Are cheap AD584 units worth it?

[zlymex]

- you could build the ultimate statistical reference with those 

I certainly could, with total of more than 100 LTZ1000(plus 40+ brand new).  The rough plan is ten times ten.

[bitseeker]

I'm now in possession  of more than 50 reclaimed LTZ1000, all tested good.

Wow, zlymex. Are you building references for us?

[zlymex]

I'm now in possession  of more than 50 reclaimed LTZ1000, all tested good.

Wow, zlymex. Are you building references for us?

Well, building 100+ refs all in their own box is a hell of a work for me.
The original idea was to build a super statistical one. My current plans is two level statistical with just only 11 boxes.

[Theboel]

Someone is very serious about precision 

[bitseeker]

The original idea was to build a super statistical one. My current plans is two level statistical with just only 11 boxes.

That sounds very cool. I'm looking forward to your thread on it.

[Andreas]

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

Hello,

thanks for the article.

some annotations:
You are in doubt wether the plastic case is equal in specs to the Metal CAN housing.
Usually it is not. With a plastic package you can measure up to 15 ppm output voltage change
 for humidity changes between 40 to 70% rh. (Depending on device).

And usually the (best) ageing spec is only for hermetically sealed device.
In most datasheets this is mentioned partly as a footnote.
Even if the hermetically device is already obsolete: see LT1027, REF102.
For the AD584 the footnote is not there (because this fact is evident or does it really play no role?)

The AD584 has a relative large input voltage dependency. (20-50 ppm/V)
So if you use a battery (non stabilized) you should not only record the temperature but also the supply voltage.

The microprocessor could be a Holtek HT68F002  (best guess).
There are not many processors with supply on pin 1 + 8.
(and since pin 4 is a output the pic12F508/627/683 devices do not fit)

With best regards

Andreas

[Mark Hennessy]

Hi Andreas,

Thanks for the feedback - you raise some good points

You are in doubt wether the plastic case is equal in specs to the Metal CAN housing.
Usually it is not. With a plastic package you can measure up to 15 ppm output voltage change
 for humidity changes between 40 to 70% rh. (Depending on device).

And usually the (best) ageing spec is only for hermetically sealed device.
In most datasheets this is mentioned partly as a footnote.
Even if the hermetically device is already obsolete: see LT1027, REF102.
For the AD584 the footnote is not there (because this fact is evident or does it really play no role?)

Yes, I'm certainly aware that humidity can play a part with plastic cases. And because I live in an old house (where the workshop is in the cellar), I've had to learn a lot about the subject - and invest in a dehumidifier

However, unless I've missed it, the datasheet doesn't differentiate between "H" and "N" in any parameter - as a result, I thought it was safest to not to speculate about it here.

The AD584 has a relative large input voltage dependency. (20-50 ppm/V)
So if you use a battery (non stabilized) you should not only record the temperature but also the supply voltage.

I did quote it (15.00V).

I agree that the line regulation is specified at 20-50ppm worst-case, but I found it was somewhat better than that. Admittedly, I didn't test it exhaustively, and in fact I think I only checked the 2nd unit, but I will double-check this aspect soon.

In the meantime, here's what I recorded here: http://www.markhennessy.co.uk/ad584_references/#unit2_summary

A 5.5V drop in supply voltage (18 to 12.5V) caused a 200uV change in the 10V output voltage, which is 20ppm overall, or 3.6ppm/V. Surprisingly good.

The microprocessor could be a Holtek HT68F002  (best guess).
There are not many processors with supply on pin 1 + 8.
(and since pin 4 is a output the pic12F508/627/683 devices do not fit)

Thanks - that's useful insight

BTW, I'm playing around with an AD588 at the moment

All the best,

Mark

[zlymex]

BTW, I'm playing around with an AD588 at the moment

I have two AD588 but I don't like them at all. The spec for noise and drift is just average, it need dual supply and draw too much current which means the self-heat will add thermal related problems such as EMF and low frequency noise.

[Mark Hennessy]

BTW, I'm playing around with an AD588 at the moment

I have two AD588 but I don't like them at all. The spec for noise and drift is just average, it need dual supply and draw too much current which means the self-heat will add thermal related problems such as EMF and low frequency noise.

