ADR1000 Datasheet

[bsdphk]

"The 399 [meaning the components on the chip/die] is of highly asymmetric design.."

Yes, I know that, and provided the thermodynamic circumstances of the metal can are constant, I am sure that will cause some measurable amount of orientation sensitivity.

But they are not constant, on all the LM399's I have had my hands om, the plastic cap can move relative to the pins, and that means a huge difference in thermal conductivity between the metal can and the plastic insulation.

So by all means attempt to characterize the orientation sensitivity of the chip design, but do so only after you either glue the plastic cap to the metal can or remove it entirely.

[Kleinstein]

The movement of the plastic cap is a good point.  Instead of glue I would consider a little bit of cotton wool or similar to keep it in a fixed position.
The can itself is relatively small and there is little temperature difference inside. So I don't think convection inside the case is an issue.

[bsdphk]

Instead of glue I would consider a little bit of cotton wool or similar to keep it in a fixed position.

My preference would be to get rid of it entirely, because it does the wrong thing for what we're trying to achieve.

And yes, that is one of my pet peves: Metrology people tend to think in terms of thermal conductivity when they should focus on thermal impedance, so let me explain:

Thermal insulation = "Unwillingness to move thermal energy" = Porous low thermal capacity material which restricts the movement of gas molecules.  Example materials: EPS foam, "RockWool" and knitted sweaters.

Thermal impedance = "Unwillingness to change temperature" = Solid high thermal capacity with low(-ish) thermal conductivity.  Example materials: Non-crystaline rocks, bricks & concrete.

The benefit of high thermal impedance is that temperatures changes propagate very slowly through the material, around 25kg of granite will get you into microkelvin territory.

The downside is that it can therefore take weeks or months to reach that thermal equilibrium.

If you want to experiment:  Sandwich your PCB between a couple of candlelight holders made from granite, concrete or mable.

[mzzj]

Thermal impedance = "Unwillingness to change temperature" = Solid high thermal capacity with low(-ish) thermal conductivity.  Example materials: Non-crystaline rocks, bricks & concrete.

The benefit of high thermal impedance is that temperatures changes propagate very slowly through the material, around 25kg of granite will get you into microkelvin territory.

Add paraffin wax and some plastics to your list. Higher specific heat capacity than granite and lower thermal conductivity.

[magic]

So by all means attempt to characterize the orientation sensitivity of the chip design, but do so only after you either glue the plastic cap to the metal can or remove it entirely.

More like: solder thin copper twisted pairs for zener sense, zener force and heater power (bonus: kovar cables with solderless joints). Place in the middle of a copper pipe and flood with polyurethane foam, surround with another layer of insulation. Then you may have some excuse to blame the remaining effect on air flow or dust inside the can.

Otherwise, external air and PCB are likely to have more effect.

[ramon]

No PU foam inside, MgO powder. Low electrical conductivity, high thermal conductivity to avoid issues with kovar. Both Over the pcb and under the pcb too.

Once the above is done, then the thermal insulation. PU or whatever you like. Then do the tilting test again, and you now  that this can only be the effect of internal die issues.

Edit: no need to say, that all of the above, without the plastic case.

[bsdphk]

Place in the middle of a copper pipe and flood with polyurethane foam, surround with another layer of insulation.

No, no & no.

See above re: Thermal impedance

[bsdphk]

Add paraffin wax and some plastics to your list. Higher specific heat capacity than granite and lower thermal conductivity.

Yes, there are some plastics in the right ballpark, and yes they are much easier to work, but polymers as a rule, cannot be bought in a usable shape.

You can get them as pipes, sticks, granules, sheets, foils, tapes, but not as a solid 5cm cube.

If you find something 5cm thick, it is likely to be at least 40cm on the two other dimensions, and weigh so much the postman will hate you forever.

(Cutting 5cm thick polymer takes a really good wide-toothed high-speed circular saw. Too low speed or too thin-toothed and the friction will melt the polymer and your saw will get stuck.  Do not ask me how I know this :-)

All that said:  Any lump of high density polymer (HDPE, ABS etc.) will give more stable temperature than what we usually see on pictures here.

