Ref80 - A competition to ADR1001?

[branadic]

"REF80 is a buried zener, internal heater compensated voltage reference. The device is designed with an extremely low temperature drift and long-term drift. REF80 is also a low noise device."

https://device.report/m/158081276c8014e611d4be1e2e7e80c15bf10773f92c56a97763852ea16d52fd.pdf

Has TI really worked on a competition to ADR1001? HP3458A reading in figure 2-3 indicates a zener voltage of 7.577 V.

-branadic-

[mendip_discovery]

"REF80 is a buried zener, internal heater compensated voltage reference. The device is designed with an extremely low temperature drift and long-term drift. REF80 is also a low noise device."

https://device.report/m/158081276c8014e611d4be1e2e7e80c15bf10773f92c56a97763852ea16d52fd.pdf

Has TI really worked on a competition to ADR1001? HP3458A reading in figure 2-3 indicates a zener voltage of 7.577 V.

-branadic-

oooh, that sounds interesting. I guess AD have kinda dropped the ball a bit with the ADR1001 as they have only managed to produce a new ref much like the old one but not as good (YMMV).

Looks like this is very early info so will wait to see what happens.


EDIT:
Looking at the site there is a REF70,
https://www.ti.com/lit/ds/symlink/ref70.pdf

Long-term stability: 1k hrs: 35 ppm
Coefficients: 2ppm per deg C

even newer is the REF54,
https://www.ti.com/product/REF54

[iMo]

Let us wait on REF90.. With the opamps and 5V/10V dividers in the oven.
What is interesting to see is the pcb layout of the eval kit - how are the slots around the package arranged..

[jbb]

Having another vendor for high precision references would be welcome.

I wouldn’t expect a drop-in replacement, but having a competitor to Analog Devices in the game is a good thing.

[laichh]

Identical file was found on ti.com domain:

https://www.ti.com/lit/ug/snau304/snau304.pdf

From the schematic in Figure 4-1, this REF80 doesn't have the derived output (+5 V, +10 V...) like the ADR1001, only the zener output at ~7.6 V is available.

Traditionally, TI's alternative is often (if not always) cheaper than ADI. ADR1001 1ku listed price on ADI is $66.34, I would expect this REF80 to be range between $35 ~ $50 each. 

[ArdWar]

Interesting, the pinout is MostlyCompatible™ with ADR1001.

Is this TI's first foray into buried zener technology? I don't think their other references are buried zener (LMx99 don't count )

[Andreas]

What about REF102 (Burr Brown Division)?

[3roomlab]

nice
the slotting they use is so small
like size of the 2.54mm pin (~ 0.4mm ?)
* mr imo is right, more like 1mm slot

[Kleinstein]

What about REF102 (Burr Brown Division)?

The REF102 is a nice 10 V reference, but the metal case seems to be no longer available.


How good the Ref80 will actually be is still unclear. There is not even a preview data-sheet. It may as well be more competing with the LCC version or the ADR1399.

The slightly higher voltage is a bit odd and rather high for zener plus PN junction.

[iMo]

nice
the slotting they use is so small
like size of the 2.54mm pin (~ 0.4mm ?)
..

The pitch of the package pins is 1.27mm. Thus the slots are aprox 1mm wide, imho.
The slots' layout seems to help with the mechanical stress as well.
So let us copy that layout for the ADR1001 as well..

[Kleinstein]

The slots are for mechnical stress and to keep the thermal loss low, though much of the heat loss can still be via the copper traces. It thus makes sense to have separate drive and sense pins for the output and also separate grounds for the reference and the heater.

[iMo]

The slots are for mechnical stress and to keep the thermal loss low, though much of the heat loss can still be via the copper traces. It thus makes sense to have separate drive and sense pins for the output and also separate grounds for the reference and the heater.

I've been planning the pcb for 2y already  and perhaps the other owners do too, let us continue in the ADR1001 thread and let us design the pcb finally..

[laichh]

A dedicated page for REF8EVM is now available on ti.com.

https://www.ti.com/tool/REF8EVM

Still no DS was found.

