Early 3458A A9 ref, T106382=LTZ1000A ?

[TheSteve]

I just had a look at the A9 board on my reasonably early 3458A. Instead of being marked LTZ1000A the ref is marked LT T106382 date code 8809.

Has anyone seen this before?
Perhaps the same thing with a part # from before they decided on "LTZ1000"?

[TiN]

I'm 99.9% sure it's same old LTZ1000A 
If you want to know for sure, you'd need to drill the head off..

[Macbeth]

That means there is a 0.1% chance it is an LTZ1000CH (non A), the resistors should reveal the truth... 

[chuckb]

That's a neat find!
The LTZ1000 was available a few years before 1988 so they had that base part number already. According to Pickering, then other people made suggestions for improvements and that was the LTZ1000A.

I am quoting John (Pickering) literally:

“... I have a Datron reference module schematic, using the LTZ1000, dated early 1987. I believe that the first product we released using it was the 4910 but it may have been the 1281. I know we evaluated and analyzed it for quite long period beforehand so at the latest that would be during 1986 but may have been 1985. (Unfortunately I've lost my evaluation report which contained a lot of useful data but was on paper only!).

If the heater resistors are set up for 95 deg C (15k/1k?) you could measure the heater voltage (really Q401 Base voltage) on pin 1 of the J401 connector to determine if the chip is a LTZ1000 or a LTZ1000A.

[manganin]

based on the date code 8809 - the LTZ1000ACH is excluded from the party ... (with 6 sigma certainty it is an LTZ1000CH)
In the early days of the 3458A all special parts had a custom HP tag ...

The HP part numbers were different, for example see the op-amp in the photo.

The T-number usually means a pre production test sample. My guess would be the -A version based on the date code.

[Pipelie]

Hello Voltnuts - based on the date code 8809 - the LTZ1000ACH is excluded from the party ... (with 6 sigma certainty it is an LTZ1000CH)
In the early days of the 3458A all special parts had a custom HP tag ...

The parts on the photo from TheSteve look glossy - is there some conformal coating (Parylene) used ?
Does anyone know the public P/N of the MOLEX connectors being used J400 - J401 ?

Merry Christmas to all Voltnuts
Flinstone

What about mine? I thought it is LTZ1000ACH, NOT check it yet.

[Andreas]

Has anyone seen this before?

Hello,

strange. The LT1013 is blacktopped.

Was that a usual thing of LT at that time?
Blacktopping is something that I only know from counterfeit parts.

with best regards

Andreas

[quarks]

afaik all early HP 3458A had this 1826-1860 HP part number for LTZ1000 and later switched to original LT part number

https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg440601/#msg440601

edit/correction: in a conversation with Flinstone I realized I mixed up numbers, my reply only refers to 1826-1860, therefore I corrected my reply.

[TiN]

Indeed time to stable level is very different between chip versions. My A9 with CH chip takes about 5 minutes to settle at it's level (including lower temperature setpoint, 15K+100K), while standard ACH'es are seconds from cold state, due to 4 times better thermal insulation from the package.

[Dr. Frank]

All of my 5 LTZ1000 stabilize their oven temperature of 50°C within 15..30sec, in accordance with the datasheet value. See table inside my presentation of these 5 new kids on the block.
The reference voltage takes longer, about 30min to <1ppm ..about 2h for full stability of <0.05ppm.
Therefore, all of you should explain / define, to which criterion you're relating to.
Frank

[TiN]

My numbers were to 1ppm DCV readout match.
E.g. 3458A (cold) connected to known LTZ ref output. Power on, start timer countdown till DCV readout reaches better than 1ppm tol from previous recorded value (measured by same unit, once it was hot for days).

Heck, my CH-A9 3458 will even fail self-test if you run it one minute after the power on.

[VintageNut]

Week 14 of 1991

1826-1860 9114

[TheSteve]

Did some testing today, from cold to within 1 PPM of nominal value was 6 minutes. So most likely not an "A" version based on TiN's data.

The parts on the photo from TheSteve look glossy - is there some conformal coating (Parylene) used ?

