature.



What could be the advantage? R7 divider is still critical, but no extra gain stage for 7V --> 10V is required, thus lower noise of the 10V? Demands on oven stability are decreased? However, the oven is small and onboard so could be well regulated?
What are your thoughts?

-branadic-

Theoretically, that might work, but will probably damage Q1, due to restriction:

"Pin 4: Substrate and Zener Negative. Must have a higher
positive value than Pin 7. If Q1 is zenered (about 7V) a
permanent degradation in beta will result."

If Q1 zeners already <7V, this will not work.

Frank

[Andreas]

If Q1 zeners already <7V, this will not work.

If you look at the Absolute Maximum Ratings then Q1/Q2 must not be reverse biased above 2V.

A normal transistor is specced at 5 V  maximum reverse bias and will zener around 7-8V.
(and usually is a good noise generator in this mode).

So the chance that Q1 does not zener above 3-4V is very little.

With best regards

Andreas

[Insatman]

Just a fun photo.

I walked into my little lab today and took a snapshot of volt nut fantasy.   It's a LTZ1000 based reference that has been burning in for around 18 months since it's last tweak.  The truth is this number on my old fluke 8505A varies by up to +/-30uv at times.  But the reference is in a cardboard box, powered by a cheap walwart and has unshieled leads going to the meter.   Not to bad considering.   I have a better housed primary reference.  This was one of three boards I built and the best was selected for my reference chassis.

[SilverSolder]

Long term stability of +/-30uv is only 6ppm,  that is nothing to sneeze at.  Nice job!

[F64098]

Hello everybody,

after waiting a long time for all that good stuff to arrive, i finished my first two boards yesterday.

Inspired by the boards of branadic and Dr. Franks schematic i made my own board, adding a double RC-filter at the input of the LTC1052
and move the power supply to a separate board (not built yet).
The PCB fits into a TEKO 394 housing and it's enough space left for the power supply and maybe a heater regulator for the whole housing.
As you can see, in spite of everything, i forgot to order the LTC1052 
R14 will be measured and ordered after the burn-in period. The yellow jumpers can be used to connect the floating copper planes to top or bottom ground.
The boards are nearly without errors, all parts are fitting well and they are functionally. Just the drill holes for the two types of transistors are both a little bit to small.
Unfortunately one of the LTZ1000 seems to be defective. Zener and heater are good, but the junction of the two emitters / Pin 7 has a problem.
On both 70k-resistors i measure too much voltage compared to the running board. It seems, both transistors can't drive any current. Is that a known error?
Bought them as "used but tested" on ebay a while ago, but used them yesterday the first time.
I plan to build another two boards. Is it advisable to look for non-A LTZ 1000?

Best regards

Frank

[hwj-d]

Are this resistors from Edwin G. Pettis? I didn't got my payed sets since february from him

[exe]

I have questions about bc369.

1) What is special about it?

2) It seems it's not bc369 in the circuit, is it? bc369 seems to be to-92.

[F64098]

Are this resistors from Edwin G. Pettis?

Yes, ordered last year as a "Halloween offer".

I didn't got my payed sets since february from him

Be patient, they are really handmade and i also had to wait 5 months.

Best regards

Frank

[F64098]

I have questions about bc369.

1) What is special about it?

Nothing. Just a simple cheap PNP from my standard distributor.

2) It seems it's not bc369 in the circuit, is it? bc369 seems to be to-92.

I made the layout for TO-92 and TO-126 just for experimental purpose. I used the BD176, because there is a tiny little layout error for the TO-92

Best regards

Frank

[Andreas]

why PNP?
The standard positive cirquit uses a NPN

with best regards

Andreas

[F64098]

 
I confused the "original" BC639 with the BC369.
But i used a NPN-Transistor and soldered in a BD176.
The layout is already corrected (bigger holes) and the TO-92 pins are gone...

Best regards

Frank

[beanflying]

Still waiting patiently for a few sets from Edwin too. Hope he is slaving over a bobbin winder and not on Summer holidays 

[Andreas]

But i used a NPN-Transistor and soldered in a BD176.

According to my datasheets the BD176 is a PNP with BD175 as NPN complement.

with best regards

Andreas

[F64098]

 

Okay, i think my license to solder is finally lost...
Time to join the Supervised Thinking.

Changed the transistor to a 100% REAL NPN and look, both units perform _slightly_ better...
But on the other hand,  this embarrassing mistake saved both LTZs from being grilled by a defectice 1kOhm setpoint resistor.
Handled them all with maximum care and bent them with a special tool but one of them is really dead.

Best regards

Frank

[3roomlab]

Still waiting patiently for a few sets from Edwin too. Hope he is slaving over a bobbin winder and not on Summer holidays 

hmmm I have been wondering for some time
did anyone have load life specs for his resistors? xx ppm yy C @ zzzz hrs 90min on 30min off ?

[iMo]

How are the Edwin's resistors manufactured? Any details on that? Is the resistive wire enamel coated? Bifilar winding (low L)? For such a resistance like 70k/12k you would need a pretty long and thin wire to wind on that small core, imho..