Author Topic: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x  (Read 237716 times)

CDN_Torsten, Rosco and 1 Guest are viewing this topic.

Offline gamalot

  • Super Contributor
  • ***
  • Posts: 1300
  • Country: au
  • Correct my English
    • Youtube
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #375 on: February 10, 2017, 06:54:51 pm »
I don't think they have any website or other contacts details except a phone number, do you speak chinese?  :-//

I have colleagues that speak chinese - a phone number or email address could be a good start ...
Thanks

Flinstone

+86 21 6377 2411

Good luck  :)

Offline CalMachine

  • Frequent Contributor
  • **
  • Posts: 477
  • Country: us
  • Metrology Nut
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #376 on: February 10, 2017, 06:56:49 pm »
technix is Chinese, he's the one who's been talking to them in person, he lives near the factory.

I got mine from Technix.  I have not had a chance to characterize any of them yet, however :(  They do seem like the legit low noise ones!
All your volts are belong to me
 

Offline gamalot

  • Super Contributor
  • ***
  • Posts: 1300
  • Country: au
  • Correct my English
    • Youtube
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #377 on: February 11, 2017, 01:06:36 pm »
Hello VoltNuts - has anyone got a street address of 17th radio of factory, mail address - as one of colleagues - who speaks chinese - will try to contact them locally ?

I also wonder in which IC process these are made to manufacture those so low noisy ... - some competitors manufacture them in silicon epitaxial planar process but those will not be low noisy.

Can anyone forward the datasheet from 17th radio factory about those 2DW232 ... 2DW235 series of zener diodes ?

Thank you very much !
Best regards

Flinstone

Address : No.175, Dongjiadu Road, Huangpu District, Shanghai, China

No email address published.
« Last Edit: February 11, 2017, 01:08:20 pm by gamalot »
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #378 on: February 13, 2017, 02:06:17 am »
Hello VoltNuts - has anyone got a street address of 17th radio of factory, mail address - as one of colleagues - who speaks chinese - will try to contact them locally ?

I also wonder in which IC process these are made to manufacture those so low noisy ... - some competitors manufacture them in silicon epitaxial planar process but those will not be low noisy.

Can anyone forward the datasheet from 17th radio factory about those 2DW232 ... 2DW235 series of zener diodes ?

Thank you very much !
Best regards

Flinstone

Address : No.175, Dongjiadu Road, Huangpu District, Shanghai, China

No email address published.
That address is on their officially registry but the office have moved to Pudong. I live in Shanghai though, and their office is within public transit range for me.
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #379 on: February 16, 2017, 07:15:42 am »
I am going to try and create a better board for my 2DW23x - to get some long-term drift data - which at the moment is missing information.
 

In the meantime, I found a source for some minature peltier coolers. This would open the possiblity of replacing the heating loop used for thermal stabilisation with a Peltier cooler to do the same thing. The regulation temperature could be set to co-incide with expected room temperature - say 23C.

Advantages,

  - less thermal stress on the 2DW23x when cycling between use and non-use. Reduced hysteresis issues. Also between the time that the part is manufactured and the time that it is deployed/used.

  - It feels elegant to balance the heating effect from the 20-30mA zener current with a cooling loop. At the moment there needs to be enough headroom above the ordinary heated temperature for the temperature regulation loop to have room to work in various ambient settings.

  - The poor power efficiency of Peltier devices is discouraging. But if we accept that the reference is already high-current then it seems less of an issue.

  - lower thermal gradients over the device - because the regulation setpoint is closer to ambient temperature.
 
  - More amenable to the carrier board solution - to keep the precision resistors at the same temperature as the 2DW23x. This is because of the lower second-order TCR affects of the Z201 around 25C .

Disadvantages,

 - Is it just really over-complexifying the design, and assembly?

Some more details, http://blog.julian1.io/2017/02/16/2dw232-circuit-7.html
 
The following users thanked this post: TiN

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #380 on: February 16, 2017, 03:12:41 pm »
  - The poor power efficiency of Peltier devices is discouraging. But if we accept that the reference is already high-current then it seems less of an issue.

People always seem to get this one wrong. If you use a Peltier device at the right point on its characteristic curve(s) the efficiency is quite good. It's easy to get a  decent CoP (Coefficient of Performance, the ratio of heat moved to energy supplied) if you pick the right point on the curve. The higher the deltaT across the device, the lower the efficiency, the higher the heat flow as a proportion of the rated maximum the lower the efficiency.

