Are cheap AD584 units worth it?

[Muxr]

There is also this AD588 board.. http://www.ebay.com/itm/111840217453?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

It appears to have better tempco than the best AD584 reference I have.

Lets see....

AD588BQ:
Digikey - $49 Qty 100
Mouser - $47 Qty 100

Completed board eBay from China - $14 Qty 1

Yah... instills trust...Not.

Same goes for the AD584 based references on eBay. We know they are all salvaged.

[Bud]

Ok, but then we can't assume these tests characterise the manufacturer part, these tests only characterise a particular salvaged part, who knows how much abuse it handled, if it was a production reject, a test part, or still can be a fake.

[alm]

Agreed. The AD584 units had a fair amount of variation in the amount of noise, and something like that could easily be the case for the AD588 boards. But the only way to tell is to gather data from a large number of units bought over a time span of at least a couple of months.

Of course depending on how they were treated, they might just be well aged references .

[Muxr]

Ok, but then we can't assume these tests characterise the manufacturer part, these tests only characterise a particular salvaged part, who knows how much abuse it handled, if it was a production reject, a test part, or still can be a fake.

There is no free lunch. As you put it the brand new part alone costs way more (in both cases AD584 brand new off mouser is $60 just for the part as well). So there is definitely a catch to these ~$15 references from China.

I have 3 ebay AD584 references and one AD588BQ one. And so far AD588BQ sample I have is an improvement over all the AD584's I've tested (at least as far as tempco and noise is concerned (except for occasional popcorn noise), haven't had it long enough to establish long term drift).

But everything you said stands. It's just my samples, the parts are of questionable sources. I am just sharing my data. Your mileage may vary.

[Marco]

The only thing is, it requires both a +17V and a -17V rail.

With 16V rated capacitors

[Muxr]

The only thing is, it requires both a +17V and a -17V rail.

With 16V rated capacitors

oops, that's actually my bad. I was thinking of my own prototype reference, the one from China does have 16V rated caps but it requires +15V/-15V.

[Gyro]

Ok, but then we can't assume these tests characterise the manufacturer part, these tests only characterise a particular salvaged part, who knows how much abuse it handled, if it was a production reject, a test part, or still can be a fake.

There is no free lunch. As you put it the brand new part alone costs way more (in both cases AD584 brand new off mouser is $60 just for the part as well). So there is definitely a catch to these ~$15 references from China.

Ouch! That makes me feel a bit guilty about the tube of 6 unused AD588BQs I scored on ebay a couple of years back for £9.50 including postage. It was genuine UK seller who clears out old production line stores. I think we were both rather surprised when no one else bid! There is still the odd legitimate bargain out there if you keep your eyes open. You guys in the US seem to have it easier in that respect.

[Marco]

oops, that's actually my bad. I was thinking of my own prototype reference, the one from China does have 16V rated caps but it requires +15V/-15V.

But one of the set of caps is in front of the linear regulators, with room for the drop out you'll exceed the 16V.

[Muxr]

oops, that's actually my bad. I was thinking of my own prototype reference, the one from China does have 16V rated caps but it requires +15V/-15V.

But one of the set of caps is in front of the linear regulators, with room for the drop out you'll exceed the 16V.

The datasheet for the AD588 says it can operate with rails down to +/-12V.



Table 3 even references a test with 10.8V +/- rails.

With +15 and -15 supply rails I measured 12.95V (so the drop out is 2V) on the lower rail (at the chip) so it's fine, however it is pretty dumb that they used 16V rated caps pre regulation 

edit: I should also add. The caps are after the rectifier and the rectifier has about 0.7V drop as well. So you could technically supply 16V-16.7V +/- rails to the board without operating beyond the cap spec.

[jordanp123]

For anyone who has one of these units how have these "Recycled" units been holding up?

[leonerd]

For anyone who has one of these units how have these "Recycled" units been holding up?

I'm just about to get mine re-measured by a cal lab, so I'll let you know when I get a result.

I just hope they don't laugh too much at the look of the quality of the thing. In actuality it isn't too bad inside, it just looks cheap.

[CJay]

Ouch! That makes me feel a bit guilty about the tube of 6 unused AD588BQs I scored on ebay a couple of years back for £9.50 including postage. It was genuine UK seller who clears out old production line stores. I think we were both rather surprised when no one else bid! There is still the odd legitimate bargain out there if you keep your eyes open. You guys in the US seem to have it easier in that respect.

I'm assuming you have plans for those chips and couldn't be persuaded to part with one?