But ask your wife first:  There is 22% chance she have a pair of geologic candlelight holders she never uses.

[chris_11]

The issue with the covers is much more the prevention of thermal flow variations than pure isolation. Even a thin heat-shrink cover over the Kovar leads and package will do. Thermal air movement will create noise through low frequency temperature changes in the Kovar leads, which have about 40uV/K to copper similar to lead/tin to copper. Check low frequency performance on a LT1000 without a cover. Then test a simple heat shrink tube (no need to shrink) just cover the leads and package.
The lattice is heated and on constant high temperature. The PCB is on a lower temperature, so a thermal gradient is always present. Voltage variation equal to 1/f noise is created when the thermal gradient over the leads varies with air movement.

BR
Christian

[ScoobyDoo]

Hello folks
                The ADR1000AHZ is now available from the ADI online webshop.

The MOQ has been set as low as 5 (one) to reach all Voltnuts - (MOQ = 5).
The part will be monitored for a period of at least six months to understand if this is a viable (sustainable) path for ADI - so pls. order as much as you can ...
ADI is checking this MPN in order to understand if ordering process goes smoothly and will remove existing issues with this MPN = ADR1000AHZ

Best regards
ScoobyDoo

[Echo88]

Is the AD7177-2 capable of using the ADR1000AHZ directly for its Refinput?
The maximum ratings for the AD7177-2 state:

AVDD1, AVDD2 to AVSS: −0.3 V to +6.5 V
Reference Input Voltage to AVSS: −0.3 V to AVDD1 + 0.3 V

So i would have guessed the Refinput accepts 6.8Vmax in at AVDD1 = 6.5V, suitable for a max 6.67V from a ADR1000AHZ.

Or is it foolish of me to think that might work due to substrate diodes starting to conduct/other reason?

https://www.analog.com/media/en/technical-documentation/data-sheets/AD7177-2.pdf

[razvan784]

It's not wise to run chips at their maximum ratings, nor for the inputs to exceed the supplies.
What I would do is divide down the reference to something like 6V and use that as both the supply (through a power buffer of course) and reference input.
Dividing 6.6V to 6V will attenuate the error due to the drift of the divider approximately 10 times, as discussed by Dr. Frank in the LTZ thread.
You can use a thin-film array as indicated by John Pickering and as discussed in the Statistical Arrays thread.
The recent paper by N. Beev also linked on this forum indicates the low-noise types (TDP, NOMC etc)

[magic]

These are stress ratings only, operation is not implied, blah blah...

They specify reference input range as 50mV beyond the rails, assuming you aren't using the internal buffer of course. What's beyond 50mV? Dragons... maybe OK at 25°C, maybe not.

[TiN]

Thermal image of bare chip during 2 minute operation. Hottest point is die epoxy attach. Die itself looks cooler due to different emissivity and reflection parameters. Resistors for heater are 11.5kOhm/1 kOhm, zener current 4.05mA (120 Ohm).

[chris_11]

They specify reference input range as 50mV beyond the rails, assuming you aren't using the internal buffer of course. What's beyond 50mV? Dragons... maybe OK at 25°C, maybe not.

In CMOS circuits you inject substrate or well currents if you go under the lowest -supply or above the +supply, unless there are special additional structures used. Typical there are diodes similar to a FET Body diode and/or. ESD structures with similar effect. So 50mV will be o.k. since you do not get significant currents. But for sensitive applications even 300mV might be to much and there are always spikes from noise etc. So do not exceed the supply boundaries on any input unless the data sheet explicit allows for that.
If in doubt, check with a nA meter at what point currents starts to flow.

BR
Christian

[chris_11]

Thermal image of bare chip during 2 minute operation. Hottest point is die epoxy attach. Die itself looks cooler due to different emissivity and reflection parameters. Resistors for heater are 11.5kOhm/1 kOhm, zener current 4.05mA (120 Ohm).

Is that a cut open TO package from the top or how did you make the image?

[magic]

Definitely looks like that. The orange rings are glass seals around pins and you can also see where is the pin connected to the can. The image shows thermal emissivity of the various components first and foremost but temperature differences between identical materials may be comparable.

