Wow, good catch. I can't imagine that does anything good to the reference.
-branadic-
Why wouldn't it be good? The LT6655HLS8 family has better specs than than the other versions. So mechanically/thermally it is reliable. AD claims tha mechanical stress has no mentionable effect on the package. Since it has lower thermal mass it is easier to isolate. Would it not have better specs?
Would it not have better specs?
compared to what?
A Ford T-Model (SMD ceramic) is of course better than a plastic "Bobby Car" (MSOP-Package).
But a Mercedes (Thermally isolated TO-46 package) is still a class better.
What do you expect from the LS8-package?
Measuring air speed?
or measuring PCB humidity?
with best regards
Andreas
AFAIK they offer a ADR1000 derived version also in a ceramic case.
The package as shown on the picture looks quite thick. So if they are really into it, why not a SMD version, but with some changes:
1) slightly lower temperature
2) a 3 layer case with ceramic, isolation, ceramic and only relarively thin contacts between the 2 parts in a way that this is not bridged by solder.
With more pins they would be able to realize Kelvin connection. The ADR1399 die would allow Kelvin connection... Coming soon...
...probably tomorrow...
What do you expect from the LS8-package?
Measuring air speed?
or measuring PCB humidity?
Completely avoiding Kovar and FR4 materials and being able to do custom hybrid without the need to open a metal case and re-bonding.
Actually we need
KGD version ! [
... waiting Noopy
pictures ]
Andreas,
the HLS (it is NOT a LS8) package is known to be immune to humidity and also is very tolerant to mechanical distortion. Being so small makes it very easy to add some protection against air flow. AD has experience with it: LTC6655 has been out there for several years. I believe this was their test case.
And indeed: having extra connection pins is an added advantage. All this may make the ADR1399 in this style just even more beteer than the ADR1399AHZ. Closing up to the performance of LTZ100 (Already usable beyond 6.5digt) at a fraction of the price. I bet ADR1000 will follow in a few months.
the HLS (it is NOT a LS8) package
Hello,
can you give me a pointer to a datasheet with the HLS package? (cannot find one on the AD page).
I always thought H is the temperature range (-40 .. 125 deg C) and LS8 is the package.
with best regards
Andreas
Completely avoiding Kovar
How do you get this idea together with hermetically tight packages?
And why are my LS8 references attracted by a magnet?
with best regards
Andreas
Why do you get together hermetic and package? Hermetic to what? We all live surrounded by something, no matter if you live in the coast, on the mountains, under deep sea, over geostationary orbit, inside a 1m thick Pb chamber, or inside a 12m diameter ultra-pure heavy water neutrino detector pool ... you will always be permeable to something.
Your LS8 references case material could be magnetic. Is that bad?
Andreas,
"can you give me a pointer to a datasheet with the HLS package? (cannot find one on the AD page)"
That was my original question. The image makes us suppose there is (or will be) a ADR1399 variant in that type of casing.
For properties of the package please refer to the specs at AD (with an open mind?).
Is that bad?
No.
Kovar (which is needed for hermetically tight packages) is magnetic.
I am shure you will find out what long term stability and hermetically tight packages have to do for each other with a little google search.
with best regards
Andreas
High-end references from AD seem to get the LS/LCC package treatment. Two examples are the LTC6655 and ADR4550D:
Hello,
thanks for trying to clarify.
While the LS8 (or formerly LCC package) is well known I had the impression that RikV has info on a new "HLS" package which might avoid the disadvantages of the LS8-package. But obviously he has no new info.
In one early brochure (New Releases August 2012) the humidity sensitivity of the LS8-package was given with 10 ppm/25 %rH change. But since this sensitivity is related to the FR4 PCB material and not to the LS8-package itself it was left off in later data sheets giving only the hint that the PCB mounting is critical.
Generally all SMD packages suffer more from PCB stress than CERDIP or even TO Packages.
