Are cheap AD584 units worth it?

[leonerd]

I see you can get an AD584 built up in a nice case all ready for use, for peanuts cost these days. Eg.

  http://www.ebay.co.uk/itm/KKMOON-High-Precision-Voltage-Reference-Module-AD584kH-Single-button-Operation/381300503613

The AD584 offers 15ppm/degC, which I'm quite sure is adequate for my needs. But are these units actually worth it though, or would I be better off paying a little more for something nicer?

One concern I have with this unit is that it has outputs at 5V, 7.5V and 10V, yet the power input to the unit is only 5V. So to generate those voltages it's going to have to boost it up somehow. Looking carefully at the board, I believe I can see an inductor and a big diode in what looks like a classic boost-mode SMPS. Surely that's going to put out a lot of switching noise, if you want a stable voltage.

So would I be best to ignore these units and look for something that takes 12V in directly? Or maybe I could post-mod it, take out that switching unit and feed a stable (linear-regulated) 12V directly?

[acbern]

Depending on your expectations re. accuracy it may be enough. But keep in mind, this is a not temperature regulated device, with non-hermetic, not pre-aged components. You will have all kinds of drifts, easily a few 100ppm in a couple of months. You have (from what I can see), no traceable accuracy data of the device used to calibrate it. All, to be fair, not expectable at that price. So for a 3.5 or 4.5 digit DMM, it is probably enough.

[tszaboo]

They are selling it less than half the booking price of the reference itself?? Sure...
Anyway I dont think anything more than a 3-5 PPM/K reference voltage with pre-regulation is really worth it. I mean I have a REF5050A in an altoids box with a 9V battery, which was  run for a week, and plugged into a 3458A, that will be better than this.
I mean it is probably fine to calibrate a 8-10 bit ADC with its front end, but nothing more.

[mojoe]

I bought four of these just recently: http://www.ebay.com/itm/181863456673

These are the "L" version chip, which has better specs and is also hermetically sealed. It is no longer made. I suspect that these were salvaged from something, which may actually be a good thing. These are probably well aged. You don't get that funky 15V battery that is in the picture.

According to the spec sheet, you should power this chip with at least 15 V, although it will work at lower voltage. You have to look around for an older specsheet, as the new one doesn't list the "L" version.

I currently have the third of the four I bought hooked up to a bench supply and my HP 3457A. I let each one cook for a few days and monitor the voltage output at 10 V. The first two were better than spec and pretty stable.

One thing I noticed about the three tested so far is that the first 3-4 times you power it on, you get some random voltage that is lower than 10 V. I suspect that the chip is oscillating. I plan on checking this with a scope over the weekend. I noticed that there are more caps on these boards than the circuit in the specsheet. I understand that sometimes a capacitor on the output can make the chip oscillate. I haven't traced the circuit yet. Also slated to do over the weekend.

One amusing item - each board came with a small folded slip of paper with very precise voltages listed for each of the four output voltages. So, at first glance, it looks like someone went to the trouble to check each board. That's what you think. All four slips of paper had the exact same numbers printed on them. And of course, these numbers didn't match the actual output voltages. Sometimes I wonder, why do they bother with this type of deception, when it is so easy to disprove?

[Arhammon]

I bought four of these just recently: http://www.ebay.com/itm/181863456673

One amusing item - each board came with a small folded slip of paper with very precise voltages listed for each of the four output voltages. So, at first glance, it looks like someone went to the trouble to check each board. That's what you think. All four slips of paper had the exact same numbers printed on them. And of course, these numbers didn't match the actual output voltages. Sometimes I wonder, why do they bother with this type of deception, when it is so easy to disprove?

I bought this board from this seller six months ago. This board hasn’t come with paper. The seller said he did not understand what a paper, I want from him.  Apparently, a lot of people asked for a refund and he had to print fake papers.
I’m feeling is better to buy the cheapest AD584KH, like this:
https://xdevs.com/review/ad584_vref_test/

[Andreas]

These are the "L" version chip, which has better specs and is also hermetically sealed. It is no longer made. I suspect that these were salvaged from something, which may actually be a good thing. T
One amusing item - each board came with a small folded slip of paper with very precise voltages listed for each of the four output voltages. So, at first glance, it looks like someone went to the trouble to check each board. That's what you think. All four slips of paper had the exact same numbers printed on them. And of course, these numbers didn't match the actual output voltages. Sometimes I wonder, why do they bother with this type of deception, when it is so easy to disprove?

Hello,

up to now my opinion on these devices was:
You get at least some measurement values which are from a calibrated instrument.

Although for a bandgap reference you can usually expect drifts from 100-200uV over several months.