So do not exceed the supply boundaries on any input unless the data sheet explicit allows for that.
If in doubt, check with a nA meter at what point currents starts to flow.

It's worse than that, because D-S ADCs sample and hold the input in some internal capacitors which are switched by means of MOSFETs, so there is likely leakage from those capacitors too if their potential forward biases the body diodes.

[iMo]

Btw, why the glass seals are 12C higher than the surrounding metal?
It looks to me like "the Blackbird vehicle faster than the tail wind" problem.. 

[Kleinstein]

The reference has an essentially constant temperature. Nearly all the details visible is a diffenrence in the emissivity. So the picture is not eally showing temperature, but more like optical properties in the IR range, more like a normal photo, just in the IR.

[KT88]

Is the AD7177-2 capable of using the ADR1000AHZ directly for its Refinput?
The maximum ratings for the AD7177-2 state:

AVDD1, AVDD2 to AVSS: −0.3 V to +6.5 V
Reference Input Voltage to AVSS: −0.3 V to AVDD1 + 0.3 V

The answer is a clear NO.
The (operating-) specs are stated for 4.5 to 5.5V. Anything outside that values is not covered by the specs.
There is an abs. voltage spec for the external reference (Vref = (Ref+) - (Ref-)) which is AVSS (min) to AVDD1 (max) with buffers enabled and AVSS-0,05V (min) to AVDD +0.05V (max) with buffers disabled.

Abs. maximum ratings only state what a chip can survive.

Cheers

Andreas

[rigrunner]

Thermal image of bare chip during 2 minute operation. Hottest point is die epoxy attach. Die itself looks cooler due to different emissivity and reflection parameters. Resistors for heater are 11.5kOhm/1 kOhm, zener current 4.05mA (120 Ohm).

How are you deriving 4.05mA at 120Ohm?

If I plug the numbers into the formula on page 7 of the data sheet I get a slightly different figure.

11K5/1k heater should give 51.265 C die temperature.

Using the R1= 0.658V-0.0022*51.265 over 4.05mA minus 7 Ohms  R1 resolves to 127.62 Ohms.

To get 120 Ohms R1 with the same heater value I need 4.29mA.

I'd like to understand where I'm going wrong with the data sheet calculation 

[TiN]

I felt like formula on page 7 slightly wrong, as there are 7 ohms in series were lost and they used 0.0022 in one spot, but 0.002 in other. 

Actual heater set was 11.435/1k (13kOhm + 95kOhm in parallel), so bit different. I'd also think removed cap would affect temperature quite a bit due to no protection from air drafts/heat loss outside.
That imagery is more of entertainment value, not metrology value.

[rigrunner]

I felt like formula on page 7 slightly wrong, as there are 7 ohms in series were lost and they used 0.0022 in one spot, but 0.002 in other. 

The example in the data sheet is also wrong in that it doesn't appear to subtract the 7 Ohms. The 129.5 Ohm example result should be 119 Ohm (using the .0022)

Actual heater set was 11.435/1k (13kOhm + 95kOhm in parallel), so bit different. I'd also think removed cap would affect temperature quite a bit due to no protection from air drafts/heat loss outside.
That imagery is more of entertainment value, not metrology value.

It's pretty

[d-smes]

The datasheet refers to a "recommended layout" on page 1 but none is given.   Is there other documentation that shows this recommended layout?

[leighcorrigall]

You can get them as pipes, sticks, granules, sheets, foils, tapes, but not as a solid 5cm cube.

If you find something 5cm thick, it is likely to be at least 40cm on the two other dimensions, and weigh so much the postman will hate you forever.

(Cutting 5cm thick polymer takes a really good wide-toothed high-speed circular saw. Too low speed or too thin-toothed and the friction will melt the polymer and your saw will get stuck.  Do not ask me how I know this :-)

What? 

Foam isn't that difficult to source. Here is an 8x8x8 inch^3 polyurethane cube for example:
https://www.mcmaster.com/8882K16/

As for cutting... please don't use a power saw. Instead, use something like a (hot) knife and cut manually:



 

Tip: If you use a hot knife (or a cheap soldering iron), do not keep it in one position too long or you might start a fire