Fluke 732B DC Standard Teardown

[free_electron]

Right. The last guy that was using the standard forgot to plug it back in mains power. The end result : a cold reference and a dead battery ... Time to take out the screwdriver and pop it open. By now the Pavlovian conditioning of real volt-nuts is kicking in and they should be salivating heavily ....

You saw it first on EEVblog !

Here is the beastie:
http://www.siliconvalleygarage.com/eevblog/fluke732b/std1.jpg
Yes boys and girls , an honest to -whatever deity you adhere to- metrology grade Fluke 732B DC standard. Making an exact 10.00000 volts as well as a very accurate 1.018 volts. All provided you connect it right and treat it right.

http://www.siliconvalleygarage.com/eevblog/fluke732b/std2.jpg
Covers off. The main control board is essentially a power supply and a battery charger/conditioner. To attain its precision this machine needs to run off its built in 12 volts 7aH lead-gel battery pack. When in idle, the machine is connected to a mains supply line that keeps the battery charged. When doing the calibration you float this unit by unplugging it from the mains supply. The battery charger has a fast charge and trickle charge mode. A monitoring circuit built around some lm339 comparators monitors battery voltage as well as oven temperature. A flipflop made with a 4011 latches any fault condition in either charger / battery or oven and turns off the 'in CAL' led at which point it is -game over- : you lost your reference... a switching inverter makes a negative 6 volts as additional feed for the reference oven.

The lead gel pack is a standard battery pack as you find in motorcycles , UPs systems and alarm systems. Nothing special.
http://www.siliconvalleygarage.com/eevblog/fluke732b/std3.jpg

http://www.siliconvalleygarage.com/eevblog/fluke732b/std4.jpg
The business end of the machine : the source unit. This module is where all the black magic happens. The top has a couple of BCD switches ( these are NOT potentiometers !) that allow fine-trimming of the reference source. This can only be done if the source has stabilized for several days. The alignment is done against a primary standard using a null-voltmeter. Any fiddling with these requires a few hours of 'settling time' ... the reason for not using potentiometers is simple : the mechanical stresses and vibration may cause the wipers to shift ever so slightly giving an undesired drift. with these switches this will not happen.

Time to do a bit of stripping ..

http://www.siliconvalleygarage.com/eevblog/fluke732b/std5.jpg
The first thing we notice is a big isolated pack. That's where the oven is. A kapton foil semi-rigid PCB comes out of the foampack as well as a single teflon wire. The big grey blocks are ferrite sleeves that sit over the banana jacks.

http://www.siliconvalleygarage.com/eevblog/fluke732b/std6.jp
A close-up of the the front PCB. this pcb holds the 4 LED's as well as a 555 timer to make the 'low bat' led blink a single gas surge-arrester connects the
COM terminals to the chassis. At the bottom you see the circuit board holding the BCD switches.

It's strip-tease time.. ( By now the volt-nuts are sitting at the tip of their chair frantically waving dollar bills around .... )
http://www.siliconvalleygarage.com/eevblog/fluke732b/std7.jpg
Taking the cover off the isopack reveals .. more foam... drat...
http://www.siliconvalleygarage.com/eevblog/fluke732b/std8.jpg
Removing that bit of foam shows a massive chunk of metal with, what appears to be a circuit board, inside.
This image shows the entire interconnect. The semi-rigid PCB splits of into 2 branches. one goes to the front board holding the BCD switches. The other end goes to a stiffener board and is screwed directly onto the brass banana terminals. The original design had the 4 terminals screwed down. but apparently there is something going on as they deliberately drilled out one terminal and connected that using the white terminal. The semi-rigid cable holds a trace and you can clearly see the drilling was done afterwards. I ohmed the connection out and there is no cut in the flex , so this is not a 'repair' but an intentional modification. This wire carries the 1.018 volts. Possibily there was a leakage or some other weird effect present requiring additional isolation.

Removing the source from its isopack reveals a single power transistor bolted on the case. Surely that can't be the heater ?
http://www.siliconvalleygarage.com/eevblog/fluke732b/std9.jpg

Stripping the outer shell reveals the truth : this is the heater ! An Alox (aluminum-oxide) substrate based power resistor !
]http://www.siliconvalleygarage.com/eevblog/fluke732b/std10.jpg

Flipping the package over shows the semi-rigid interconnect and.. another heater ! hoho ! These guys aren't kidding around.....
http://www.siliconvalleygarage.com/eevblog/fluke732b/std11.jpg

http://www.siliconvalleygarage.com/eevblog/fluke732b/std12.jpg

The side shows two cavities: one that holds special temperature compensated resistors , the other holding 3 hybrid circuits.
http://www.siliconvalleygarage.com/eevblog/fluke732b/std14.jpg

http://www.siliconvalleygarage.com/eevblog/fluke732b/std15.jpg
One hybrid is soldered onto the bottom circuit board. This seems to be a simple resistor network. ( the top one in the picture )
The bottom hybrid holds a bunch of opamps , transistors, a weird metal can with 4 legs... and.. another hybrid soldered on top of it ? wtf ? hybrids soldered on hybrids ?

It turns out this top hybrid is an ultra precision resistor chain. this hybrid has resistor ladders and is covered with a quartz plate that hermetically seals them away. These resistors are laser trimmed through the glass plate. The structure is so that they can cut away segments of the resistor to increase the value.
http://www.siliconvalleygarage.com/eevblog/fluke732b/std16.jpg

The image below shows a closeup of the trimming structure. They can trim both up and down by shooting open either a horizontal or a vertical bar in the grid. This has got to be super expensive...
http://www.siliconvalleygarage.com/eevblog/fluke732b/std17.jpg

A PMI ( absorbed into Analog devices now ) precision OP97 opamp apparently does some work ...
http://www.siliconvalleygarage.com/eevblog/fluke732b/std18.jpg

The mystery metal can had a little sticker on top... the only markings on the side ar KFXFAAB 9006 ( i stuck a small mirror in there to look at the backside ). googling yielded absolutely zippo... time to peel off the little sticker.. using tweezers i managed to peel it off..

http://www.siliconvalleygarage.com/eevblog/fluke732b/std19.jpg

And tadaa : a linear technologies part. LTFLU-I CH 9006. judging from my standard the 9006 is the production date of this mystery part. the letters LTFLU-1 seem to indicate a partnership between Linear Technologies and Fluke.
Several sources claimed that these standards used the LTZ1000 , which is clearly not the case... it doesn't have enough pins.
Digging around on the internet if found some references to this part on a Japanese website. It is a device similar to a Motorola  SZA263 that was used in the older generation fluke calibrators. motorola discontinued it an Linear picked it up. It is a simply circuit consisting of 1 transistor and a zener diode in a hermetically sealed can. The black magic happens at semiconductor level. This zener diode is not constructed at the surface but buried in the silicon with the transistor built on top of it.

-edited as the images don;t show

[SeanB]

You wanted a new one .........

[free_electron]

nope. it's assembled again. has been under power for a whole day and it's dead stable... i'll let it charge thoroughly, ship it 'hot' to the callab ( under battery power ) and it'll come back 'hot' (powered up during transport ) and live happily ever after.

[ddavidebor]

Oooooh O.O amazing piece of equipment

[Rufus]

You can download the user manual from fluke with schematics and a theory of operation chapter.

http://us.flukecal.com/products/electrical-calibration/electrical-standards/734a-voltage-reference-dc-voltage-standard?quicktabs_product_details=4

[amyk]

Here you can see the resistors in the oven taken out(!):
http://bbs.38hot.net/thread-177-1-1.html
Interesting shape to them...

[Dr. Frank]

Hello free_electron,

nice that you have a 732B as a teardown object (and happily working ever after).
Thank you for sharing.

As the schematics for the reference circuitry is not officially available, also not in the 732B manual, do you have complete planar pictures of upper and lower side of the PCB(s) in the oven assembly?

Would be interesting to re-engineer the schematics of the LTFLU version.

Frank

[quantumvolt]

I might be blind or have misunderstood, but it seems that the schematic of the reference board with the encapsulated Zener+Transistor is not here?

Well - if so, this manual for 732A contains a schematic (1983) that might be similar to early 732B (?):

http://www.ko4bb.com/Manuals/Fluke/Misc/732A_Users_Manual.pdf

[amyk]

The thread in that Chinese forum above (I fixed the link) contains some RE'd schematics...

[PA0PBZ]

I was wondering about the 1.018 volts, so to answer my own question:

The Weston cell contained two electrodes immersed in a solution of cadmium sulfate, and produced a predictable 1.018 volts DC at room temperature.

[robrenz]

Very nice, Thanks for taking the time to show this.

[saturation]

Most excellent, thanks freee-!

[ElectroIrradiator]

That is some serious adult content right there.

What fate befell the individual, which let the battery run out?

[bingo600]

Damm i'm envious ... 
Why's all the nice affordable used-stuff available "over there".

I'm drooling heavily , and want one.....

Thanx for the nice teardown 

But give it to Dave for  the next contest ... hint..hint

/Bingo

[PA4TIM]

They allways tell the 732B is a totally different class as the 731A, now I see why.  My 731 A  http://www.pa4tim.nl/?p=2376  is not more as a reference, an opamp and some resistors.  I made a LM399 standarsd, still planning to ovenise the whole thing and buils it in the 731A housing.
 It is pretty stable but the temp in my lab is not. . It changes about  40 uV on 10V something like 8 degrees temp change and relative humidity. ( 0.5 ppm / C )  As soon as I can log lemp and rH% I will log the voltage deviation over time.

[branadic]

I made a LM399 standarsd, still planning to ovenise the whole thing and buils it in the 731A housing.

Do you want to share your schematics? What does it look like?

I've build the "10V Buffered Reference" from LinearTechnology datasheet with LT1001 in ceramic package with 5ppm/K 0805 chip resistors for gain setting and a crystal heater on top (40.8±1.5°C but <0.1K regulation accuracy). The reference is now running several weeks on a 34401A under lab conditions (~22.4±0.5°C) and I can observe a drift by ±10µV correlating with humidity in the lab (corner value 1 RH<40%, corner value 2 RH>45%).
Can't say if this is a sensitivity of my reference or the 34401A.
I'm still searching for the ultimate circuit. One idea could be a LM334 current source in hermetical package feeding the LM399 combined with a chopper amp at the output.
So I'm interested in your solution.

[SimonSatCom]

Very nice teardown. At the moment I'm trying to build my own precision reference using the LTZ1000A.

http://www.simonthenerd.com/LTZ1000A.htm

I have used the Fluke 732B as kind of my reference so it is nice to see some real pictures from inside the beast.

[VR2XGI]

any Fluke 732A teardown??

[VR2XGI]

Here you can see the resistors in the oven taken out(!):
http://bbs.38hot.net/thread-177-1-1.html
Interesting shape to them...

38hot is a very good forum , I have been there many year