3458A - black edition

[oPossum]

VFDs can be reliable, if done right.  I have a couple of 1970s era HP 59304A GPIB displays that use some type of gas discharge technology.
(Attachment Link)

Probably a Panaplex display. Similar technology to Nixie tubes but has a cathode for each segment rather than a cathode for an entire digit.

[TheSteve]

While we wait for high res pics we can at least see the new A3 in low-res courtesy of Keysight.

[syau]

While we wait for high res pics we can at least see the new A3 in low-res courtesy of Keysight.

U180 still there 

FPGA and associated 5V interface chips similar to those newer A3 board and the 8051 became SMD.

[MadTux]

What you could think of is *parallelizing* references for low noise. 10 chips ca. 1/3rd of the noise.

Why isn't this done on chip/die level already?
Like doing the LTZ1000 reference circuit like 16 times on a small ovenized chip assembly to get 4x lower noise than normal LTZ1000.

[Kleinstein]

On chip level a larger area device can be used instead of multiple parallel one. AFAIK the LTFU already has 2 zener structures in parallel.
For the LTZ1000 more current for the reference also gives more heat and this requires a higher temperature or more power to the heater and possibly a different case design. In principle the LTZ1000 reference could run with a slightly higher current for lower noise - but with possible downside of more drift.

For a DMM 2 parallel references have an additional potential benefit over just noise reduction: 2 reference allow some basic consistency check to detect excessive drift of one reference. So it adds to the reliability.

Because the new boards are meant as a direct 1:1 replacement, a additional 2nd ref. board would be difficult to implement. The space and power would not be so bad, but it would also need some signal routing and different firmware. So this would be more like a point to the wish-list for a possible future new design.

[iMo]

What you could think of is *parallelizing* references for low noise. 10 chips ca. 1/3rd of the noise.

Why isn't this done on chip/die level already?
Like doing the LTZ1000 reference circuit like 16 times on a small ovenized chip assembly to get 4x lower noise than normal LTZ1000.

I think the yield (the number of "good" chips/dies you get off a silicon wafer) with voltage references is low.
Therefore putting 16 references into a single chip (a single monolithic silicon die) would be risky and the yield would be much lower.
The good option would be a "hybrid chip device" or "multichip module" - ie. 16x LTZ1000 individually preselected dies put on a ceramic carrier, with some supporting stuff (thin/thick film technology), put into a larger hermetic metallic package.

[Henrik_V]

Put 20 on the die, and offer a 4/8/16 version .. laser laser

[Macbeth]

U180 still there 

FPGA and associated 5V interface chips similar to those newer A3 board and the 8051 became SMD.

Is U180 still in production? I figured they had a finite supply of old stock but with a new board and meter are Keysight actually still getting them manufactured in that old school package?

[niner_007]

U180 still there 

FPGA and associated 5V interface chips similar to those newer A3 board and the 8051 became SMD.

Is U180 still in production? I figured they had a finite supply of old stock but with a new board and meter are Keysight actually still getting them manufactured in that old school package?

Of course it is in production, how could they still manufacture new meters or sell the bare boards, if they didn’t

[MadTux]

Of course it is in production, how could they still manufacture new meters or sell the bare boards, if they didn’t

New old stock, ie they made a big load of U180 dies during production run and stashed them somewhere?
Now they only need to package those. Once they run out of U180 dies, no more new 3458A?

[Kleinstein]

The U180 hybrid chip is a main part of the ADC. Redesigning this would be quite some effort with expensive testing. So it is absolutely no surprise to still find it. With the relatively large chips inside, the old THT case may also be a good choice for a hermetic case.

I don't know if they still have a current production or just sufficient NOS, either finished or just bare dies. The quantities are still small, so if won't be production every week anyway, more like every few years a new batch.

Just for the mounting holes and cable lengths they can't make the boards much smaller - so th new board picture is not such a large surprise. Staying rather close to the old version is natural even of the new chips are smaller.

[notfaded1]

I don't see any reason why you wouldn't keep producing them... if it's not broke... don't fix it. 

[e61_phil]

I don't see any reason why you wouldn't keep producing them... if it's not broke... don't fix it. 

But it is...

The ADC drift problem seems to be a problem over many years now.

[Kleinstein]

For the drift problem it is not so clear if this is really solved. It is also a relatively expensive part to produce.
So there are some arguments to avoid the U180 chip in a new design, but for the 3458 it is the essentially part for the ADC. 
So no surprise to still see the old style U180 - it would not be a 3458 anymore without it.

[voltampere]

Not really hires....



But a bit clearer,


regards

Joerg

[Dr. Frank]

Many Thanks, 
There's the Wekomm logo in the right bottom corner, a small non occupied 'test structure' beneath, probably a hex inverter 7404, 5 inverters in parallel, driving something, maybe a 'pigtail' optoisolator (U10).
No 8051 µP, instead that ALTERA CPLD, probably 5M1270Z, everything else as described.
Let's search, which fast comparator they've used..

Frank

[iMo]

TL3016?

[branadic]

Enough evidence that wekomm made this board, nice, thanks 
Unpopulated Broadcom / Avago HFBR-1521Z and related logic gate for external clock.

-branadic-

[Kleinstein]

The comparators looks like TL3016 with additional clamping on the input, so they don't need a +-18 V supply. It is readable for U181.
This looks like a sensible choice.

The part around U181 still looks as odd as the original:  Buffer the signal with an LM358 to send it to a super fast comparator to drive a resistor of some 600 K. 
This part of the circuit never made sense to me and chances are the comparator is disabled most of the time, except during some self tests.

[TiN]

Dr. Frank
8051 uC is little TQFP chipie with a label. So everything is there

[chickenHeadKnob]

Where is the integrator cap? that part of the layout looks changed as well.

[Kleinstein]

The integration cap is C119, right at the bottom of the board.
 There are some changes to the placement, but not really much. The general layout is similar.
There are also some changes to the circuit and values: e.g. 2 diodes each at the slope amplifier, larger resistors and thus a smaller capacitor at the output of the LT1001 in the integrator.
No longer the large decoupling caps (C120,C121) at the integrator (at least not on the top side - there could be an tantalum cap on the bottom, but I doubt it).

[Echo88]

Is it correct that theyve never corrected the U180-microcrack-issue?

[iMo]

Posters close to we-komm there in DE may ask them on U180 details..

[Dr. Frank]

The statement from Wekomm was (I think I already posted it here), that U180 is still in actual production by KS and has seen several revisions over the years.