Calibratory D-105 DC Precision Voltage Reference Standard

[Terabyte2007]

so after reading a very hard to read thread ...

can more than 1 person who is still using his 10v ref give a updated "picture" of/about its stability/drift ?  ... (or did he really got banned just because he could not speak tech?)

i am interested to know if it is still a viable unit, since there is no other premade/pre-cal units made or on offer @ this price. or is there a better option @ USD100 ?

I will upload some long term data shortly. I purchased mine about the time this thread started and have been doing some long-term tests. I will try and post an update later today. I do have a full-time job and to find time to tie up my equipment for long term testing is challenging to say the least.

[Excavatoree]

That argument makes no sense.  He's making so much money per hour he can't spend time to lay the components out neatly, nor build it in a workmanlike manner but he can't afford to hire someone to design a PCB, get them made, and even assemble them?   

If he's turning down thousands per hour to build these, he's doing it for rewards other than money.  One would think these would include wanting to at least make a product that doesn't look like complete crap.   He should be smart enough to at least twist the strands together.

[JohnnyBerg]

I will upload some long term data shortly. I purchased mine about the time this thread started and have been doing some long-term tests.

I did a teardown of this device. The reference IC is a REF02P (with Bur Brown logo) in a socket!!

I removed the copper foil, to have a look temperature compensation. I have my doubts about that.
Besides that, drift is specified +/-, what if the compensation works(if it works) in the + direction, and the device drifts in the - direction?

Why do you think this device can beat the specs in the datasheet?

[Terabyte2007]

That argument makes no sense.  He's making so much money per hour he can't spend time to lay the components out neatly, nor build it in a workmanlike manner but he can't afford to hire someone to design a PCB, get them made, and even assemble them?   

If he's turning down thousands per hour to build these, he's doing it for rewards other than money.  One would think these would include wanting to at least make a product that doesn't look like complete crap.   He should be smart enough to at least twist the strands together.

I highly doubt he is making thousands of dollars an hour. His eBay "Sold" counter places the units sold in an auction at 30-40 sold or something like that. Regardless, it was quite apparent from some of his posts that there might have been a psychological problem involved. For all we know he may have some handicap issues either mentally or physically. Before any gets bent, I am saying this as a 3rd party observer to all the posts by Awesome14. I could be wrong, but that's what all of this lead me to possibly believe. It really comes down to taking constructive criticism as a positive instead of a negative. It's a way to improve your skills and you as a person. That was not the case.

[Terabyte2007]

I will upload some long term data shortly. I purchased mine about the time this thread started and have been doing some long-term tests.

Why do you think this device can beat the specs in the datasheet?

I never said that it could. I am only following through with the tests I stated a while back that I would do. I believe I have mentioned in previous posts that I felt it would not meet the seller's advertised specifications. Instead of speculation and attacks, I am merely trying to prove or disprove to best of my ability the claimed accuracy and stability of a product.

[JohnnyBerg]

Why do you think this device can beat the specs in the datasheet?

I never said that it could. I am only following through with the tests I stated a while back that I would do. I believe I have mentioned in previous posts that I felt it would not meet the seller's advertised specifications. Instead of speculation and attacks, I am merely trying to prove or disprove to best of my ability the claimed accuracy and stability of a product.

Ok, fair enough.

IMHO what you get is printed on the IC itself. It is a Burr Brown REF102P

Tried to find a datasheet for this IC, but for me it is unclear what the precise specs would be.




note: In September 2000, Texas Instruments acquired BB. What does that tell us about the IC, that has still the BB logo on it?

[Terabyte2007]

note: In September 2000, Texas Instruments acquired BB. What does that tell us about the IC, that has still the BB logo on it?

Interesting... I can't make it out from the picture but can you find a date code on that IC?

[paulie]

Why do you think this device can beat the specs in the datasheet?

Well certainly not anymore.

Even ignoring the obvious rampant bias and prejudice it's unlikely anyone here is in a position to verify if it even meets let alone beats any spec. Once you dismantle or even open the device this becomes impossible. The designer and seller stated this clearly and no one here has any right to dispute that. Of course many will try.

My early prediction that no valid performance conclusions would arise from this thread so far appears true. No matter positive or negative once you violate user instructions all bets off.

And try reading the posts once in a while. Awesome never made any such statement about thousands per hour. I was sarcastically highlighting how badly he has been misquoted and words simply fabricated. It should have been obvious from my facetious addendum:

See... you guys aren't the only ones that can put words in his mouth.

ps. People being human will however see what they want to see rather than what is.