So the real "high end" references for me are the TO packaged LTZ1000, LM399 followed by the AD586/AD587 in CERDIP Package.
The LS8 package has of course advantages against other SMD packages like MSOP or SO-8. But I generally measure a factor 2 less standard deviation on a AD586LQ against a LT1236AILS-8. Both in a ovenized measurement condition.
with best regards
Andreas
Andreas,
I understand you are reluctant to accept the (H)LS package as a full value in the small world of high performance references. Probably you are right, but wouldn't it be better to rely on figures and experience than just on an "emotional feeling". Why not do real, objective testing when such device comes available and produce some figures, data that allows to place them where they belong: Excellent, good but only for.., just average?
After all is that n ot what this blog is all about? Experiment and search for inmprovement?
"Generally all SMD packages suffer more from PCB stress than CERDIP or even TO Packages". Yes, but since the thing is hardly 5*5mm, how bad is that suffering? (again: figures).
So, once again, has anyone any idea if or when this could be available?
The LM399 / ADR1399 are heated referenes running quite hot. With the To46 case much of the temperature difference is over the legs.
With the SMD case much of the temperature difference would likely have to be in the PCB. So the performance may very much depend on the PCB - it may need something like a high temperature flex PCB or other PCB material with good temperature stability. Normal FR4 may show aging at 90 C and humidiy could be problem for the PCB material if not running 24/7.
With other parts the ceramic SMD packages do react to board stress and chances are this would also apply to the new chip.
The stress is similar, no matter what size. The stress only gets smaller, if higher then wide, So maybe below 1-2 mm.
Isn't it scientific practice to transfer experience from a non-heated reference A in LS8 package to a heated reference B in the very same package? And since there is experience but also some numbers given for reference A in LS8 package it is valid to assume very similar results for reference B in the very same package.
ADR1399 in LS8 though has its right to exist, it might be intended for precision application on automated SMD assembly lines, but it will probably not be good enough for metrology use, which calls for the TO package. ADI can prove me wrong on that.
There might be some workarounds to cover the well know issues of strain transfer from the PCB to the ceramic SMD package and from there to the die itself, such as Flex PCBs or Rigid-Flex PCBs, with the ceramic package sitting on the flex part. But even then the heat transfer of such a ceramic package calls for a pretty well insulating die attach. I wonder how ADI addressed that challenge on ADR1399 but also ADR1001. Hopefully we will know soon.
So let's differ between precision vs. high precision, high precision vs. ultra-high precision or ultra-high precision vs. metrology use and call it a day
-branadic-
Haven't seen any mention of die attach in this thread - will a flexible die attach and bond wires isolate SMT package stress from the die?
Also - PCB slots are usually voltnut voodoo, but in this case might help reduce package stress?
We could really do with a lot more data on both points.
Edit: I'm amused branadic mentioned it at the same time as me.
Haven't seen any mention of die attach in this thread - will a flexible die attach and bond wires isolate SMT package stress from the die?
That is sort of challenging. During wire bonding you don't want the die to move, as you introduce heat and ultra-sonic to the die and bond wire. A moving (flexible) die would make that pretty hard of a job.
I'm not aware though of a die attach that has shear viscosity properties; that can be made flexible after wire bonding or that can be removed somehow after the wire bonding process to have the die floating, since we all know that air is the best insulator, right?
Edit: macaba probably meant flexible SMD assembly, not die attach.
Edit2: I do not recommend the use of ICA (conductive adhesives) as they are prone to humidity and their CTE is pretty large, worsening the issue of strain transfer even more.
-branadic-
I meant die attach. This is a problem to be solved at the source, not by voltnut wizards of the internet!
With the die attach compound used in the LTZ100A (400°C/W) the heat transfer into the PCB would be still reasonable. A heat spreading (GND-) layer under the package would reduce both absolute temperature as well as thermal gradients. Last but not least a ceramic package won't require Kovar...