Lets summarize:
- in every case you get re-stamped devices
  (the sample of AD that I bought from local distributor has laser engravement and the PIN1 marker on bottom side of the marking)
  so it is easy to change a JH device to a LH device.

- you do not get necessarily measurement values of your device which would be valid at least some weeks.

So the worth of all is not more than the parts price.

With best regards

Andreas

[acbern]

I would not even bet these are not faked parts. The Fluke ref. and others have been faked as we have read here.

[Muttley Snickers]

I've also been looking into these modules, it seems that the AD584LH is still readily available on some of the cheaper basic yellow PCB modules on Ebay, most of the acrylic cased units that I have seen now have the AD584KH onboard unless you are willing to pay almost double the price for one with the AD584LH and then there is no guarantee as to what you are getting or how good a reference they really are.

Anyway Scullcom does an excellent video and comparison of a few units, certainly worth a look if you are interested.

   

[mojoe]

Looking at what I received, I don't think that these parts have been faked or re-labeled. They certainly look like they have been used before. In any case, in the short term (two days each), all four of mine exceed the voltage tolerance spec for the "L" part.

From the specsheet (only tested at 10 V so far):

Maximum Error for Nominal Outputs of: 10.000 V     AD584J = +/- 30 mV     AD584K = +/- 10 mV     AD584L = +/- 5 mV

The worst one of the four was 3.5 mV high. All four have been high.

The last one that I still have hooked up is only fluctuating about +/- 40 uV and that could be induced noise on my test leads. I need to make some shielded test leads.

These certainly aren't a high precision voltage reference, but should be good enough for checking the average hobbyist handheld DMM. I bought four of these because they were so inexpensive and to satisfy my curiosity. I don't need them, as I have a much higher precision and more versatile voltage reference that is currently out for calibration.

[mojoe]

Anyway Scullcom does an excellent video and comparison of a few units, certainly worth a look if you are interested.

Yes, I watched that a few nights ago. The first one is essentially the same as what I linked to. The last one was heavily discussed here some time ago. A grossly overpriced unit made by someone who thinks he has some secret knowledge that makes his POS better than it should be (IMO). If he is reading this and takes offense to my OPINION, suck it up. I won't get into an argument with you. And please don't quote the bible at me.

[hydrogen maser]

I have one of these and it looks exactly what the OP linked to. How good is it? No idea, I bought it because it was cheap and looked interesting. Mine has a sticker on the back that says "Hp 3458A" and then lists what was "measured" for each voltage setting. The number next to each voltage is in ink pen and goes out to 5 decimal places. Would I trust this thing for anything critical? Nope but it is OK to see if a multi-meter is way off.

[mojoe]

Just curious, what numbers did you get on that slip of paper? Here is what I got (printed) with all four.

2.5V     2.49942V
5V     5.00037V
7.5V     7.50042V
10V     10.00066V

Same meter, temperature and date on all four. Just amazing how all four of them put out the exact same voltages 

[hydrogen maser]

Sticky label on the back written in ink pen:

2.5v: 2.49994
5.0v: 4.99888
7.5v: 7.49776
 10v: 9.99581

[mij59]

In 2014 I bought the acrylic case version.

Printed sticker  on the back,  year 2014,  temperature 21 C, using a Agilent 34401A, with  written measurements.

Done some measurements with the KS 34465A and the Fluke 8846A, both units powered up for at least 1 hour, voltage reference was powered up for 1/2 hour before taken the measurements, ambient temperature 21,1 C according to the Fluke 87V  with thermocouple temperature probe.   

range   on sticker     KS 34465A   Fluke 8846A

2.5V     2.50070       2.50068          2.50070
5.0V     5.00035       5.00035          5.00037
7.5V     7.50418       7.50418          7.50421
10.0V  10.00285     10.00286        10.00292

KS 34465A bought in 2014, Fluke 8846A last calibration date 2008

[tszaboo]

Same meter, temperature and date on all four. Just amazing how all four of them put out the exact same voltages 

Well that preatty much sums up what I think about these references. One should know that calibration is some degree trust and some degree technical aspect.
How much do you trust a random chinese ebay seller? (with free shipping, sells like hot cookies)

[Muttley Snickers]

I've been meaning to order a couple of these modules for a few weeks and today bit the bullet on one of each, so I should end up with both a AD584LH and AD584KH reference to have a play around with.

I did PM another member for a link to the Ebay site where they got theirs from but that particular seller was not listed with Ebay AU so I simply chose a couple of random sellers that had the most colourful negative feedback, a bit of a worry really.

[tronde]

Just curious, what numbers did you get on that slip of paper? Here is what I got (printed) with all four.