This is part of my survey of cheap Chinese voltage references - I have one from KKMOON which seems quite nice so far, and am awaiting delivery of another type. Links here:

http://www.ebay.co.uk/itm/172024077034
http://www.ebay.co.uk/itm/171927104601

As before, I'll put up some photos, schematics, basic measurements, etc.

I'm not chasing the ultimate reference here - that's another project - rather, this investigation started as a result of a forum conversation elsewhere. Most people who were interested in these eBay units are using 3.5 or 4.5 digit meters, so for them, the AD584 modules are probably fine.

I haven't had a chance to fully study the AD588 datasheet yet, but it certainly seems like a nice step up from the AD584 in terms of basic initial accuracy and tempco. Line regulation looks worse at a glance. Yes, the power supply requirements seem like a PITA. But against that, the extra op-amps in there might be useful for making a precision current source or something similar - sort-of like a "super-LM10", if you see what I mean 

Cheers,

Mark

[pelule]

@ mojoe
You don't get that funky 15V battery that is in the picture.

I regular order that 15V batteries in China since several year. Never had any issue yet.
http://www.aliexpress.com/item/free-ship-15v-10F20-primary-battery-dry-battery-multi-meter-heavy-duty-battery/923034386.html

PeLuLe

[Andreas]

A 5.5V drop in supply voltage (18 to 12.5V) caused a 200uV change in the 10V output voltage, which is 20ppm overall, or 3.6ppm/V. Surprisingly good.

Hello,

for a volt-nut there are better references:
A LT1027CCN8-5 I have measured around 14uV (<3 ppm) over 8.5 - 15.5 V. (0.4 ppm/V)
(3-8 ppm/V in data sheet).

Other references like LT1236AILS8-5 are even better (around 0.5-1ppm over 8-15V).

But of course the measurements done are not the same as stated in the datasheet.
In the datasheet they use pulsed measurements with no self heating effects.
Whereas I use constant voltage over longer times. (reflecting more the reality).

With best regards

Andreas

[Muttley Snickers]

My AD584KH unit arrived yesterday nicely packaged in bubble wrap and the first thing I did was to take a few measurements and document the readings, I then hooked the unit up to a power supply at 5 volts to see what the battery state was at and the initial current draw was around 300 mA which rapidly diminished over about ten minutes to not much at all.

I then opened the unit up to insert a clear insulating shim between the battery and the PCB as others have done and the battery in this particular one appears to be rather crappy and slightly swollen, the unit initially took quite a while to settle on a final value in all selections enough for me to have serious concerns about the product but I found after about five hours of continuous testing with various meters that the module now settles in only a few seconds and is relatively consistent.

The nominated values on the label appear to be correct and a few of the meters that I tested with were in agreement with the label, certainly well within the specifications of both the module and the meters used. I noticed when putting in the shim that there was a fair bit of flux residue on the board and opened the unit back up to clean it off and during the cleaning process the negative lead for the battery came away with not much effort at all.

I am now waiting on the AD584LH module to arrive and hope to be able to take both units up to a fellow members place to run these things on some better gear than I have here.

The Ebay seller I got the unit from.
http://www.ebay.com.au/itm/High-Precision-4-Channel-2-5V-7-5V-5V-10V-Voltage-Reference-Module-AD584KH-/181873998781

The website that is on the screen print.
https://shop58034511.world.taobao.com/

[Kjelt]

My AD584KH unit arrived yesterday nicely packaged in bubble wrap and the first thing I did was to take a few measurements

Mine arrived also wednesday, unbelievable quick from china to europe within a week, but I think since it was a german package it got delivered from a european warehouse or so.
My seller on ebay was tomtop_sales.
I took some measurements and they were spoton. I do not recognize the unstable behavior mine was very stable but I did not charge the battery, could that have something to do with it?
Anyway thanks for the shim advice will do that this weekend.

