Electronics > Metrology

Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x

(1/129) > >>

zlymex:
Brief facts
 - 2DW23x series is a Chinese brand temperature compensated Zener reference
 - There are two identical zener diodes connected in back to back manner in a 3-lead TO-39 package.

 - Norminal voltage 6.0V t0 6.5V
 - Norminal zero-TC current ranges from 5mA to 15mA depending on types.
 - Datasheet(in Chinese) http://www.chinaeds.com/zl/%E5%B8%B8%E7%94%A8IC%5C2%5C2D%5C2DW230.pdf

Why 2DW23x
These 2DW23x series(2DW232, 2DW233, 2DW234, 2DW235) has been around for many year and was probably the only 'reference grade' zener in China. Previously known as 2DW7C and changed name about ten years ago. There were/are many manufacturers making these devices and I have tons of them. The performance varies according to makers very much and generally are poor (aging rate, tempco, noise). I've never use them in my designs/circuits.

However, people discovered that there is this Shanghai 17th Radio Factory making this particular Diamond brand(there is a diamond symbol on face of each device) with exceptional low noise, even lower than that of LTZ1000. I didn't believed, but there are a lot of buying/teardown/discuss about it:
http://bbs.38hot.net/forum.php?mod=viewthread&tid=49306
http://bbs.38hot.net/forum.php?mod=viewthread&tid=84620
http://bbs.38hot.net/forum.php?mod=viewthread&tid=119921
http://bbs.38hot.net/forum.php?mod=viewthread&tid=120264
http://bbs.38hot.net/forum.php?mod=viewthread&tid=120731

So, I built my own noise meter recently and bought some samples. At first, the noise tested is not good, 1.2uVp-p, same or slightly inferior to LTZ1000. However, some one soon pointed out that the power supply might be the problem. After I changed the power source to batteries, I got even better result than others: 0.4uVp-p at 5mA, which is one third of LTZ1000. I swapped devices, altered current, measured my other LTZ1000 based voltages(4910, 732A, 732B etc) as reference, and I bought more 2DW23x from other e-stores, they all show this ultra-low noise characteristic.


Availability
I bought about 200 unit of 2DW233 from here: https://item.taobao.com/item.htm?spm=a1z09.2.0.0.gpf2Zk&id=35815633601

Then I bought 550 unit of 2DW234 from another seller.

There are many sellers at Aliexpress selling these cheaply but only buy those with diamond mark on the top such as
http://www.aliexpress.com/item/Free-shipping-2DW233-DIP3/32433963421.html
http://www.aliexpress.com/item/hot-spot-10pcs-2DW232-new-original-in-stock/32637084815.html
https://www.aliexpress.com/item/High-quality-10Pcs-2DW234-2DW235-2DW236-New/32527935820.html
Be warned, I got these links by searching only. Someone was buying one of the entries and did not receive the package in 75 days.


My plan to test/use
1. A very simple circuit(batteries+resistor) to test/verify noise, done.

2. A simple circuit to test/seek the zero-TC point, partially done.
For a given device at given zener current, there is a convex shape of voltage-temperature curve and there is a peak voltage which is the zero-TC point for temperature.

Because of this, for a given device at given temp, there is a convex shape of voltage-current curve and there is a peak voltage which is the zero-TC point for zener current.
The type number(2DW232, 2DW233) indicating different nominal zener current for this zero-TC point. For instance, the zero-TC current of 2DW232 is 5mA, the zero-TC current of 2DW233 is 7.5mA, the zero-TC current of 2DW234 is 10mA, and so on.
But in reality,  there is not much difference. A device of any type may show zero-TC current at any value ranging from 4mA to 20mA, therefore, there is need to test every device for it's zero-TC current.

3. Accelerate aging
Boil at 100 degree C and freeze at -15 degree C, repeat the cycle many times.

4. A circuit to test aging rate etc.
 - build many to test simultaneously
 - characterize TC
 - test by continuous power up, power down/up, mild thermal shock, mechanical shock

The zener current(5mA to 10mA) is supplied by an foil resistor R1 from the final 10V. The current should be stable to 25ppm annually.
R2 and R3 are foil divider pair, the exact value is no very important but this 10V should be stable to 15ppm annually(assuming 2DW233 is stable).
I'll use AD707AH because I have a lot of them and the noise is not too bad(0.23uVpp, 14pAp-p).

5. Others
I've built one prototype but the noise is not as low as expected probably because there is no heat sink/thermal lag.



Reference
There is a Chinese military standard SJ 50033/150-2002 on 2DW230-2DW236.
http://doc.mbalib.com/view/7ff38949056ec076fcf871d4e67ec0f4.html


More datasheet from the industry
http://doc.mbalib.com/view/1a1945b2dcc8911e2a82b0f5ff52fc0f.html

I also found an old article(dated 1986, attached but in Chinese), the author were obviously from Shanghai 17th radio factory describing this 'superb' zener of Diamond brand.
He said, from 1982, the factory began a profound revolutionary change of the process structure of the 2DW230-2DW236 series making it rival similar products abroad in term of low noise and low drift. There are some test results in the article but the author did not tell anything about their new process. There was someone who tested an 2DW234 dated 1987 and got 3.2uVp-p noise, compatible with what have described in the article(http://bbs.38hot.net/forum.php?mod=viewthread&tid=120264). He also tested another 2DW234 dated 2013 and got 0.57uVp-p noise. This means that there must be another change of the process structure in Shanghai 17th radio factory between 1987 to 2013. One thing for certain is that they changed the internal connection of two zener from common anode(as can be seen from above two internal diagrams) to common cathode.

I took apart one(of 2DW234), here is the photo by my card camera plus a magnifier

Someone at 38hot(archwang) took a much better inside photo of 2DW234 dated 2013(see attached)

zlymex:
Low frequency noise test
By my DIY noise meter(https://www.eevblog.com/forum/metrology/diy-low-frenquency-noise-meter/), I tested some 2DW233.



For a given zener, the LF noise is inverse proportional to the square root of applied current. I plotted the curve for some well known zeners with 2DW23x which shows 2DW232 is about 2.54 times better noise wise or 6.5 times better current wise.



T.C. test



Aging test

Kleinstein:
With the two references in one case one could use the second zener as a normal diode to sene the chip temperature.

Form the internal pictures shown in one of the links, it looks like the chip itself is covered in a drop if transparent glue (e.g. epoxy). So heating to more than 100-120 C (and the speed of cooling from there) might have an influence on the glue and thus performance / aging.

With the relatively high power, there can be temperature fluctuation due to turbulent air flow - this might look very much like 1/f noise. Often one covers sensitve circuits with a kind of foam / fiber.

zlymex:

--- Quote from: Kleinstein on September 28, 2016, 03:29:06 pm ---With the two references in one case one could use the second zener as a normal diode to sene the chip temperature.

Form the internal pictures shown in one of the links, it looks like the chip itself is covered in a drop if transparent glue (e.g. epoxy). So heating to more than 100-120 C (and the speed of cooling from there) might have an influence on the glue and thus performance / aging.

With the relatively high power, there can be temperature fluctuation due to turbulent air flow - this might look very much like 1/f noise. Often one covers sensitve circuits with a kind of foam / fiber.

--- End quote ---

I think you are absolutely correct on these points. I'm going to design an temperature oven by using the forward biased diode. I also boiled 200 pieces for high temperature cycle but not higher.

Cerebus:
All very interesting, thanks for posting this.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version