Ultra Precision Reference LTZ1000

[tatel]

These guys from lankamicro.com are actually from Shenzen.

They have very strange pricing, you can pay the same for a PTF65 no matter if it is a 0.1%, 5 ppm one, or a 1%, 10 ppm.

So perhaps it's not even legit.

[The Soulman]

That website sure looks like a scam.

Just did a "cold-start" on my ptf56 euipped ltz1000, It was powered off for 20 hours to cool down.
I've only measured the 7v output.
After applying power it took a minute to reach the original value and after that it overshoot by some 4ppm and came back down to the original value within 30 minutes and remains rock solid after that.

One other reason I've used multiple ptf56 (0,125W) was to help with power dissipation, ptf65 (0,25W) is better but might require some calculations to make sure it is enough.
Also, higher dissipation means higher drift. (check the load life stability in the datasheet).

At the (digikey) price of ptf65 is close to the Caddock USF340 series (0,33W and much larger surface area) and Vishay S102 (0,6W), S102 are also available at https://www.buerklin.com/en/search/s102/

[tatel]

That website sure looks like a scam.

Yep, I couldn't be bothered to even send an email. If it's to good to be true, it's too good to be true. To get a 0.1%, 5 ppm PTF65 for at most 2.62 €, and perhaps even for less than a dollar, is too good to be true.

Just did a "cold-start" on my ptf56 euipped ltz1000, It was powered off for 20 hours to cool down.
I've only measured the 7v output.
After applying power it took a minute to reach the original value and after that it overshoot by some 4ppm and came back down to the original value within 30 minutes and remains rock solid after that.

One other reason I've used multiple ptf56 (0,125W) was to help with power dissipation, ptf65 (0,25W) is better but might require some calculations to make sure it is enough.
Also, higher dissipation means higher drift. (check the load life stability in the datasheet).

At the (digikey) price of ptf65 is close to the Caddock USF340 series (0,33W and much larger surface area) and Vishay S102 (0,6W), S102 are also available at https://www.buerklin.com/en/search/s102/

I finally got 0.1%, 5ppm/K RESI MMFR2568s from LCSC at about €3.29 each, about €2.80 each buying ten. Even 1%, 10 ppm PTF56s from digikey are about €4.99. I compared the datasheets and MMFR specs looked to me not so bad, compared to "tipycal" Vishay specs. So i took the gamble. Datasheet attached.

[Simmed]

have you played with it in the temperature/RH chamber yet ?

[dietert1]

In some tests metal film resistors exhibited largely variable excess noise, and as far as i remember more than metal foil resistors.
You could look here: https://www.eevblog.com/forum/metrology/resistor-noise-test-setup-article-interesting-reading/msg6052857/#msg6052857

Regards, Dieter

[tatel]

In some tests metal film resistors exhibited largely variable excess noise, and as far as i remember more than metal foil resistors.
You could look here: https://www.eevblog.com/forum/metrology/resistor-noise-test-setup-article-interesting-reading/msg6052857/#msg6052857

Regards, Dieter

Understood, Fortunately, as previously stated,  for the most critical resistors I'm going with FLCY foil resistors (R12 and R13) and perhaps NOMC array (R4 and R5). I hope metal film is good enough for the remaining, not so critical resistors. As far as I can see, PTFs are also metal film over ceramic core. The same for Ohmite MOX700. Neither MMFRs nor MOX700 nor PTFs have any "current noise" spec in the datasheet. However FLCY do have that spec. From the paper you linked, it seems that NOMC arrays are not bad, either.

Do you mean all resistors on a LTZ1000 reference should be metal foil? I was under the impression it would be overkill. Am I wrong? Right now I can only hope I din't screwed it up. Info from The Soulman is encouraging, however.

Perhaps this is the reason Kleinstein said he doesn't like PTFs so much anymore?

Anyway, thank you very much for this chunk of information that was going under my radar.

have you played with it in the temperature/RH chamber yet ?

No, but I'll do in the future. That's for the medium/long run, however.

Regards

[Hydron]

Also interested in how well the RESI parts work - the PTF ones have rocketed in price

[Andreas]

Also interested in how well the RESI parts work - the PTF ones have rocketed in price

For me the RESI parts look like a copy (quality?) of UPF25/UPF50 resistors from TE

I have measured T.C. of 2 samples of UPF50 with 1K.
The result was not bad (below 2 ppm/K and no visible hysteresis on slow gradients = 0.12K/minute).

https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg462301/#msg462301

But of course this covers not humidity sensitivity and ageing drift.

with best regards

Andreas

[tatel]

After reading the paper linked by dieter1, I have been looking for metal film on silicon core resistors, which seemingly wouldn't have that excess noise.