[innkeeper]

I'd love to do some casual comparisons with other references with my kkmoon.
not sure id shell out the bucks to have a cal lab check it., but be nice to cross reference to see how close they are.
i hit ham fests in the northeast USA a lot if anyone else hits them.

[BU508A]

Ouch! That makes me feel a bit guilty about the tube of 6 unused AD588BQs I scored on ebay a couple of years back for £9.50 including postage.

I bought yesterday a tube of 14 AD588BD for around U$ 63,- including shipping, taxes, customs etc. Date code is 9239.
http://www.ebay.com/itm/142510140268

Edit: URL changed from .de to .com

[Gyro]

Ouch! That makes me feel a bit guilty about the tube of 6 unused AD588BQs I scored on ebay a couple of years back for £9.50 including postage. It was genuine UK seller who clears out old production line stores. I think we were both rather surprised when no one else bid! There is still the odd legitimate bargain out there if you keep your eyes open. You guys in the US seem to have it easier in that respect.

I'm assuming you have plans for those chips and couldn't be persuaded to part with one?

Yes, mostly allocated (my son wants a couple) but I can probably spare one. PM me your address and I'll stick one in the post. 

Damn, BU508A scored better spec ones!


P.S. If anyone is looking, there appear to be some AD588BQ spec parts available from Germany at what looks like a good price. No idea about the provenance but possibly less likely re-marks from China?  http://www.ebay.co.uk/itm/AD588BQ-Very-High-Precisons-Reference-Voltage-IC-TK-1-5ppm-C-/182764566859?epid=1269035643&hash=item2a8d9df94b:g:Dy8AAOSwi8VZUkGL

[BU508A]

Damn, BU508A scored better spec ones!

Aha? I thought, the BD and the BQ have the same specifications? I did not spot any difference between them in the datasheet.
My impression was, the BD is older than the BQ (probably the BQ is RoHS compliant or something)
What spec's are you referring to? *scratching head*

[Gyro]

Ah no, the ones I bought were JQs. Yours BDs (BQs same) are better.... ~1.5ppm vs my ~3ppm. 

I probably confused you with that new ebay link that I spotted.

Edit... Oops, No I did put BQ in my original post didn't I. Mine are definitely JQ. 

[BU508A]

Edit... Oops, No I did put BQ in my original post didn't I. Mine are definitely JQ. 

No worries. Cup of tea? Some bisquits? 

Cheers,

Andreas

[2N3055]

BD version was made in brazed ceramic package (D-16) (with metal lid), whereas BQ is Cerdip (C-16).. Both hermetic.
Specs seem to be the same.   Brazed ceramic package should be a bit more robust to pin force.. In practice I don't think it matters..

[Gyro]

Let me try and get my suffixes right today! 

I think the BD version was last included in the AD588 Rev A spec. It seems to have gone in Rev B anyway.

I've attached a copy for reference, just in case...

[tszaboo]

Careful with these old AD datasheets. They used accelerated aging to determine the long therm drift, and Linear technology actually debunked the values in them. After this some drift values were changed some two magnitudes (!!!).

[Gyro]

Hard to be able to draw comparisons. Both the Rev A and the latest Rev M datasheets show long term drift as 15ppm/khr @25'C. That implies that either:

A. AD are still using the accelerated method
or
B.  The old accelerated aging method actually correlates with long term aging
or
C. The recent dies have much lower drift than the older ones.

That's got me curious. 

[2N3055]

From what I can see on AD588BQ, after initial 10-15ppm drift in first 500-600 hours, now it is very low, below what I can measure  with what I have...

Long term drift for first 1000 hours will be dominated by what it will drift in first 100-200 hours.. Curve is exponential.  If manufacturer would just plug in references for 50 hours before shipping, they could specify much smaller long term drift..

But long term drift will easily be swamped with temp hysteresis effect it you use it trough wide temp range.
AD588BQ that I have seems to be extremely good in that regard as far I can tell.

[tszaboo]

Well, if you want to look into it, check out ADR290 datasheet revision A and the current one. And Linear technology DN229.  It should tell a story. I find it amusing TBH. Now AD buying Linear, who will point out these?

15 ppm without maximum value for a hermetic package... sounds about right.

[Gyro]

Oh yes, I knew about the LT debunking of the AD XFET references.  I guess in the future AD will be responsible for keeping themselves honest! 

I suspect that with a hermetic package buried Zener it doesn't really matter how you age them (within sensible limits), they will always tend to have good long term drift. It may be that AD always did decent sample drift evaluation on these parts, just skimped on the mushrooming number of cheap plastic packaged non-zener types.