Trimming the LTZ1000 tempco
Has anyone tried to minimise the LTZ1000 tempco? The unheated tempco
is high, so it could help to reduce the drift caused by the
imperfections of the temperature control system.
Datasheet page 6 shows a way to "null" the overall tempco
cds.linear.com/docs/Datasheet/1000afc.pdf
As far as I understand (most likely I don't) adding the R1 doesn't
much change the zener current. But the op-amp output voltage goes
higher to maintain the balance which increases the transistor Ic which
slightly changes the VBE tempco. But could we get the same result by
adjusting the 70k resistor instead? Adding a component is always a
source of drift. There is probably a reason to do it as shown in the
datasheet.
Will
Hello Will
slightly changes the VBE tempco. But could we get the same result by
adjusting the 70k resistor instead? Adding a component is always a
source of drift. There is probably a reason to do it as shown in the
datasheet.
Will
I tried the influence of the 70k Resistor on tempco since
I had to build mine of a 50K + a 20K precision WW resistor
and wanted to save the 20K resistor.
The change in tempco on my LTZ1000A #1 is from about 48ppm/K
with 70kOhms to 52ppm/K with 50kOhms.
(I measured this by changing the temperature setpoint voltage divider
from 12K5 + 1K to 12K5 + (1K || 22K) )
The schematic used you will find on 06.11.2010 on this list.
I have spent much time to get away common mode effects away
from the most sensitive point of the cirquit. (Wich is the setpoint
voltage for the temperature).
With best regards
Andreas
The change in tempco on my LTZ1000A #1 is from about 48ppm/K
with 70kOhms to 52ppm/K with 50kOhms.
A rough estimation:
Change in Ic when R1 is set from 0 to max value 200 ohms is equivalent
to changing the 70k resistor to 60k. I haven't prototyped or
simulated, so there can be more complex interactions that I haven't
included.
Maybe it is not even possible to "null" the tempco in a way simple
temperature compensated zeners allow to do that? The trick in the data
sheet seems not to be very effective.
Another alternative is to adjust the 120 ohm resistor which changes
the zener current.
It is surprising how slavishly all equipment manufacturers have copied
the original datasheet circuit during the last 20+ years.
Will
Will willvolts@gmail.com writes:
The change in tempco on my LTZ1000A #1 is from about 48ppm/K
with 70kOhms to 52ppm/K with 50kOhms.
A rough estimation:
Change in Ic when R1 is set from 0 to max value 200 ohms is equivalent
to changing the 70k resistor to 60k. I haven't prototyped or
simulated, so there can be more complex interactions that I haven't
included.
Maybe it is not even possible to "null" the tempco in a way simple
temperature compensated zeners allow to do that? The trick in the data
sheet seems not to be very effective.
Another alternative is to adjust the 120 ohm resistor which changes
the zener current.
It is surprising how slavishly all equipment manufacturers have copied
the original datasheet circuit during the last 20+ years.
Hi Will,
I was wondering about this too, it is not clear to me how important the
actual circuit values are. In the canonical "7V positive reference
circuit" we have R2=70k, R1=120 ohms.
OK, R1=120 ohms gets you 5mA zener current at the operating point (when
it starts to turn on the reference transistor Q1 at ~0.6V Vbe).
And R2=70k gets you a 100uA Q1 collector current, when Q1 is turned on
such that its collector voltage equals its base.
But these are awkward values when shopping for precision
resistors. R1=100 ohms and R2=50k would be easier. Yielding 6mA diode
current and 130uA. Is there anything wrong with that? Or are the
5mA/100uA magic values in some way? Perhaps they have been designed for
zero TC at 5mA - like a reference diode is selected for 7.5mA.
The (unstabilised?) "low noise reference" on page 1 has a 30k collector
resistor. And the "averaging" dual reference on P6 has 150ohms zener
resistor and a 5mA collector current!
--
John Devereux
The 3458A reference board has 111 ohms and 74.25 kohms which are not a
standard values either. It is also running 30 degrees higher than the
datasheet circuit, so maybe it minimises the tempco around that higher
temperature (pure guessing). Or HP was just stocking those values for
other purposes at the time the board was designed.