So far the only thing i have found is this: https://resistor.com/product_silicon.html

These are SMD parts. Datasheet attached. It does show current noise spec, which would under -35 dB for 1R and -30 dB for 10M However best TCR would be 25 ppm/K

Not really sure about trading 5x TCR for some -6 dB (IIRC metal film over ceramic are about -30 dB?)

[The Soulman]

I don't understand why/how resistor noise (any variety) would have a noticeable impact on a reference's noise performance with the zener itself being a dominant source and after that any other active component?

Or are my (back of the envelope) calculations/presumptions wrong? (could well be)

[tatel]

UPF25 seem to be cheaper on LCSC than MMFR2568 -about €2.05 vs €3.29. Alas, only ten hits and no stock.

[Kleinstein]

The drift, TC and noise of the resistors is attenuated by some factor of some 200 or more depending on the resistor position.
This way the resistor excess noise is not a problem - it is only that noisy resistors also have the tendency to have more long term drift.
How noise a resistor type is depends on the type and value not all thin film on ceramic are the same. There are both rather noisy ones and rather low noise ones. From my limited tests I had rather noisy 15 ppm/K THT parts from ACROL (~ -25 dBi ?) and rather low noise Susumu RR 25 ppm/K (<-40 dBi AFAIR). With many datasheets the given noise specs are more test limits and not actual performance.

The TC should also no be that bad if 10 ppm/K or better resistors are used. Even 25 ppm/K resistors can be OK as the drift over the more limited use range tends to be better than the specs.
The likely most critical parameter is the long term drift and here data are very sparse.

[tatel]

The likely most critical parameter is the long term drift and here data are very sparse.

So datasheets don't give us the information we would need for this purpose. Moreover I see not all OEMs use the same standard, and it seems some OEMs use more than just one standard.

I guess load life is the best we can look at? And, perhaps, moisture resistance?

Anyway i compared specs betwenn UPF, MMRF and PTF. Here it is FWIW.


                                                                             UPF25                      MMRF2568            PTF56
---------------------------------------------------------------------------------------------------------------------------------
Tolerance                                                               0.1%                           0.1%                    0.1%
---------------------------------------------------------------------------------------------------------------------------------
TCR                                                                      5 ppm                         5 ppm                   5 ppm
---------------------------------------------------------------------------------------------------------------------------------
Endurance/Load life                                            0.2%+0.05R            0.05%+0.05R            0.04%
---------------------------------------------------------------------------------------------------------------------------------
Damp heat with load/Moisture resistance           0.2%+0.05R            0.05%+0.05R            0.08%*
---------------------------------------------------------------------------------------------------------------------------------
Pulse/short time overload                                   0.1%+0.01R                0,02%                    0.01%*
----------------------------------------------------------------------------------------------------------------------------------
Resistance to soldering heat                              0.05%+0.01R          0.05%+0.05R            0.02%*
----------------------------------------------------------------------------------------------------------------------------------
Note: * = Typical Vishay

[dietert1]

Some years ago i studied DALE RS02B resistors. Those are wire wound cement power resistors and well made with copper wires. When using them at low power they are very stable. Typical specs are 1% 25 ppm. In my tests i found many with < 5 ppm/K at 20 to 40 °C and  they exhibited ppm level long term drift.
I found those resistors in a E1490 VXI plugin hand-made for a metrology application (ebay deal).
The same resistors are also available embedded in aluminium heatsinks, which may be useful for temperature control.

Regards, Dieter

[tatel]

Have you checked Rhopoint Components?

-branadic-

Emailed Rhopoint and got no answer. Ten days ago used their website asking for a quote again. This time got an answer. It seems they don't work with private customers. So far they ask just for company name, but guess i'll need to give a valid european VAT number at the end of the day.

Not a problem, but thought it would perhaps be useful to let other people know about it. Rhopoint doesn't seem to work with private customers.

[tatel]

Attached the quote from Rhopoint

Price isn't that bad, but the MOQ -20 pcs/line- is hard to swallow. I think I'll look at Mouser before making any decision. Probably double the price and same wait, but hopefully no MOQ. We will see.

Out the window goes the Caddock USFS anyway.

[tatel]

Ordered from Mouser, 4 each, FLCY 10K and 4K. About €110. Lead time some 20 weeks... I hope that NOMCT array will be good enough to set oven temperature.

Thank you all for your help

[MiDi]

NOMC or other similar RNs from vishay in conjunction with unheated ZTC compensation for oven setpoint are good enough, interdigitation recommended.
For lower noise the 120R/100R can be reduced to e.g. 50R, that approximately doubles the zener current and halves the noise.