[JohnnyBerg]

note: In September 2000, Texas Instruments acquired BB. What does that tell us about the IC, that has still the BB logo on it?

Interesting... I can't make it out from the picture but can you find a date code on that IC?

Nope, the only other code on the IC is OBW405B

[JohnnyBerg]

OLD ... and WELL aged ? which could mean ... good?

Hmm .. references only age when a voltage is applied ..

[JohnnyBerg]

E14 have them about USD12 each ... hmmmm

Have a closer look 

There are different versions .. I assume that the REF102P (BB type) is now REF102AP (TI type)
When buying 10, they are € 3,67 (~ $4,-).
Look at the T.C. 



We can only guess where the BB types came from ..

[nctnico]

I would only rely on what is specified in the datasheet.

[JohnnyBerg]

I would only rely on what is specified in the datasheet.

+100 

[JohnnyBerg]

the datasheet is not even the actual working item

Perhaps there is someone out there, with an original Burr Brown datasheet?
In a book, on paper? 

And as it is a plastic IC, it has attracted some moisture over the years. What will that do with stability??

[6thimage]

the datasheet is not even the actual working item

Perhaps there is someone out there, with an original Burr Brown datasheet?
In a book, on paper? 

The only thing I can find online is http://www.pci-card.com/ref102.pdf. It still refers to the 3 different grades, but also uses R, S and M, for A, B and C (respectively). But these were for TO-99 cans, the plastic DIP is either A or B.

[JohnnyBerg]

inside the TI datasheet, there is a version with 3 outputs into 1 (fig 14) with price of 10 less than USD100 ... now that should make someone a really really good stable reference i bet?

**edit ... its over USD100 for 10 here. but its a diff model

Define good and stable? 

IMHO "good" cannot be quantized.

I used the REF5025A a lot, that has a typ. TC of 3ppm/°C after 1000 hrs. Thats gives a very stable reading on a 6,5 digit DMM (within 1~2 count typical). I suspect it is better then the REF102P

n.b. figure 14 says that it has extremly low noise. We did not talk about noise .. yet

My fellow dutchman and forum member  Blackdog did a lot of research on references in parallel.

[Excavatoree]

And try reading the posts once in a while. Awesome never made any such statement about thousands per hour. I was sarcastically highlighting how badly he has been misquoted and words simply fabricated.

I merely explained how that argument is illogical.  I never claimed he or anyone else had made it. 

[cncjerry]

The Geller you guys keep comparing him with, is that Uri?

[Joe Geller]

It's a relatively common name; no relation that I am aware of. -Joe Geller

[drakke]

I don't want to hijack a thread but it seems to have degraded already so ...

are these sub-$100 voltage calibrators good value for hobbyist/student use?

I see a lot of them on Ebay.

[Joe Geller]

Adding a thermistor and cheap voltage trim components [i.e. carbon-film resistors, and cheap pots] does not [magically or otherwise] improve the specifications of *ANY* monolithic voltage reference.  While you may be able to improve the TC and final voltage initially, these trim settings cannot hold due to the cheap components used.  [And there would still be a very small drift if higher performance and higher cost components were used].

Using an epoxy packaged monolithic voltage reference has well known "issues" with temperature excursions and humidity changes.  Putting all of this into a plastic box and "sealing" this [with whatever] does not [magically or otherwise] make these "issues" go away.  Water vapor is one of the worst things to deal with, and it can infuse through almost all plastics excepting some fluorocarbons.  This would include the epoxy used to encapsulate the IC and the plastic material that was used to injection mold the enclosure.  These are like "sieves" to water vapor.

Yes, "aged" means: power-on time [for the Zener] and time [at operating temperature] for the internal and external intentional and unintentional [parasitic] resistors.  Only *some* of this aging can be accelerated by baking a high temperatures for a long time, and then following that with [progressively smaller] temperature cycles.  The rest of the aging is due to the Zener current, and with a monolithic voltage reference this can only be accomplished with a long power-on period [months to years].

So yes, the performance of *ANY* monolithic reference begins with the data sheet, then becomes worse as you add more components to try to "fix" the shortcomings of the reference.  If you use cheap components to do this, then expect even a less pleasant outcome.  Also, not all voltage reference IC's are them same.  Even if they are from the same fabrication wafer-- each reference can have vastly different drift characteristics that no amount of aging can modify.  So, out of a given batch of reference IC's, a few of them will have very good drift characteristics and some of them will have much worse drift, while the majority of them will be close to data-sheet specifications.