It would be interesting to do some voltage vs temperature plots with
different zener currents (from 3 to 8 mA for example). A potentiometer
between the power rails and a high value resistor from the wiper to
the temperature control voltage divider will do the job. 2mV/C change
is probably an assumption accurate enough to calibrate the
potentiometer scale in degrees. We don't need high accurary or good
long term stability, just an idea of the overall behavior which tells
us the direction to go when looking for the lowest tempco.
It is hard to believe that such a small variations in current would
affect stability. Except through selfheating in unheated references
but that is not an issue in this case.
Will
Hi Will,
I was wondering about this too, it is not clear to me how important the
actual circuit values are. In the canonical "7V positive reference
circuit" we have R2=70k, R1=120 ohms.
OK, R1=120 ohms gets you 5mA zener current at the operating point (when
it starts to turn on the reference transistor Q1 at ~0.6V Vbe).
And R2=70k gets you a 100uA Q1 collector current, when Q1 is turned on
such that its collector voltage equals its base.
But these are awkward values when shopping for precision
resistors. R1=100 ohms and R2=50k would be easier. Yielding 6mA diode
current and 130uA. Is there anything wrong with that? Or are the
5mA/100uA magic values in some way? Perhaps they have been designed for
zero TC at 5mA - like a reference diode is selected for 7.5mA.
The (unstabilised?) "low noise reference" on page 1 has a 30k collector
resistor. And the "averaging" dual reference on P6 has 150ohms zener
resistor and a 5mA collector current!
John Devereux
Looking at the data from various manufacturers of Zeners it seems that increasing the current makes the tempco become more positive.
Date: Sat, 1 Oct 2011 17:19:59 +0300
From: willvolts@gmail.com
To: volt-nuts@febo.com
Subject: Re: [volt-nuts] Trimming the LTZ1000 tempco
The 3458A reference board has 111 ohms and 74.25 kohms which are not a
standard values either. It is also running 30 degrees higher than the
datasheet circuit, so maybe it minimises the tempco around that higher
temperature (pure guessing). Or HP was just stocking those values for
other purposes at the time the board was designed.
It would be interesting to do some voltage vs temperature plots with
different zener currents (from 3 to 8 mA for example). A potentiometer
between the power rails and a high value resistor from the wiper to
the temperature control voltage divider will do the job. 2mV/C change
is probably an assumption accurate enough to calibrate the
potentiometer scale in degrees. We don't need high accurary or good
long term stability, just an idea of the overall behavior which tells
us the direction to go when looking for the lowest tempco.
It is hard to believe that such a small variations in current would
affect stability. Except through selfheating in unheated references
but that is not an issue in this case.
Will
Hi Will,
I was wondering about this too, it is not clear to me how important the
actual circuit values are. In the canonical "7V positive reference
circuit" we have R2=70k, R1=120 ohms.
OK, R1=120 ohms gets you 5mA zener current at the operating point (when
it starts to turn on the reference transistor Q1 at ~0.6V Vbe).
And R2=70k gets you a 100uA Q1 collector current, when Q1 is turned on
such that its collector voltage equals its base.
But these are awkward values when shopping for precision
resistors. R1=100 ohms and R2=50k would be easier. Yielding 6mA diode
current and 130uA. Is there anything wrong with that? Or are the
5mA/100uA magic values in some way? Perhaps they have been designed for
zero TC at 5mA - like a reference diode is selected for 7.5mA.
The (unstabilised?) "low noise reference" on page 1 has a 30k collector
resistor. And the "averaging" dual reference on P6 has 150ohms zener
resistor and a 5mA collector current!
John Devereux
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.
----- Original Message -----
From: "m k" m1k3k1@hotmail.com
To: volt-nuts@febo.com
Sent: Wednesday, October 05, 2011 8:45 PM
Subject: Re: [volt-nuts] Trimming the LTZ1000 tempco
Looking at the data from various manufacturers of Zeners it seems that
increasing the current makes the tempco become more positive.
Hi m k,
But on my LTZ1000A the tempco is positive around +50 ppm/K at nominal
current.
