The LTFLU (aka SZA263) reference zener diode circuit

[Kleinstein]

The circuit inside the 731 and 3330 is very similar - mainly without the diodes and smaller capacitors in the 731.

Die Influence of the diodes on the TC and thus the resulting transistor current to get zero TC is rather small. My rough calculation would give some 500 ppm/K for the zener current and maybe 3-5 ppm/ for the TC at constant transistor current. To compensate it would take current that is some 5 % lower. So nothing really significant and thus not really worth having extra diodes. One diode might make current compensation slightly simpler - but no real need with a single reference.

For temperature stabilization I would not consider an peltier element: It is best to have a temperature slightly above room temperature to reduce the humidity too. Cooling is problematic as is can cause humidity problems. The peltier element itself would provide a significant heat path an thus a rather small (and hard to get) peltier element would be needed. The heater only system could have better thermal insulation and thus a more thermal low pass filtering of external variations. It is also easier to distribute the heaters - compared to using several super small peltier elements.

[BU508A]

Hello branadic,

Thanks for the flowers, but if Andreas wants me to make a picture of the chip I will go ahead. Otherwise I can send him the part back. By the way, I still have the other opened sucker on my desk.

I know, it is not a nice move to crack such things open. But, I really want to know, how the chip inside is looking.
And, beside that. the piece I sent you is a LTFLU-1CH, the one you have on your desk is a LTFLU-1ACH. Curious, if there is a difference.

Maybe in a few days we'll know more.

If you want to send them back to me, please let me know. I'll pay for the returning.

Thanks,

Andreas

[cellularmitosis]

Thanks for the flowers...

You guys are a hoot 

Thanks for a great thread, all!

[branadic]

Opened the part today that I received from Andreas and yes it is again looking like a real LTFLU. Couldn't find a significant difference between A and non-A version, maybe I have to take a closer look. I took a few pictures, I'm sure Andreas will share them here.

-branadic-

[BU508A]

Hello branadic,

thank you very much for taking the pictures. Much appreciated it. 

It seems to me (I am not an expert in semiconductor manufacturing) that there is no obvious electrical difference between the LTFLU-1CH and the LTFLU-1ACH.  Any ideas, why they put in this "A" into the name?

The LTFLU1-ACH is the left one in these two difference pictures below.

And, as promised, here are the pictures:








Difference pictures:



Link to the website:
http://www.mounty.de/LTFLU-1CH/

And, for convenience and compariosn reasons, the link to the LTFLU-1ACH pictures:
http://www.mounty.de/LTFLU-1ACH/

@TiN: If you want, you can take the pictures and store them on xdevs.com as well.

Regards,

Andreas

[branadic]

I'm pretty sure the difference is the die attach. As you can see in the former opened LTFLU that the die attach is deviated while the latest pictures shows a well defined area for the die attach. They will not have changed the die itself.

-branadic-

[TiN]

Great shots. What setup was used for such magnification?

[mimmus78]

My SZA (older than me) just arrived.

I spent 25 USD and an unlimited amount of time searching on eBay (plus almost double $ for shipping). I was lucky TiN arrived late on this :-).

Meter is almost spot on at 10V so it should worth to salvage it and try something different.

Any good advice to where to start? I have an schematic saved somewhere form some hot Chinese BBS ... I need to start looking at it.

Inviato dal mio Nexus 6P utilizzando Tapatalk

[TiN]

What are your plans? There was no luck, TiN doesn't hunt for 8840A. 

[mimmus78]

My plans are to amuse me in some other voltage reference rabbit hole ... kinda bored of LTZ1000 ones.

Some mini hovenized SZA reference should keep me busy for some time.



Inviato dal mio Nexus 6P utilizzando Tapatalk

[Kleinstein]

An ovenized reference circuit can keep you busy for a while. However leaving out the oven does not necessary make is much faster: the oven helps to adjust the TC and without the oven the TC needs to be adjusted even more accurate. The oven could allow for a less accurate TC trimming and a good TC trimming would alow a less critical oven - so quite some choices. A crude oven might be easier than no oven.

With the SZA type reference there can be a few more options in how to build the circuit, especially if combined with a 7 to 10 V scaling.

[mimmus78]

An ovenized reference circuit can keep you busy for a while. However leaving out the oven does not necessary make is much faster: the oven helps to adjust the TC and without the oven the TC needs to be adjusted even more accurate. The oven could allow for a less accurate TC trimming and a good TC trimming would alow a less critical oven - so quite some choices. A crude oven might be easier than no oven.

With the SZA type reference there can be a few more options in how to build the circuit, especially if combined with a 7 to 10 V scaling.

Wondering if I can use the provided resistors as starting point to find optimal current for TC ... this can be a good way to begin before to think at the hoven.

Inviato dal mio Nexus 6P utilizzando Tapatalk

[Kleinstein]

The originally used resistors and circuit would be very good starting point. As that reference looks like being used without a heater should have a well chosen set of resistors. AFAIK these references come as a set with resistors. So if they were to sell these refs, this would be more like a set of chip and 2 or 3 resistors.