[Muttley Snickers]

No this unit was definitely unstable regardless of the charge state and was still drifting dramatically by tens of millivolts for about an hour with a stable 21 degrees, after a few power cycles the unit seemed to stabilise and is now pretty well consistent, I also hadn't heard of anyone else having this problem and intended to give it some time before contacting the seller, as it is I am very happy but again time will tell.

Also I forgot to mention that I added another piece of clear insulating sheet to the top of the board to stop inserted banana plugs shorting anything out which others previously mentioned, this is a good quick fix and saves modifying the output jacks.

[Kjelt]

Ok so now I have seen that my benchmeters and multimeter are pretty much within hobbieist level of calibration, what are you all going to do with these things ? 

[bitseeker]

Excellent question, Kjelt, and one that has kept me on the fence about getting devices like these or the more finished voltage/current standards. I suppose that if I had a lot of equipment and standards that spanned all the required ranges, I could have the standards calibrated and then calibrate everything else accordingly.

[Muttley Snickers]

The AD584LH module that I ordered a few weeks ago arrived today and I was just about to complain to the seller in regards to the delay after receiving an item sent email weeks ago, anyway it was really poorly packaged and was not supplied with any documentation whatsoever yet the Ebay listing indicates that it would be supplied with values.

It was around eight bucks so I'm not going to whinge about it and just glad that it turned up at all, I will however notify the seller to amend the listing so others who may be reliant on these values aren't disappointed, the unit is currently on the bench being tested and so far it seems to work.



   

[Mark Hennessy]

Perhaps he's decided to stop giving out the fake certificates? No certificate is better than a fictitious one. But yes, I agree the listing should be amended - as it stands, it's "not as described", and plenty of buyers would complain about that.

What's the date code on your IC?

[Muttley Snickers]

The date code on this AD584LH module is 1015 and looks exactly like the one in the picture that I posted earlier in the thread, I am hopefully going  to have a play around with it tomorrow and look at getting rid of those jumper links which are a bit of a pain, perhaps a decent rotary switch and a proper box might be in order for the interim.

[guy48065]

What can be done to boost the current capability without sacrificing the excellent power and load regulation? I'd like to use mine as a precision 10V excitation source, with about 100mA output at a minimum. The AD584 is limited to 10mA.

[Kleinstein]

The AD584 does not have separate voltage drive and sense lines. So all one can do is adding an low offset amplifier. This could be something like an OP07 or similar with a transistor to do most of the power driving.

[Andreas]

What can be done to boost the current capability without sacrificing the excellent power and load regulation?

Hello,

I do not find that 20-50ppm/mA are a excellent value (compared to the LT1027 with 3-10ppm/mA).

In AN86 from Linear Technology there is a cirquit with a LTC1152 + LT1010 as buffer which can deliver the 100mA.

with best regards

Andreas

[Mark Hennessy]

What can be done to boost the current capability without sacrificing the excellent power and load regulation?

I do not find that 20-50ppm/mA are a excellent value (compared to the LT1027 with 3-10ppm/mA).

But remember that it's all relative.

Compared to the alternatives at his disposal, and considering his needs, it might well be excellent. Certainly, most bench power supplies drift rather more - but they are perfectly fit for purpose...

Now that I've acquired a handful of these things and am wondering what to do with them, building a "precision" bench power supply around them is one option. It won't be as good as the LT1027, but it'll be better than the usual grade units. Although I'm thinking that I'd be better off using the AD588 (better specs than the AD584, and sense lines, of course). It might be as simple as adding pass transistors and making use of the internal op-amps - but I haven't given that serious thought yet.

[Kleinstein]

For delivering more power an external power stage like that with a LTC1052 and LT1010 (or just an transistor) it not that bad. Especially with a AZ Op, it can very well compensate for load changes.

[Kjelt]

In the meanwhile I have received both AD584 units, the one in the plexiglass box a AD584K is marked with some kind of mill, datecode 9210 so week 10 1992 which seems legit.
The other one is the AD584L it looks brand new, the markings are lasered (you can see the individual dots under a microscope) and has a datecode 1015 so week 15 2010.
The big question is, did AD make these still in 2010 and were they indeed lasered or is the last one a fake chip? Anyone that can give a decisive answer on this one?