So you would have to reduce the current. But this would increase noise.
with best regards
Andreas
Andreas Jahn wrote:
----- Original Message ----- From: "m k" m1k3k1@hotmail.com
To: volt-nuts@febo.com
Sent: Wednesday, October 05, 2011 8:45 PM
Subject: Re: [volt-nuts] Trimming the LTZ1000 tempco
Looking at the data from various manufacturers of Zeners it seems
that increasing the current makes the tempco become more positive.
Hi m k,
But on my LTZ1000A the tempco is positive around +50 ppm/K at nominal
current.
So you would have to reduce the current. But this would increase noise.
with best regards
Andreas
Then you can easily trim the tempco as illustrated in the diagram taken
from the bottom of page 6 of the current datasheet.
Bruce
Will wrote:
Datasheet page 6 shows a way to "null" the overall tempco. As far as I
understand (most likely I don't) adding the R1 doesn't much change the
zener current. But the op-amp output voltage goes higher to maintain
the balance which increases the transistor Ic which slightly changes
the VBE tempco. But could we get the same result by adjusting the 70k
resistor instead?
Andreas wrote:
I tried the influence of the 70k Resistor on tempco since I had to
build mine of a 50K + a 20K precision WW resistor and wanted to save
the 20K resistor. The change in tempco on my LTZ1000A #1 is from about
48ppm/K with 70kOhms to 52ppm/K with 50kOhms.
Will wrote:
A rough estimation: Change in Ic when R1 is set from 0 to max value
200 ohms is equivalent to changing the 70k resistor to 60k. I haven't
prototyped or simulated, so there can be more complex interactions
that I haven't included.
It seems that the page 6 trick slightly reduces the tempco but it
still stays in the 50ppm/C area. Or I have forgot something in my
calculations. I don't have a spare LTZ1000 available at the moment to
run the tests but maybe Andreas has time in the future.
The original unheated tempco is surprisingly high when compared to
ordinary temperature stabilised zeners. Better stability at the
expense of tempco?
Another alternative is to adjust the 120 ohm resistor which changes
the zener current. It could be more effective way to alter the tempco
but there can be some side effercts?
Will
2011/10/6, Bruce Griffiths bruce.griffiths@xtra.co.nz:
Then you can easily trim the tempco as illustrated in the diagram taken
from the bottom of page 6 of the current datasheet.
Bruce
A tempco of +50ppm/K with a 7V output is equivalent to +350uV/K or about
0.18 x Vbe tempco (~ -2mV/K) which can be achieved by using a series
resistor of about 22 ohms or so.
The resultant 110mV or so increase in the voltage across the 70k
collector resistor will only have a relateively minor effect on the Vbe
tempco as The Vbe tempco is relatively insensitive to small (0.14%)
changes in Ic.
Increasing the Vbe tempco by 18% will (if achievable at all) require a
large decrease in Ic.
Bruce
Will wrote:
Will wrote:
Datasheet page 6 shows a way to "null" the overall tempco. As far as I
understand (most likely I don't) adding the R1 doesn't much change the
zener current. But the op-amp output voltage goes higher to maintain
the balance which increases the transistor Ic which slightly changes
the VBE tempco. But could we get the same result by adjusting the 70k
resistor instead?
Andreas wrote:
I tried the influence of the 70k Resistor on tempco since I had to
build mine of a 50K + a 20K precision WW resistor and wanted to save
the 20K resistor. The change in tempco on my LTZ1000A #1 is from about
48ppm/K with 70kOhms to 52ppm/K with 50kOhms.
Will wrote:
A rough estimation: Change in Ic when R1 is set from 0 to max value
200 ohms is equivalent to changing the 70k resistor to 60k. I haven't
prototyped or simulated, so there can be more complex interactions
that I haven't included.
Not true, how on earth did you arrive at this erroneous conclusion?
It seems that the page 6 trick slightly reduces the tempco but it
still stays in the 50ppm/C area. Or I have forgot something in my
calculations. I don't have a spare LTZ1000 available at the moment to
run the tests but maybe Andreas has time in the future.
Your analysis is faulty.
The original unheated tempco is surprisingly high when compared to
ordinary temperature stabilised zeners. Better stability at the
expense of tempco?
Another alternative is to adjust the 120 ohm resistor which changes
the zener current. It could be more effective way to alter the tempco
but there can be some side effercts?
It may not be possible to achieve this without exceeding the maximum
specified zener current.
Will
Bruce