[PCB.Wiz]

Page is now up, with data

https://www.ti.com/product/REF80

1-99     $84.841
100-249   $82.289
250-999   $68.510
1,000+   $63.790

[laichh]

Well, in summary, the REF80 is about half the TC & double the noise vs. the ADR1001. Can't tell about the long-term stability as the chart provided is only up to 1000 hours.

Parameter	| LT LTZ1000ACH	| ADI ADR1001	| TI REF80	| Unit
Released	| 1987	| Jan-2024	| Oct-2024 (Preview)	| Year
Zener Voltage	| 7.2	| 6.6	| 7.6	| V
Noise (0.1~10 Hz)	| 1.2 @ 5 mA	| 0.6 @ 4 mA	| 1.2 @ 15 mA	| µVpp (Typ)
Temp Coef	| 0.05	| 0.2	| 0.05	| ppm/°C (Typ)
Price (1ku)	| $59.92	| $66.34	| $63.79	| USD

[floobydust]

I have an issue with piling on the output capacitance in order to achieve that datasheet glory low noise number.
The EVM ref output film caps- one SMT, the other Wima the only through-hole part really for total 20uF+0.22Ω seems heavy handed. Sigh.

That doc is looking pretty sloppy for August, TI lazy copy'n'paste errors galore:
It's not a REV54EVM, isn't is a REF8EVM? 
"4.2 PCB Layouts"
"REF54EVM is a two layer board."
Uh no, figures 4-4, 4-5 show layers 3 and 4. Isn't it a REF8EVM? Who gets paid to do such low quality editing work.

"4.1 Schematics"
"The schematic for the REF8EVM is illustrated in Figure 3-3"
Uh no, figure 4-1 shows the schematic.

Heater voltage can be 10-42V for 75-18mA hot T_SET=115°C for around 0.75W

[Andreas]

Well, in summary, the REF80 is about half the TC & double the noise vs. the ADR1001. Can't tell about the long-term stability as the chart provided is only up to 1000 hours.

Also interesting the zener current vs. noise.

with best regards

Andreas

[ArdWar]

What kind of temperature control are they using?
Fig 9-2 hints something two stage, not fully linear, and VREFZ is gated before OPSTBL is asserted?

REF80 samples the value of the resistor at the time of startup and the value of TSET is modified accordingly. Small drift in resistor value will not impact the TSET temperature as T-SET pin samples discreet value.

They put digital mgmt inside?
Now instead of worrying about R_TSET drift we worry about how good the sample and hold circuitry holds its value?

[Kleinstein]

For the temperature there seem to be only 4 discrete temperature to chose: 115 C, 105 C, 95 C and 85 C. So all relatively high.
Looks like a digital solution for the set temperature. The actual regulation is likely still analog even though a long time constant is also attractive to realize digital.

The supply current for the zener part is also quite high (15 mA) and thus quite some self heating and much below 85 C may not be that practical anyway, if the thermal resistance is high.
Overall it's a power hungy solution.

The DS numbers look good for the long term drift, but not so much with the noise. The shown noise is relatively high at the shown higher frequencies and the true LF noise below the 1/10 Hz may not be that bad. The actual drift will also depend on the PCB, though no critical external parts.

[tszaboo]

What kind of temperature control are they using?
Fig 9-2 hints something two stage, not fully linear, and VREFZ is gated before OPSTBL is asserted?

REF80 samples the value of the resistor at the time of startup and the value of TSET is modified accordingly. Small drift in resistor value will not impact the TSET temperature as T-SET pin samples discreet value.

They put digital mgmt inside?
Now instead of worrying about R_TSET drift we worry about how good the sample and hold circuitry holds its value?

It's likely not a sample and hold circuit. TI uses this in many ICs now: They sample a voltage divider (external resistor) on startup and then set up a DAC based on this sample. That's why you have ICs (for example voltage regulators) where there are specific values for external resistors which set up specific values of voltages. And it allows very low power consumption, since you can shutdown the sampling of the resistor, which is clearly not a requirement here.

