EEVblog #427 - HP 3478A Multimeter Teardown

[BravoV]

[uprightsquire]

The 3478a that I've been cleaning up today is date coded '82 and doesnt have the optocouplers... There are also a few other small differences between the newer and older boards

[BravoV]

The 3478a that I've been cleaning up today is date coded '82 and doesnt have the optocouplers... There are also a few other small differences between the newer and older boards

Photos please, especially the isolation part using transformer instead optocouplers, must be interesting to see the differences.

PS : Suggesting not to use external hosting for the photo like you did using tinypic at your other thread. Just use this forum's image attachment feature.

[EEVblog]

My photos here:
http://www.flickr.com/photos/eevblog/sets/72157632806646675/

Dave.

[uprightsquire]

By '82 I mean '84

[BravoV]

Thanks, its obvious not only the opto couplers vs toroid transformers, also the newer 3478A board has slightly different layout.

Did a quick&dirty merging both photos between Dave's (upper one) vs yours (bottom) at the isolation parts.





PS :
Just a reminder on Dave's selling this particular HP 3478A at ebay -> http://www.ebay.com.au/itm/271158344204 , proof that Dave's is newer, hopefully this will contribute higher bidding price on this HP 3478A meter Sagan's fund, hurry.... one more day left.


.

[EEVblog]

BTW, my board is marked Rev A 

Dave.

[uprightsquire]

BTW, my board is marked Rev A 

I have rev c.

[Zad]

If anyone is interested, I did a tear-down of my Tektronix DMM4020 (a Fluke 8808a by any other name). No custom part numbers in this, no prizes for guessing the voltage reference!



Lots of optical isolation on the serial driver section. It even had its own isolated feed from the custom main power transformer.



As with the HP3478A, the PCB has segregated areas for digital and analogue systems:



Just 4 Philips screws and a cal sticker to remove to get into the DMM4020, I love equipment that is clearly designed to be maintained over a long timescale.  There are a few more photos here: http://s222.photobucket.com/albums/dd237/Zadpics/Tektronix%20DMM4020%20Bench%20Multimeter/

[Jonny]

How tall is metal shield beside the crystal? Does it meet the lid to form more of a box over the circuit?

[EEVblog]

How tall is metal shield beside the crystal? Does it meet the lid to form more of a box over the circuit?

No. Maybe an inch or so tall.

Dave.

[c4757p]

The 3468A has a full can over the analog section. Interesting that they thought it was not necessary for the higher-spec 3478A.

[BravoV]

The 3468A has a full can over the analog section. Interesting that they thought it was not necessary for the higher-spec 3478A.

Probably because the whole 3468A's case is all plastic instead of metal ? 

[c4757p]

  Yeah, that would be it.

[BravoV]

Excluding cost, physical size and ease of installation factors, technical wise whats the pro & con for using opto coupler vs inductive for the isolation like above ?

Also just noticed at the toroid windings, example at the red & blue windings (are they pri & sec ?), they're not even wound in a complete single turn, looks like only half turn ? 

[obsoletemac]

Nice video, thanks Dave.
There seems to be no line voltage selector on the 3478A. Is it possible to rewire a 120V unit for 240V without replacing the transformer? i.e. are there dual primary windings?

//c

[pgross]

Yes, it is possible to change the mains voltage.

Remove the back bezel and slide off the case. Underneath the ribbon cable to the HP-IP port there is a single pole connector and some pins for the input voltage ranging from 100-240 V.

The actual line frequency (50 or 60 Hz) is selectable by the small dip switch on the back.

/ Peter

[staxquad]

Dave's unit has a ser. of 2911A65826, falling into the post 2619A37795 units with a different design to the pre, and part of the serial could be the date code, the 1st, 2nd, 3rd and 4th digit, SN : 2911A66xxx - Means : (29+60 = 89) = 1989 , Week 11 , A(merica) (info from bingo600)
Also, if you don't know the serial no, but "TERMINALS" shows up instead of "TERM" for the front-rear switch front panel, it's a late unit (for those aiming to purchase a later model).

[Wartex]

Excluding cost, physical size and ease of installation factors, technical wise whats the pro & con for using opto coupler vs inductive for the isolation like above ?