No coating on the board, but the LT1013 is coated on top. No rework ever done to the board, it is all original.

[Dr. Frank]

Did some testing today, from cold to within 1 PPM of nominal value was 6 minutes. So most likely not an "A" version .

Nope, These 5 min are related to the stabilization of the complete circuit, and this is
practically  independent from A or non-A version.
You need to monitor the Ube of the reference element, like I did, and there's the difference.
Non-A takes 10...30sec, A version a few seconds only.
Definitely search for my contribution and study the graphs
Frank

[Pipelie]

There is an simple way to test it is ACH or CH.

For CH version, the resistor between PIN4 and the metal can is very low,  hundreds to Thousands ohm,

but for the ACH version , the resistor is really high, if you use fluke 187 to measure it ,you'll get "OL"

grab your DMM and test it, prove me wrong ,if you can.

[Dr. Frank]

There is an simple way to test it is ACH or CH.

For CH version, the resistor between PIN4 and the metal can is very low,  hundreds to Thousands ohm,

but for the ACH version , the resistor is really high, if you use fluke 187 to measure it ,you'll get "OL"

grab your DMM and test it, prove me wrong ,if you can.

That makes sense, as pin 4 is zener negative and also substrate.
If all LTZ1000 (non-A) used an electrically conducting chip glue epoxy, then pin4 is coupled to the case by the substrate bulk, having some resistance.

As the chip of the A version is thermally isolated, the chip glue epoxy is very probably an electrically non conductive material.

I'll check that in running condition, by probing the potential of the case of my LTZ1000 reference versus pin4.

Frank

[TheSteve]

The resistance between pin 4 and the can is beyond what a 34461A can measure - so a LTZ1000A(or very similar) it is. Which also makes sense as the board has no additional resistor for tempco compensation.

[Dr. Frank]

Indeed, the case of one of my LTZ1000 sits stable at +488mV, and the 120Ohm resistor at +476mV, measured with a 10 MOhm voltmeter.
Frank

[razvan784]

Does anyone know the public P/N of the MOLEX connectors being used J400 - J401 ?

22-17-2052
and they mate with 22-10-2052 on A1.

[Pipelie]

Does anyone know the public P/N of the MOLEX connectors being used J400 - J401 ?

22-17-2052
and they mate with 22-10-2052 on A1.

you can find it on Mouser, and Mouser #: 538-22-17-2052

I ordered 10 pieces from Mouser several mouths ago.

[mimmus78]

I just had a look at the A9 board on my reasonably early 3458A. Instead of being marked LTZ1000A the ref is marked LT T106382 date code 8809.

Has anyone seen this before?
Perhaps the same thing with a part # from before they decided on "LTZ1000"?

Well yes someone already have seen this before ...

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[TheSteve]

Sorry, still don't get your post. My LTZ1000 has an internal Linear part #, yours has an HP part number which is well known and documented.

[mimmus78]

Sorry, still don't get your post. My LTZ1000 has an internal Linear part #, yours has an HP part number which is well known and documented.

Yes just noticed too ... I deleted useless post.

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[Cerebus]

Hello,

strange. The LT1013 is blacktopped.

Was that a usual thing of LT at that time?
Blacktopping is something that I only know from counterfeit parts.

with best regards

Andreas

My guess is that they relabelled, with HP part numbers, some standard production LT1013s that already had LT part numbers  because, for whatever reason, they didn't have any unencapsulated dies around to fulfil an order from HP.

[Edwin G. Pettis]

It was/still is a very common practice for OEMs to have parts labeled with their in-house numbers even if they are off-the-shelf common parts, in some cases it was because of some kind of selection process.  This practice became very annoying once the OEM stopped supporting the equipment, they often didn't bother to release an equivalent's list.  In some cases I was told that some OEMs ended their runs, the specification sheets were destroyed and even the manufacturer of the parts had to destroy their copies as well so later down the line they couldn't answer any inquiries about the part even when they had made it.  Sometimes they didn't even know they had made the part in the first place, all the paperwork was gone.

Great practice isn't it?