If you use small deltaTs (say 10 degrees) and under-rate the device (say ~20%) you can get a CoP of 3.5 without trying too hard. So, a 60W peltier will maintain a temperature difference of 10 degrees, removing 12W of heat for a 3.4W supply of power - change that to 20 degrees and the CoP falls to 1.5, 30 degrees and the CoP falls below unity.

So, for a 300mW semiconductor device to be cooled, pick a 3W Peltier device, a 33 C/W heatsink for the hot side of the Peltier device and you'll get your device to 10 degrees below ambient for a power budget of 100 mW.

Anybody got a syringe I can use to squeeze the magic smoke back into this?
 
The following users thanked this post: Smoof, julian1, rachaelp

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #381 on: February 16, 2017, 04:38:18 pm »
For testing stability, I would take the raw (e.g. 5.3 V for just the zener or 6.2 V for zener+diode) voltage as the critical part. I don't think one would need super stable resistors for the first test, as the temperature of the resistor(s) could be stable as well. Before doing a long time test, it might be a good idea to look for thermal hysteresis.

There is an advantage in using a temperature above room temperature: this also makes that circuit part low relative humidity. Cooling below room temperature can be critical as this could even lead to condensation. So even with a peltier element a temperature around 30 C (for the resistors) is likely more suitable. So more like no active cooling if the room temperature is low.
For good efficiency, a peltier element should be operated at reduced current, if the temperature difference is small. AFAIK the current for best efficiency goes down proportional to temperature difference. So 20% of the nominal current at 10 C difference is about right. Still a good heat sink for the hot outer side is needed.
 
The following users thanked this post: julian1

Offline z01z

  • Regular Contributor
  • *
  • Posts: 151
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #382 on: February 17, 2017, 10:23:27 am »
Some more details, http://blog.julian1.io/2017/02/16/2dw232-circuit-7.html
A cautionary note about the foil resistors you've mentioned in your blog: you might get lucky and get one with the TCR as good as the typical range mentioned in the datasheet but only the worst case range is guaranteed. You might want to check out the TC measurement topic.
 
The following users thanked this post: julian1

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #383 on: February 17, 2017, 11:26:52 am »

A cautionary note about the foil resistors you've mentioned in your blog: you might get lucky and get one with the TCR as good as the typical range mentioned in the datasheet but only the worst case range is guaranteed. You might want to check out the TC measurement topic.

Thanks for the heads up on expectations for the z201. I haven't put a lot of thought into it and was even kind of buying on price. My initial intention was to use s102 which are 2ppm/C. Then I saw the z201 range with apparent 10x better TCR with only a modest difference in price.

I think I am probably juggling a couple of different goals. The important thing is to create a reference suitable to be left powered on for six months in order to obtain some longer-term drift data. For this task - perhaps inexpensive resistors would be fine as Kleinstein notes. So long as data was sampled (perhaps once per week) under identical room temperature conditions.

For price/performance, Vishay ptf56 look impressive with 5-10ppm/C. The 10k ptf56 are less than $1. In a divider ratio they're probably about 2ppm. I tested voltage sensitivity of the 2DW232 in series configuration to current and it's about 1/80x.

I am not sure how to account for resistor hysteresis except perhaps to sample and record their resistance - or by buying more expensive ones...
 

Offline bertik

  • Contributor
  • Posts: 30
  • Country: ch
    • W's Laser Web Site
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #384 on: February 17, 2017, 11:59:54 am »
I am following this with interest, by having similar plans. Ultimatively it will be about a low noise, constant current laser diode driver but the basic reference stage should be similar. So far I was using Peltier control of the Zener temperature, but that in a test setup and not for real use, where it would be overkill and anyway a PWM control should better not be used anywhere near a low-noise reference.
 

Offline MrBungle

  • Supporter
  • ****
  • Posts: 75
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #385 on: February 18, 2017, 05:21:45 am »
I am following this with interest, by having similar plans. Ultimatively it will be about a low noise, constant current laser diode driver...
Trying hard not to go off topic, but have to say I like the cut of your jib!  :-+
I'm following for the exact same reason  ;D
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #386 on: February 18, 2017, 09:17:09 am »
How good the resistors for a long time drift test need to be is a difficult question. If the refs are really good, it is well possible that the resistor(s) set the final drift performance of the full reference circuit. If the references show a lot of drift, it would be a kind of wast using very good resistors, bit we don't know up front. How good reference drift can be resolved also depends on the equipment available. So with an equipment that might be good to a few ppm/a one should use resistors that do add much more error. To get a 1 ppm stability for the voltage would need something like 50-100 ppm for the resistor.
At least the two resistors for setting the gain should be measurable (e.g. check the voltage ratio with an DMM). So at least for those resistors the PTF56 should be good enough, especially as the power is rather low for those. For the final current setting resistor the PTF56 could be about the lower limit on how good the resistors should be for a long time stable reference.