Also just noticed at the toroid windings, example at the red & blue windings (are they pri & sec ?), they're not even wound in a complete single turn, looks like only half turn ? 

high voltage AC will make it thru the inductor

[robbak]

Excluding cost, physical size and ease of installation factors, technical wise whats the pro & con for using opto coupler vs inductive for the isolation like above ?

Also just noticed at the toroid windings, example at the red & blue windings (are they pri & sec ?), they're not even wound in a complete single turn, looks like only half turn ? 

Add ease of use, and speed: You can drive the LED in an optocoupler straight off the data bus, and the signal from the photo-transistor on the other side is ready for the data bus. The toroids need driver circuits to drive them and sensor circuits to turn what comes out of them back into a useable signal on a data bus.
Power, of course: those toroids would need a reasonable power level to drive them.
And speed: the inductance of those toroids would limit the clock speed that you could use them on.

[Hypernova]

Two down sides for opto's:

1) Life time, for fast signalling if you choose to drive the LEDs near their rated current they fail after a few years, for hobby use this is generally negligible.
2) Size, compared to other IC scale isolators (capacitive from TI and transformer from ADI) opto's use air gaps for isolation, for a given isolation rating they may be bulkier.

[bingo600]

Damm i have hoped that Dave did replace the battery 

I have one that soon needs a battery replacement , and was thinking of using my propane iron , and the leave it powered on during the change.

Dave i think the guy buying this meter needs a "Fresh battery" installed ... Hint..Hint .. Don't you


/Bingo

[BravoV]

Dave mentioned the measured voltage was at 3.49 volt which is a big possibility it is a lithium thionyl chloride type (LTC) with typical at 3.6 volt instead of lithium sulfur dioxide type at typical 3 volt.

As these old HP DMM are using static cmos ram for keeping the calibration, the absolute minimum threshold for retaining the data at the chip is 2 volt. 

Below the typical LTC cell discharge curve, it has a really scary vertical voltage drop when its going to die. 

[IvoS]

I was also thinking to plug the unit to mains but disconnect the ground prong during battery replacement.

[Smokey]

Wow.  The ebay auction ended and the thing sold for AU $355.00.

http://www.ebay.com.au/itm/271158344204

Did you have it gold plated before you put it back together or something?  You at least autographed the case right?

[EEVblog]

Wow.  The ebay auction ended and the thing sold for AU $355.00.

Yep, more than double what I paid for it.
That's ebay Australia for you, most likely not a fan, they'd have no idea who I am.
I've seen this time and time again with stuff I've sold on ebay, people will pay silly prices to buy it from someone local in Oz instead of from overseas.
Almost every bit of test gear I've bought and sold on ebay, I've sold it higher than what I paid for it (inc postage).

Dave.

[triac]

they're not even wound in a complete single turn, looks like only half turn ? 
What you see is classified as one turn.

[belba]

Hello,

at http://matthieu.benoit.free.fr/cross/competitive/agilent/hp-part.htm you can find a cross-reference list of many HP marked ICs. I think it could be useful.

Guido

[Buga]

Why do these high end bench meters wander around in the millivolt range where as the fluke handhelds seem to hold a zero when turned on? Is it really that sensitive to noise

[c4757p]

This meter, like many other bench meters, has an input impedance specified to be "greater than 10 G" (yes, gigohms) on the low ranges. That's not enough to dissipate charge that it picks up. (If you manually switch it to a higher range, it will settle to zero.) Most of the handheld meters are 10M on all ranges.

[IvoS]

Does anyone know where I can buy a new battery for this meter?

[Dave]

Farnell, Digikey, Mouser, anywhere really.

[reagle]

Just got one for work, and did not realize that it will not take standard shrouded DMM leads. Is there a special "HP" type of leads or do I need to go down the path of chopping the shrouds off?

[c4757p]

I just chop off all the shrouds. I'm not an electrician, the times when I use any voltage above 40 or 50 or so are very rare, and I would never yank an energized plug anyway. Any time I do use dangerous voltages, I prepare an entire test setup with the power completely disconnected, power it up, and all I allow myself to do without shutting down the power is push buttons. I've got too much older stuff that won't take shrouded plugs to mess around with them.