The IC is interesting. The eval board is a bit MEH. They put a lot of effort to isolate the heat generated by the part from the rest of the PCB, insead of concentrating on air currents and thermouple effects. It also seems to use regular banana plugs, not even gold plated ones. Also, LM317  I don't think the engineer putting the ref board together has put too much tough  into it.

[macaba]

To answer the thread title, "Ref80 - A competition to ADR1001?"

Not really.

Three things that stand out to me: zener noise to supply current figure-of-merit is bad. >300mA inrush on power up is bad & no mention of a 10 ohm, or something, resistor to limit it. The price is a little high when holistically considered against competitors. There will be other (good & bad) things also but those jumped out at me today.

On the positive side though - it's evidence TI have a team who can design this level of reference and that can only be a good thing going forward. For those reading this - thanks, I hope you get approval for more products like this.

[dlebed]

Since ADI chose not to sell (!) me the ADR1001 for whatever reason, I decided to entertain myself with REF80, which I managed to get my hands on.

I only was able to build a barebone prototype board, but should be enough for basic measurements (assuming you have a low noise power supply).
Here are my results so far, after spending couple hours testing it.



- it's currently missing the thermal insulation (it's on its way)
- output filter cap is C0G + 2 x X7R for now, will need to replace X7R with film cap

I decided to run the reference at max. temperature setpoint (115C) for a month to speed up aging, then I'll change the setpoint to 85C.
My first REF80 at 115C settled at 7.568V.
External case temperature is just a little shy of 90C.

As you can see from thermal images below, board cutouts are quite effective and help with thermal isolation of REF80.
I went against datasheet recommendations and put a symmetrical round copper pour under the REF80 itself to spread the heat evenly (it anyway will be thermally insulated).

Top

Bottom


I'm running both heater and bias at 15V, the total current with well insulated REF80 is 32mA @ 15V (0.5W) (this is again 115C heater setpoint with ~25C ambient).

Measuring it with a few 3458a's and DMM7510 there is clearly an expected downward aging trend seen even over couple hours.


I have quite a few ideas in mind of what to do with it, so I will post updates here. I'll try to measure output impedance and post results soon.
Unfortunately, I don't have a good noise measurement setup, so I could not make any meaningful noise measurements yet.
If anyone has suggestions for specific tests to try with it, please let me know, and I'll see if I can give them a shot.

Here are the design files https://github.com/dlebed/ref80-basic

P.S.You can order the board on oshpark if you want https://oshpark.com/shared_projects/kTTHxNFD

[aronake]

Since ADI chose not to sell (!) me the ADR1001 for whatever reason, I decided to entertain myself with REF80, which I managed to get my hands on.

If ADI do not want to sell a ADR1001, maybe you can try Mouser:

https://www.mouser.com/ProductDetail/Analog-Devices/ADR1001AEZ?qs=ZcfC38r4PosJwibAPZPq6Q%3D%3D

17 in stock now.

Seems you live in USA, and if you do not click the boxes that you intend to use it for military purposes or resell it to some evil country, there ought to be no problem buying. I got 2 ADR1001 and one evaluation borad sent to Hong Kong which should be pretty high up on the evil list.

[dlebed]

aronake, you supposedly could buy ADR1001 much cheaper from ADI directly, but somehow they've cancelled my order.
Right about that time I stumbled upon REF80, so I just decided to try this new beast instead.

Maybe later I'll buy ADR1001 from Mouser, but I need to finish my LTZ1000 and REF80 designs first 

[iMo]

..
Here are my results so far, after spending couple hours testing it.

The cutouts - I would prefer the shape as in the kit DS - that helps with the mechanical stress a bit, not sure your design is more effective in this regard..
Aging at 115C might be significant, imho, my ADR1001 spent some 2 days at around the same temperature (by my mistake, call it fast burn-in) and it moved down with the voltage as well. Also I would not run it at 115C for a longer period, however.
PS: internal 115C with the case 90C sounds a bit odd, the temperature difference should be much higher, imo.