Ideally one would do a long term drift test with more than one unit - they may not behave all the same.
With similar diodes one might be able to do this with several zeners in series  - thus only one precision current source and switchable test points to measure the voltage.

How much resistor drift cause voltage drift depends on the zener resistance. There are a few crude data on this, but it might be worth checking the actual used refs at a current closer to the one used. This could be an important property for an reference and with lower voltage zener diodes this figure is expected to be not as good as the classical 6.2 V + forward compensated zeners.
 
The following users thanked this post: julian1

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #387 on: February 21, 2017, 09:21:54 pm »
Just a comment on resistors- trust no one! Measure the actual resistors you're going to use, immersing them in warm/hot mineral oil and cold. Solder some and recheck the values. The resistor you know is far better than the promise of a data sheet.  8)
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #388 on: February 25, 2017, 11:32:58 am »
PCB boards look quite nice,



The circuit works, even if there's a bunch of goofy things that could have done differently.



I'll make one last stab at adjustable current and set-point trim with Bourns 3250 trimmers - since trimmers are just so convenient. I believe I can get a reasonable current range for the 2DW232 TC0 with only around 5% of the total divider resistance on the trimmer - with it's poor TCR.

If overall circuit tempco is still poor then everything will be fixed resistors.
 

Offline BFX

  • Frequent Contributor
  • **
  • Posts: 376
  • Country: sk
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #389 on: February 25, 2017, 06:51:56 pm »
PCB boards look quite nice,



The circuit works, even if there's a bunch of goofy things that could have done differently.



I'll make one last stab at adjustable current and set-point trim with Bourns 3250 trimmers - since trimmers are just so convenient. I believe I can get a reasonable current range for the 2DW232 TC0 with only around 5% of the total divider resistance on the trimmer - with it's poor TCR.

If overall circuit tempco is still poor then everything will be fixed resistors.

It's so nice could you please provide schematics , gerber and so?
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #390 on: February 25, 2017, 09:28:23 pm »
Quote
It's so nice could you please provide schematics , gerber and so?

I've made a bunch of mistakes with the prototype. The main one being not to pull the temp-sense and temp-set signals out to the pin-headers to make it easy to monitor and dial in settings.  :palm: Instead, II need to probe at the op-amp input pads to get these readings. It would also be good to have easy test-points for data-logging resistor drift/hysteresis. I'll put up the source files when I'm confident that some of these useability issues are fixed.
 

Offline BFX

  • Frequent Contributor
  • **
  • Posts: 376
  • Country: sk
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #391 on: February 25, 2017, 11:38:33 pm »
Quote
It's so nice could you please provide schematics , gerber and so?

I've made a bunch of mistakes with the prototype. The main one being not to pull the temp-sense and temp-set signals out to the pin-headers to make it easy to monitor and dial in settings.  :palm: Instead, II need to probe at the op-amp input pads to get these readings. It would also be good to have easy test-points for data-logging resistor drift/hysteresis. I'll put up the source files when I'm confident that some of these useability issues are fixed.
Discussing mistakes might be also helpful 8)
 

Offline CalMachine

  • Frequent Contributor
  • **
  • Posts: 477
  • Country: us
  • Metrology Nut
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #392 on: March 20, 2017, 12:40:04 am »
Finally got a quick and dirty prototype made just so I can get my RPi logging set the way I want it before the big boys are made!  Here is what I have so far...  I pretty much just copied the circuit provided by Zylmex.

Preliminary logging results to come!

All your volts are belong to me
 

Offline Pipelie

  • Regular Contributor
  • *
  • Posts: 172
  • Country: cn
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #393 on: March 20, 2017, 01:31:29 am »
 :-+
looking forward to see  the test result.
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #394 on: March 24, 2017, 09:43:22 am »


I've put another board together. The resistors used are,

    Vishay ptf56-Z (5ppm/C) for OP feedback divider and temp-set divider.
    Vishay z201 (0.2ppm/C) for 200ohm current limit resistor.
    Bourns 3250w for t-set fine adjustment.

Unfortunately, the non temp stabalized components are pretty temperature senstive.