[reagle]

It's just a shame to cut brand new Pomona probes, but oh well
Oddly enough, my 884x flukes of the same vintage take shrouded probes just fine. Probably due to higher voltage rating on the inputs or maybe just a bit newer

[c4757p]

I think it's probably due to Fluke. They had experience making handheld stuff for use by electricians, so I imagine they'd upgrade safety sooner. HP's been mostly lab equipment with few exceptions.

[SeanB]

Frankie ( Iloveelectronics) has the adaptors in his Ebay store cheap. Then you can leave the leads intact and place the adaptors in the unit permanently.

[bingo600]

Dave mentioned the measured voltage was at 3.49 volt which is a big possibility it is a lithium thionyl chloride type (LTC) with typical at 3.6 volt instead of lithium sulfur dioxide type at typical 3 volt.

It sems to be a Panasonic 3v Lithium Cell , whatever that means. LiIon , LTC or
I just changed mine , pict here.
https://www.eevblog.com/forum/beginners/changing-battery-in-a-hp-3478a-can-i-use-a-3v-lithium/msg205742/#msg205742

/Bingo

[Wim13]

Dave mentioned the measured voltage was at 3.49 volt which is a big possibility it is a lithium thionyl chloride type (LTC) with typical at 3.6 volt instead of lithium sulfur dioxide type at typical 3 volt.

It sems to be a Panasonic 3v Lithium Cell , whatever that means. LiIon , LTC or
I just changed mine , pict here.
https://www.eevblog.com/forum/beginners/changing-battery-in-a-hp-3478a-can-i-use-a-3v-lithium/msg205742/#msg205742

/Bingo

No it is a: Panasonic Poly CarbonmonoFluoride Lithium batt. 3.0 Volts to the datasheet.

I replaced the batt in mine 3478, and there was also the same  panasonic 3 V, see datasheet attached..

But why is the voltage 3.4 Volts.., same as above. The datasheet says 3.0 Volts..??
does the battery charges itself with zero point energy ?

The unit i have is from 1992, so he batt is 20 years.

Just repalces it by a HQ 910a, also 1200 mAH

[Rutger]

Does anyone know where I can buy a new battery for this meter?

I got mine at Battery Plus here in the US for about $ 5.00, local pickup so no shipping cost.

[madshaman]

It's just a shame to cut brand new Pomona probes, but oh well
Oddly enough, my 884x flukes of the same vintage take shrouded probes just fine. Probably due to higher voltage rating on the inputs or maybe just a bit newer

Pomona makes leads that have retractable shrouds for this purpose.

[george graves]

This meter, like many other bench meters, has an input impedance specified to be "greater than 10 G" (yes, gigohms) on the low ranges. That's not enough to dissipate charge that it picks up. (If you manually switch it to a higher range, it will settle to zero.) Most of the handheld meters are 10M on all ranges.

That's crazy.  I had no idea. I too was taken back a bit when I first turned it on.  "why the heck is it floating like that?!?!?"  Anyways -  I just picked one up on ebay for about $130 USD.  Calibrated last in 2006 - tested it a bit today - calibration seems spot on.  That's fine for my needs.  Thanks Dave/EEVblog for turing me on to this meter.  Seems like the best bang for the buck out there in the 5 1/5 digit meters.

Two things I've been thinking of...

1.) I do wish it had a back light display.  I'm tempted to pull it out and see if I can't shoot some leds into the side of the glass.  But at $60 for a replacement -  eeek - not sure I'm brave enough to try.

2.) For 4 wire resistance measurements - I know I can use any leads I want - but why are Kelvin leads better?  And is there a poor-man's alternative?

[EEVblog]

2.) For 4 wire resistance measurements - I know I can use any leads I want - but why are Kelvin leads better?  And is there a poor-man's alternative?

Nothing special about "kelvin leads", it's the "kelvin connection" that is important. i.e. those two sense wires, they have to be connected directly to the exact point you are trying to measure. How you do that depends on what you are trying to measure.
Simply having 4 wires connected to normal crocodile clips, or say some sharp probe pins is a good enough poors mans version for general use.

[alm]

Kelvin leads are nothing special, I guess the most important reason why they're expensive is the limited market. The main thing they offer is convenience: clipping four alligator clips to each resistor gets old real fast. It's also hard to get four clips on a tiny 0805 resistor.