The SVR program was always designed as an inexpensive way to transfer a calibration point.  It was never intended to be a  FLUKE 732B substitute for under $100.

Each SVR-T board was studied for hours with small temperature variations to determine sensitivity for that particular chip.  Relatively high quality IRC trim resistors (not $25 each Vishay resistors) and a relatively high quality Vishay thermistor were used for the correction, which was a small vernier made by the trim circuit (not a 1:1 correlation between trim drift and output voltage change). (The series resistor with the thermistor was standard 1% metal film, as it was a small correction on the trim input, already a small part of the output adjust.)

Burn in was just run time to work out past larger 200 hour change exhibited by many of the series.

Following the correction, each board was tested for 7+ more hours to verify the correction.  Comparison was made by Null measurement with a 34410A on a mV scale to the FLUKE 732B.  Many sets of data show good short term performance, which was the goal of the transfer program. 

[Joe Geller]

It is disappointing that the SVR program attracted such strong reaction.  I had the idea for an inexpensive transfer reference around 2003 and bought the FLUKE 732B back then to try make it work.  I learned later that there had been similar efforts with LM399 over the years, but as far as I know, no low cost boards for hobbyists and engineers.

My idea was a small board similar to the "International Crystal" small board kits from the 70's that old timers might recall.

There have been any number of similar approaches from boards to small boxes, at least from a couple of years after I began to sell AD587 based references on eBay. 

One of the most difficult aspects of this sort of product is the wide variety of what users are looking for.  For example, I got several complaints from Simpson 260 owners that turning the trim had no effect on the output voltage.  Others, as discussed at great length here wanted it to be a long term stable voltage standard.  Ironically, many of them have done quite well over years, but my post is not to debate that point.

I wish you all great luck in your own designs.  I see new design ideas all the time.  It is very likely that a great long term stable reference may emerge.  Possibly for under $100.  Great fun!

[Galaxyrise]

The AD587JQ and it's "better" brother the AD587UQ are the only monolithic 10V references that I have been able to find that are still sold in hermetic packages. 

Linear has a few LT1021 flavors in metal can still. The 10V is hard to find in small quantities, but it'd be cheaper to stack two 5V anyway since the 10V is an "M" flavor.  If you're willing to stack two, the LT1236 and LTC6655 come in LS8.  I've been able to compensate the TC of the 1021 (presumably it would work for the 1236 as well) to nearly zero.

[timb]

Apparently not banned but more likely just hiding out. And all reviews of his product so far indicate performance somewhat better than advertised. However this has not deterred detractors.

No, short term performance is *on par* with the datasheet.

Just because you plug it into your meter and it reads 5.0000V does not mean it's a good standard.

Nobody (including you and the seller) knows what it will read in six months or sixty days or what happens if you're in Florida in July with a 90% RH.

Until we have that data the only thing we can do is estimate performance based on similar devices, the construction, grade of components, etc., which is exactly what we've been doing. We're not detractors, we're real world engineers.

I don't know why you're White Knighting this dude so hard, but it's getting sort of old.


Sent from my Tablet

[HighVoltage]

The SVR program was always designed as an inexpensive way to transfer a calibration point.  It was never intended to be a  FLUKE 732B substitute for under $100.

Each SVR-T board was studied for hours with small temperature variations to determine sensitivity for that particular chip.  Relatively high quality IRC trim resistors (not $25 each Vishay resistors) and a relatively high quality Vishay thermistor were used for the correction, which was a small vernier made by the trim circuit (not a 1:1 correlation between trim drift and output voltage change). (The series resistor with the thermistor was standard 1% metal film, as it was a small correction on the trim input, already a small part of the output adjust.)

Burn in was just run time to work out past larger 200 hour change exhibited by many of the series.

Following the correction, each board was tested for 7+ more hours to verify the correction.  Comparison was made by Null measurement with a 34410A on a mV scale to the FLUKE 732B.  Many sets of data show good short term performance, which was the goal of the transfer program.

Your SVR-T has worked very well for me. I take it to a local professional calibration lab and get a certificate of calibration. At the same time I get my Agilent 34410A calibrated and take both items to my client and use it to calibrate his equipment to 51/2 digits. So far, this procedure and the results have passe every audit. Thank you for a great transfer reference.