I discovered it is possible to test resistor contribution to circuit tempco by spraying them with isopropyl and observing the effect of evaporative cooling on the ref output. The z201 shows no obvious change, while the ptf56 produce a 100uV-200uV change.

The goal was to gather some longer term drift data - but I suspect the ptf56 will need to be exchanged for z201 for that.
 

Offline SvanGool

  • Regular Contributor
  • *
  • Posts: 135
  • Country: nl
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #395 on: March 24, 2017, 10:06:08 am »
@julian1

Your graph is showing app. 3.2PPM/°C and the extreme 200uV change is still less than 33PPM over the whole temperature delta.
Maybe I have my goals lower set than yours, but for a 6.5 digit DVM this is good enough and I wouldn't use this reference for anything better than that, or am I underrating the device?
# Don't hurry, the past will wait. #
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #396 on: March 24, 2017, 07:49:22 pm »
Even with some sensitivity to room temperature, one can still look for long term drift. If in doubt, just make sure the temperature is approximately the same. So if the temperature is within 2 K, the error would be only 6 ppm. Closer temperature mating is possible.

There just is a limit on how much drift is detectable. Higher quality resistor could permit a lower detection limit, but I don't expect the reference to be that good. One parameters that determines how good this reference is for a long term stable ref. is the zener resistance in relation the the current. The number of R_zener * I_zener / U_zener is the relevant figure of merit. To a first approximation this gives how much drift of the current setting resistors transfers to the output voltage. Due to the lower zener voltage the R_zener is expected to be higher compared to the more normal 6.8-7 V compensated zeners.


At least the effect of the resistors would the amplifier and the set temperature can be checked for additional aging drift, by direct measurement of the voltages. So one would know if there is a lot of drift and could in principle correct (add resistors or numerically).
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #397 on: March 24, 2017, 09:12:21 pm »
@julian1

Your graph is showing app. 3.2PPM/°C and the extreme 200uV change is still less than 33PPM over the whole temperature delta.
Maybe I have my goals lower set than yours, but for a 6.5 digit DVM this is good enough and I wouldn't use this reference for anything better than that, or am I underrating the device?

Well I feel encouraged by your positive view. The circuit is rather more fiddly to set up than I want - in terms of setting the 2DW232 current to fixTC0, and then dialing in the heater stabilizer to cancel second-order temperature effects.

But having done that work - it's frustrating that everything ultimately comes down to (expensive) resistor choices rather than the $1 cost of the semiconductor. This seems to be universally true of voltage references regardless of device - microsemi zener, lm399, ltz1000,
 

Offline julian1

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: au
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #398 on: March 24, 2017, 09:39:28 pm »

There just is a limit on how much drift is detectable. Higher quality resistor could permit a lower detection limit, but I don't expect the reference to be that good. One parameters that determines how good this reference is for a long term stable ref. is the zener resistance in relation the the current. The number of R_zener * I_zener / U_zener is the relevant figure of merit. To a first approximation this gives how much drift of the current setting resistors transfers to the output voltage. Due to the lower zener voltage the R_zener is expected to be higher compared to the more normal 6.8-7 V compensated zeners.


Good information. The high current - over 20mA would also tend to dominate. And that might also be another argument to use better resistors for the main divider.

I can certainly extract drift characteristics over and above tempco effects - either by controlling conditions at the time of measurement, or statistically - by multivariate regression. But I am wondering if there is an issue that I would be measuring ptf56 drift rather than 2dw232 drift.

The current pcb board didn't pull the main feedback divider out to the pin-headers. I assumed that with the op-amp output and z-ref available then that value could be derived  - albeit with a small difference due to the op-amp input offset voltage. But that may have been an oversight.
 

Offline SvanGool

  • Regular Contributor
  • *
  • Posts: 135
  • Country: nl
Re: Ultra Low Noise Reference 2DW232, 2DW233, 2DW23x
« Reply #399 on: March 25, 2017, 12:14:47 am »
It's frustrating that everything ultimately comes down to (expensive) resistor choices rather than the $1 cost of the semiconductor. This seems to be universally true of voltage references regardless of device - microsemi zener, lm399, ltz1000.

That is funny, there is another thread going on which is tending in the opposite direction: https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg1168867/#msg1168867 and later. I think that you have to make the best out of the components you selected to use and Kleinstein gave some good advises on that path. Maybe you haven't squeezed your design yet to the limit, don't give up yet, I'm indeed already positive on your current results.
# Don't hurry, the past will wait. #
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf