HP 3455A teardown

[daqq]

Hi guys,

I took a few snaps while inside my "new" meter:
Calibration module and misc: http://www.daqq.eu/?p=577
The logic and display: http://www.daqq.eu/?p=645
And all the analog goodness: http://www.daqq.eu/?p=695

I also recommend reading the service manual - the theory of operation especially is good reading.

The attatchement: The calibration module.

[alm]

Nice teardown!

This 1977 HP Journal issue which discusses the design of the 3455A is also an interesting read:
www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1977-02.pdf

[kilobyte]

I have here also a 3455A but it's defective.

There are two version of reference PCBs.

I have attached some pictures of the 11177B Version

[Vgkid]

Very nice teardown the analog side reminds me of my 7065, but that uses no custom components at all.

[TonyGreene]

The HP3455A is an economical precision meter.  I have seen them go for an average of $100 US.  For 6 digits....its hard to beat.

It does use a LM199A or LM399A voltage reference - HP used house part numbers on a lot of their products.

The fan is a problem, when it goes.  Its a three phase unit - there is an on board inverter/ring oscillator that feeds it.  And the fan is not made any more.  I found a small 12V DC fan from a power supply that fit the hole, but I had to mount it on the outside of the case.  Also when it ran off of 12 volts, it was loud.  I ended up mounting a 7808T voltage regulator inside and feed the fan, and it moves enough air and is quiet now.

This unit can see down to the microvolt level at 1 volt range, and my unit appears to have about a 3 microvolts noise, so it bobbles 3 digits.  This depends on how clean the voltage reference is.  Also as old as these units are, the voltage references have burned in and the drift will be minimal.  I had my unit calibrated 4 years ago, got before and after data, and then I had it calibrated in April this year.  The 1 volt range had changed by 8 microvolts - so I seen a drift of 2ppm a year !

[c4757p]

That fan appears to be the same unit used in the 3437A, which I just tore down. Any old 60mm fan will fit - I used this one, which is loud as hell but puts out a lot of air (the original fan is supposed to blow 35 CFM), with a 56/1W resistor in series, a 2200uF capacitor to smooth the rectified DC (the 3437A ran the fan oscillator off a rectified but unfiltered transformer tap), a common-mode ferrite around the wires and a 470uF capacitor directly on the fan.

[grenert]

What kind of resistors are those?  I've never seen a resistor package like that.  With the low tempco, I figure they are probably wirewound.

[Vgkid]

All of the round precision resistors would most likely be wirewound. I will check, and update this post in a few minutes.

[daqq]

I acquired it for around 100 EUR. A bargain, considering the parameters. The only thing I regret is that it does not have current measurement :-(

There are two version of reference PCBs.

The service manual actually lists 4 different reference PCBs - one yours, one mine, two revisions of each. See page 224 of the PDF of the service manual.

And the fan is not made any more.

Which is a good thing - the noise the whole setup makes is just horrible. That's my ONE objection to the whole design.

What kind of resistors are those?  I've never seen a resistor package like that.

Which ones? The metal case ones? I'm guessing that they are wire wound. They are in the high precission/stability stuff - see Dave's teardown of the resistance standard.

[sipo75]

I have here also a 3455A but it's defective.

What are the symptoms?

[kilobyte]

I bought this as defective with the hint that all LEDs are on after power on.

The first issues are dried out electrolytic capacitors of the voltage regulators.
So I have replaced all Elco's and after this I was able to run the selftest which says all and noting 

And a first investigation looks like a defective FET for the VREF Voltage on the Input/Autocal section, because the voltage drops down to around 8V.
After I desoldered the affected FETs the VREF was spot on 10V.
So far i had no time to continue repairing that big monster DMM.

[saturation]

Nice teardown.  Your unit is sooo clean.  The 3455a was the earliest in the series of system multimeters, HP's top bench DMMs.  Its technology was building block for the 3456a, 57a, and finally the 58a.

[alm]

Earliest? I would argue the 3455A is a direct successor to the 3450A. Which might be related to the 3440A plug-in DVM.

[sync]

The 3460A DVM is the predecessor of the 3450A.
AFAIK the chronological order is
405AR
3440A
3460A
3450A

[daqq]

Your unit is sooo clean.

That suprised me as well. While there is a air filter there infront of the fan, still, it's really clean. The analog part gets no air flow, so that one's understandable, but the digital part should be really messy.

Maybe it just hasn't seen a lot of use. I'll try to get the story of that meter - some of the more exotic tech/specialized that ended up in Czechoslovakia has some interesting history. One of my old professors had a US mil spec IC (a processor) from the early 80s about which he told a story:

3 pieces of it were smuggled into Czechoslovakia (Communist east block at the time) for reverse engineering at obscene prices through black market and who knows what. The first one was killed by part of the chemical process - as in really beyond recognizing the connections and reverse engineering. The second went that way as well. So they gave up and had the last one. It changed hands within Tesla (the state electronics company) and ended up with him.

Possibly true, possibly false, but interesting story anyhow.

[saturation]

One could say its a direct successor if, IIRC, the 3450a was top DVM in the HP line back in 1970.  However, I don't think there was an HPIB bus then so 'system' use would be whatever ad-hoc network one used to control these devices.  Also, this DVM was hard wired, rather than use some early microprocessor.

But my criteria considers the existing top DMM today, the 3458a, where did its technology evolve from?  What models preceded it that help make it what it is today?

Per the HP Journal, the core of 3458a, its accuracy, began with the 3455a. 

Also, the 3455a was among the first microprocessor driven DVM that used the HPIB bus and had 'Math functions'  [ e.g. stat functions etc., ]  that are standard in the 345x series: 3456a, 57a, and 58a.

Earliest? I would argue the 3455A is a direct successor to the 3450A. Which might be related to the 3440A plug-in DVM.