HP34401 teardown

[free_electron]

coming in a moment ... on air now ...

THe HP / Agilent 34401 has been around for a long time and is considered by many to be the 'gold standard' of the 6 1/2 digit class meters. It only shares this class with the Keithley 2001 for its time frame.
The newer generation machines like hte 34410 and 34411 share with the Fluk 88xx series ( see my other teardown )
These machines are only surpassed by the 3458A 8 1/2 digit meter that is in a league of it's own. even though Keithley has the model 2002 it doesn't come close in terms of throughput. The 3458 is the calibration lab's workhorse and used in many an ATE setup for mass volume testing due to its very high digitizing speed. THe 3458 is, literally, the machine used to calibrate the calibrators...

The frontpanel is a familiar sight for machines of this product line. Rubber mebrane keyboard and a starbusrst type 14 segment VFD display

The front panel has its own 87c51 to scan the keypad and refresh the display. A Sipex high voltage column driver does the VFD interfacing. The communication is level shifted on a 18 volts common rail. Mine is open here since two segments were shorted. i had to replace the driver chip.

Decent test equipment is always galvanically isolated and this machine is no different. An opticvally isolated inguard/outguard design we find two optocouplers bridging the domains.

The outguard consists of a 87c51 and a mp9914 GPIB controller.


THe brains of the machine are formed by an intel 80c196 16 bits processor and an asic common to machines of this lineage. The asic does frontpanel operations, contains uarts and has the a/d control logic on board.  A static ram and eprom complete the computer part.

This machine uses a charge balancing technology vcalled Multislope III.
I Have given extensive explanation on its working in another review of a power supply that uses the same approach.


The truerms converter uses the classical AD637 chip. This particular machine has a problem with a fried tantalum cap. Tantalum caps are notorious for this kind of behavior. Any tantalum capacitor , treated with kindenss and respect, may , for whatever reason, spontaneously combust for whatever reason it feels like. that, and they may as well call them unobtainium capacitors. the worlds largest tantalum mines ( australia ) have shut down , opened again , shut down again and the future of this mineral is unclear. There have been shortages and the prices of these parts wildly fluctuates... That's why they are being phased out ...


The entire analog portion of the system occupies most of the circuit board and is normally hidden under a metal shield cover. Precision and stability is derived from a thermally stabilized buried zener of the LM199 type. ( whit round thingie )

The entire precision analog part is contained in two custom ASIC's made by agilent in their own fabs.

The large ceramic part is a precision hybrid containing the trimmed resistors for all ranges.

The PLCC device holds the analog switches and current sources. there is a smaller plcc device to the right that has some other amplification and switching circuitry. I still need to clean off the flux of the PLCC and the two opamps. This machine suffered catastrophic failure but it's up and running again now. Not bad for something fished out of the garbage skip ...
I collected the asic from a multimeter module out of an 34970 that had its brains fried ... someone plugged it on 220 while the switch was on 110... Those machines have the same multimeter guts as a 34401 but in a different form factor.


The input section holds a real quality mechanical switch to select between front and back and has latching relays for certain range switching as well as a fully shielded COTO reed relay ( big red can ). Plenty of protection is available in the form of MOV's ( blue drums ) and fat diodes ( left of yellow relays: the bridge and two big diodes above ).

[Jad.z]

Are you passing your moment through a Bose–Einstein condensate?

[free_electron]

other two reviews took long....

[JoeyP]

Gosh I'm getting old. I remember when the *HP* 34401 was the shiniest new great thing. Hard to believe that was nearly two decades ago!

The truerms converter uses the classical AD637 chip. This particular machine has a problem with a fried tantalum cap. Tantalum caps are notorious for this kind of behavior. Any tantalum capacitor , treated with kindenss and respect, may , for whatever reason, spontaneously combust for whatever reason it feels like. that, and they may as well call them unobtainium capacitors. the worlds largest tantalum mines ( australia ) have shut down , opened again , shut down again and the future of this mineral is unclear. There have been shortages and the prices of these parts wildly fluctuates... That's why they are being phased out ...

That may be one reason, but the availability of modern high-value MLCC ceramics with much lower ESRs, coupled with higher frequency SMPS was another nail in the coffin of those old tantalum lemons. I refused to use them in a design, having replaced hundreds, maybe thousands of shorted/burned tantalums during the 1980-90's.