2.5V     2.49942V
5V     5.00037V
7.5V     7.50042V
10V     10.00066V

Same meter, temperature and date on all four. Just amazing how all four of them put out the exact same voltages 

You are not alone...

[Harfner]

I bought the acrylic version in March 2015 from seller diyelecmall108. It has a sticker on the back, date 2015, temp 21°C, voltages are written by hand.

 2.5V     2.5077
 5.0V     5.00147
 7.5V     7.50841
10.0V  10.00472

On arrival, I checked the 10V against another 10V standard and found it within spec of the multimeter.

I bought the LH version shown by Muttley Snickers from 2012moon816 in June 2015. It has a printed sheet dated 2015-6-23 looking very similar to one shown by tronde.
 2.5V     2.500,21
 5.0V     4.999,20
 7.5V     7.500,01
10.0V    9.998,19

[Mark Hennessy]

I recently bought a number of these, and have written about them (including schematics and measurements) here: http://www.markhennessy.co.uk/ad584_references/

The ICs are reclaimed, and therefore nicely aged (hopefully!).

The KKMOON version is good, and comes with believable calibration data. Some switching noise. There is a battery management IC controlling a dual MOS-FET, but a pair of deliberate solder blobs are shorting out the MOS-FETs (same on others that I've seen using the "2015" PCB).

My "L" version also came with the same fake cal certificate. Glad to see a decent number of folks on here have exactly the same numbers I do - along with a couple of other people from another forum. What date codes do you all have on the IC? In all the close-up photos I've seen, I've only ever seen "1015". Could they be re-printing (hence not genuinely "L"-spec), or did they simply stumble across a massive batch somewhere? The ICs are clean and shiny, so perhaps could be NOS rather than reclaimed.

Finally, I picked up a tiny bare-bones PCB unit that uses the "J"-spec IC (though it comfortably meets "L"-spec if my measurements are reliable). Useful for building into a project.

All units meet their spec, and are great for the money.

The start-up problem is caused by the capacitor surrounding the IC (pins 6 and 7). There's no external sign of oscillation that I could find when investigating. Reduce it to 10n to reduce the time spent in limbo. For testing, set the output to 10V (remove jumpers), and then briefly short the 2.5V pins - that reliably provokes it. I haven't noticed this behaviour with the KKMOON unit.

[Muttley Snickers]

Thanks Harfner for the reference figures and welcome to the forum.....

I'm not expecting too much out of these modules and will probably be lucky if they even turn up, if they arrive then I will conduct some tests over time and post my findings, occasionally I see test equipment for sale locally and just want a small reliable reference that hopefully will give me a guide.   

Many thanks.....

[Muttley Snickers]

I recently bought a number of these, and have written about them (including schematics and measurements) here: http://www.markhennessy.co.uk/ad584_references/

Wow, Mark that really is the benchmark guide to all of these modules and a brilliant well documented piece, I have printed the entire page and saved all of the associated links, thank you sir very much......

[dannyf]

Scullcom does an excellent video and comparison of a few units,

He got his for 9 pounds, right?

Mine produced the following measurements:

13.0 - 36v, 12.3882v

11.0 - 32v,  9.9011v

 8.0 - 28v,  7.4251v

 6.0 - 26v,  4.9524v

 3.3 - 24v,  2.4732v
=========    =======
  Vin --GND-- Vout

However, I only spent a few cents on some TL431A, , in serial and with 1K resistors.

Not bad, right?

[zlymex]

I recently bought a number of these, and have written about them (including schematics and measurements) here: http://www.markhennessy.co.uk/ad584_references/

The ICs are reclaimed, and therefore nicely aged (hopefully!).
...............

Very nice review.
Yes, there are many reclaimed IC in China market, and pretended to be new. There are groups of people specializing in doing so including re-tin, re-print and even leads extension for metal can. When I buy ICs here, I either choose from a reputable market with high price(but genuine) or reclaimed with original leads and original marking. I'm now in possession  of more than 50 reclaimed LTZ1000, all tested good.

[Kjelt]

I recently bought a number of these, and have written about them (including schematics and measurements) here: http://www.markhennessy.co.uk/ad584_references/ 

Thanks! I ordered the KKMoon (#1) today and another cheaper one (#2) with the better L version, going to swap them and keep the kkmoon with a l version, lets see what we get.

[Mark Hennessy]

Thank you all for the positive feedback 

And thanks to zlymex for the insights to the Chinese market place. And I can't wait to see how you will use all those LTZ1000s - you could build the ultimate statistical reference with those 

I've got a couple more articles like that in preparation - just waiting on eBay deliveries