[mimmus78]

As far as I understand the original resistors cannot be used, they are tuned for 6.7V in this instrument and not for the 10V version.
If I run the SZA at 20ma ... can I use it as self heating and use the same method of the "pulsed heater" in LTZ1000 to find the zero TC resistor?

[Dr. Frank]

Surely you can use the original resistors, if you intend to replicate a 10V reference, and if the circuit also delivers 10V as an internal reference.
At least, you should use the same currents, as in this DMM , i.e. about 3mA for the zener, and some tens of uA for the collector current.
If you want to damage the device, go ahead and use 20mA ,but heating  the reference that way, I think is quite useless, and a brute force method.
Better build an oven assembly, for 45..55°C, like in the 732A/B and use the raw 6.7V output as an ultrastable reference..the 10V output, anyhow might not be as stable,or you also include these amplification resistors in the oven assembly.
Frank

[mimmus78]

Frank,

are you sure 20mA is too much? I understand (from this thread) that SZA236 ha 4 time the area of LTZ1000 and can be run at higher current than LTZ1000.

Than there was this lines few posts ago of Bob Dobkin and he said that "You can safely bias the LTZ1000(A) with 20mA." so this should be good also for the SZA236 that has bigger area.

If than I'm not wrong Fluke 732a uses almost 8mA ... that is half of 20mA.

[Kleinstein]

For a different (higher) current one would also need a different set of resistors to keep  a small TC. If one knows how one might be able to calculate the new values from the old one. Even with a higher current the voltage will not go up by much. Going from 6.7-6.8 V up to 10 V is a second step.

The main advantage of a higher current would be slightly lower noise. The downside is more self heating and thus a higher temperature and in case a oven is used more power needed to keep the temperature. As a crude rule of thumb the oven needs at least the same power as the devices inside, possibly more.  The stability will likely not be better at a higher current.
It is more like that at 20 mA there will be a lot of heat and thus the reference would be more sensitive to orientation / air pressure.
The larger die / structure does to directly result in more current capability. The limit due to bond wires can be the same (or may be even lower).

[SZA263]

are you sure 20mA is too much?

The SZA263 should be used with a Zener current of 3mA and a Collector current of about 50 uA, otherwise will be quite impossible to find a zero Tc bias.
The collector current is used to trim the Tc: increasing current will make the Tc more positive, and vice-versa.
In the Fluke 8800 the resistors are calculated for the internal bootstrap supply of 18V, so are useless if you need 7 or 10 V.

[TiN]

Also more heat on die means larger thermal gradients, and larger thermal EMF's around other parts.
Heat and oven itself does not make the reference better or more stable, actually the opposite, accelerate aging and increase noise (typical for most of semiconductors, not just the reference/zener junction). But with ultimate precision requirements, having ambient temperature variations is larger evil than these introduced instability factors from the small oven, so hence designs like LM399, LTZ1000 exist and sell for their niche field.

[mimmus78]

>> The SZA263 should be used with a Zener current of 3mA and a Collector current of about 50 uA, otherwise will be quite impossible to find a zero Tc bias.

Mostly looking at zlymex designs I missed this part at 8810a ... so they are using 18V rail.

Scaling this resistors to have the same transistor and zener currents at 10V should be a good place to start ... but where to find this resistors?

So let me guess in 732A they can go away with 8mA because they are running in a oven?

[SZA263]

So let me guess in 732A they can go away with 8mA because they are running in a oven?

In the 732A the Zener Current is 3 mA. It is supplied via R4 connected from the 10 V Output to the Zener cathode (~6.3 V).
So it see 3.7 V across, and with R4=1.27k the Zener current is ~2,913 mA.

[Dr. Frank]

Mimmus, as I have demonstrated in this blog,ALL reference amplifiers of Fluke have exactly 3mA. I don't get it, where from you take 8mA.
Also, in particular,  what has the Motorola SZA263 to do with the LTZ1000 from Linear Technology?

So, if you intend to violate this great chip...go ahead.

You're sick and tired of the LTZ?
CORRECT...
The SZA263 probably offers a positive drift /time in contrast, so it might deliver a very nice complementary reference for your LT references,which all feature a negative timely drift, typically.

Frank

[mimmus78]

My mistake about actual current of 732a, off course it is 3mA.

Frank (and the others) I hope that with your help no SZA263 will be harmed during this experiment :-)

[SZA263]

If you need a 10 V Ref. you can duplicate the circuit of the Fluke 731B. With a better power supply, a precision opamp like OPA277 (in place of the original LM308AH, I don't like AZ opamp ), and only four high precision resistor, you can realize a very high grade reference without any oven. But since only few components are needed, ovenize them will be not a problem.

[mimmus78]

Anybody checked what proposed by (z)Lymex few posts ago:



that's to eliminate R1/R2?

I'm interested in using only this 4 resistors as I already have some Vishay metal foil resistors (R401/R402) in the right value to get something like 9.999XXXV but I do not have the second divider (R1/R2).

My idea is to check this SZA with this schematic and than if it behaves than order some vishay VHP goodness for (R401/R402) (and eventually some others for R1/R2).

How they decided the ratio of the second divider (R1/R2)? I suppose must be in something slightly more (or equal) to achieve Vzener+Vbe?
How this missing two resistors affect the temperature compensation current when you trim output with R401/R402 ratio variation?