[free_electron]

Jep. MLCC are better than tantalums , except for systems where microphony or vibration is in effect. MLCC's are piezo-reactive and will pick up the vibrartion and produce a voltage.

NEVER use barium-titanate based ceramics (Class-2) in feedback loops if the board is subjec to vibration. Class-1 is OK ( C0G or NPO type are class-1, X7R, Y5V, Z5U and others are class-II).

Tantalums are finickey thingies. I onse say a whole DSLAM go up in flames . We had a couple of big SMD tantalums. 220uF in D case. these were mounted 4 sided by side. somewhere in the center of the rack ( the rack holds identical boards space 1 inch apart ) one of these caps decides it doesn;t want to do its menial job anymore and spontaneously combusts.. It sets its neighbours on fire , burns a hole through the board and sets the caps on the adjacent board on fire as well... by the tie we could react the 'flame' had burned 5 boards... We hit the emergency stop and pulled out the board that wasn't affected yet to stop the cascade.

In 2000 the first big tantalum shortage hit becasue some aussie tantalum mine shut down. (Talion) and prices went through the roof. there's few producers of this stuff. Brazi, the aussies, the chinese ( they don't export raw tantalium. only caps. ) and then there is the conflict states like the Congo...
And that's it.

Anyway. MLCC is the way to go. tantalum has no viable application anymore ( unless you are dealing with wet-tantalum for military , hi-rel apps )

[coldframe]

Hi,

(Long before) I Have Question

On the top of Electrolytic capacitors Kapton™ Tapes
What is the Purpose?

Thanks~

[echen1024]

Ah yes tantalums. My good friend. Purchased a few on my Zhongguancun raid in China when I was first starting out in elementary. Hooked one up backwards, and BOOM! scared me shitless. Vowed never to touch those things again.

[free_electron]

Hi,

(Long before) I Have Question

On the top of Electrolytic capacitors Kapton™ Tapes
What is the Purpose?

Thanks~

probaby has to do with the soldering process. the top may get really hot in the reflow oven. kapton is good isolator and blocks infrared.

[Andreas]

Hello,

can you do a closeup of the LM399 reference (U403).
Does it come with a socket or is it soldered directly to the PCB?

with best regards

Andreas

[HighVoltage]

Hi,

(Long before) I Have Question

On the top of Electrolytic capacitors Kapton™ Tapes
What is the Purpose?

Thanks~

probaby has to do with the soldering process. the top may get really hot in the reflow oven. kapton is good isolator and blocks infrared.

I have seen this on many other HP/Agilent instruments and always wodered.
This is the first explanation I have heard, that makes sense.
Thanks

[Andreas]

This is the first explanation I have heard, that makes sense.

But usually the reflow process is done before the leaded devices are mounted....
And they are usually soldered  by wave soldering from the other side.

With best regards

Andreas

[HighVoltage]

This is the first explanation I have heard, that makes sense.

But usually the reflow process is done before the leaded devices are mounted....
And they are usually soldered  by wave soldering from the other side.

With best regards

Andreas

Also true.
So what would your explanation be for the tape on top of these capacitors?

[SeanB]

I have a few of those 637 chips around, in 14 pin cerdip. Will have to dig them out and see if anybody wants some.

[djacobow]

I'm thrilled to find out that my "new" HP 34401A has an 80C196 processor in it. I worked on that family at Intel back when dinosaurs still roamed the earth. It was an interesting architecture, where the main register file was huge by contemporary standards and also mapped directly into lower memory. I worked for a division at Intel whose mission at the time was essentially to find something useful to do with older fabs and the 196 was part of it. The 196s went mostly into HDDs, and Intel considered it a good investment to provide parts for that industry considering you didn't ship an x86 processor without a hard drive.

I never knew HP had adopted them for their gear.

[pgross]

Thanks for the Teardown.

Actually, the fried tantalum cap makes me wonder if these caps are adequate for the task.

The tantalum caps: C311(fried), C312, C322 and C323 - are all 22uF/20V types, but they are exposed to
15V. Good – and normal practice when designing power supplies with tantalums is to Voltage derate by 50%.

In other words, these caps should be retrofitted with 35V types to follow the rule of thumb and maybe save the
instrument from melt down. I don’t know if it makes matter worse if the ambient temperature rises with respect to
tantalum voltage specs.

Best Regards

Peter