Mark Sims | 29 Jan 01:41 2015

plastic caps on 3458A reference board

Most low-end 3D filament printers can print in either PLA,  ABS,  or even nylon.   PLA prints are known to
distort in a hot car.  ABS is extruded at around 250C.  I think nylon is a bit higher.   I would think that ABS
would work well.    There are now some low-ish cost  (like under $3000) stereo-lithography printers
available that print by scanning a liquid resin with a laser.  Some of those resins are good for higher
temperatures.  The low end resins have a glass transition temperature of around 100C.

Another possibility is to make a mold and cast the parts out of epoxy.   Epoxies that can handle over 400C are
Charles Steinmetz | 28 Jan 23:06 2015

Re: plastic caps on 3458A reference board

Dave wrote:

>If you want thermal insulation, then PTFE ...

John wrote:

>I[t] needs to insulate electrical and thermal and metal would not do well.

Two machined pieces of glass-filled PTFE with holes to pass screws or 
plastic rivets through should work pretty well.  You want some 
thermal insulation, but not too much (there has to be a heat flow 
path to ambient to allow the heater to work at a reasonable fraction 
of its capacity), so the thickness of the top part would need to be 
right and some experimentation might be required.  I've made a number 
of machined parts out of plain PTFE and glass-filled PTFE -- either 
one is definitely do-able.

Straight PTFE is subject to cold-flowing, so if you used that some 
thought would need to be given to the through-bolts so everything 
would stay tight.  Belleville washers are one common solution.  But 
glass-filled PTFE is harder and more stable, and is the better choice 
for this job.

I've never machined polysulfone, but it is routinely done, so that is 
another choice.  It is also available with glass filling.

Once a design was set and programmed, a small CNC mill could zip out 
enough parts to fill the next 100 years' volt-nut needs in ten minutes.

For those who haven't seen one, here is a top view of a 3458A 
(Continue reading)

dan | 28 Jan 22:48 2015

Re: plastic caps on 3458A reference board


I wouldn't completely discount 3D printing due to temperature. There 
are companies claiming to print the high nickel super alloys. 
It's questionable whether the plastics are available in the form 
needed, but laser sintering technologies appear to be out there now. 

If stock is available in raw form, for a few 'one off' items CNC would 
probably be the way to go. 

Anyone have a few bars of this stuff laying around? ;) 


> The problem is that an LTZ1000 (or LM399) runs hot enough that most 
> plastics you find lying around won't stand up to it for very long 
> (some of them melt almost immediately when you power it up). This is 
> true even of polycarbonate, which is one of the most durable common 
> plastics, and is why the original caps were made from polysulfone. 
> Unfortunately, it also means that replacements may be hard to make by 
> 3D "printing," since the process depends on melting the base material. 
> Best regards,
> Charles

(Continue reading)

Frank Stellmach | 28 Jan 22:17 2015

plastic caps on 3458A reference board


my guess for poly-sulfone for the top cap was not that bad:

The HP LTZ circuit, and also the LM399, were heated to about 95°C, and 
therefore, you need a thermally insulating plastic, which can withstand 
that temperature.

Poly sulfone is able to do so, whereas poly-styrene is OK for 70°C 
maximum, like in 45°C designs.

It would be OK also for that amateur purpose, if it's thick enough: The 
inner layer would deteriorate over time, but the outer side would still 
isolate enough.

If the whole PCB is placed inside a shielding box, the solder side cap 
can be omitted, of course.
I would place a thin sheet of poly styrene underneath the PCB to 
thermally insulate the circuit.

Frank Stellmach | 28 Jan 12:28 2015

plastic caps on 3458A reference board

Hello Joe,

yes the cap consists of two parts.
The upper one serves for thermal isolation of the LTZ1000A TO99 case 
against the environment and therefore reduces the power dissipation of 
the oven.

The part on the solder side is much more important, as it covers the 
solder joints and avoids air draught over these pins. That avoids these 
low frequency voltage variations, which are mentioned in the LTZ data sheet.

Anyhow, there are no further shieldings around the PCB, so the solder 
joints of the OP Amp and the precision resistors are exposed to that air 
draught, maybe from the fan.
That's a further engineering fault they made on this reference.

(The other faults are the 95°C oven temperature, the use of the A 
version instead of the non A, and the use of R417,200k temperature 
compensation resistor, which is necessary for the non A version only.)

The hat is not included in the BOM inside the CLIP, therefore can not be 
ordered from HP, obviously.

It is a smooth, shiny plastic, resembles the one used around the LM399H.
Wasn't latter one something like polysulphone?

Maybe suitable pieces of polystyrol foam will do the job also.

(Continue reading)

Frank Stellmach | 27 Jan 17:17 2015

3458A reference boards on ebay

I forgot to mention, that these original hp reference are a bargain for 
around 100$.
LTZ1000 /A cost 35.. 55$, 3 PWW or MBF for 10..20$ each, plus the 
additional stuff.

They only need a thermal shield around the LTZ1000A on component and 
solder side, the pimping to 55°C (by 100k PWW in parallel to the 15k 
BMF), a shielding, and a PSU.

The circuit is the same as in the LT datasheet, that's totally 
sufficient for stability.
And it's very compact, easing the shielding.
No fancy slots or other gimmicks for the PCN and components..
The stability is determined mainly by the LTZ1000 and its oven temperature.

Therefore, that's a recommendation.

Frank Stellmach | 27 Jan 16:30 2015

3458A reference boards on ebay

Hello fellow-nuts,

do you already know about our monster discussion on DIY LTZ1000s on EEVBLOG?

There are several people who have successfully assembled such boards, 
and they are pretty stable, much more stable than the HP ones, and the 
precision resistors are also easy to obtain, if you use precision 
wire-wound, which are at least as good as the bulk metal foil, for this 

The 002 reference is the same as the standard one, but is selected for 
lower drift.. assuming these run 24/365.

Otherwise, the may show big hysteresis and have to re-initialized for a 
period of time, to reach the same level of stability. standard boards 
are rumored to achieve the same level of stability after some years, also.

In DIY designs, but also in these HP references, simply reduce the oven 
temperature to 55°C for the LTZ1000A, and to 45°C for the LTZ1000. That 
will typically improve their stability by a factor of 2 for each 
reduction of 10°C, compared to the about 95°C, the original HP reference 
inside the 3458A is running (due to max. environmental temperature spec.)

To use these boards externally, simply put a metal shield around, give 
it a good ground, and a quiet, linear regulator (12V or so), and you're 
done. Once measured, you get an extremely stable (t & T) 7,1xV reference.

Amplifying to 10V by a ChopAmp will require precision resistors, which 
will account most for the annual drift, then.
(Continue reading)

Orin Eman | 27 Jan 10:16 2015

3458A reference boards on ebay

There is a seller letting a slow trickle of them out there... they seem to
be going for about $165.  The current batch of two is around $100, but will
probably be bid up.

I just received one that I won.  Seems to be working fine after a quick
breadboard lashup.  I'll be making an enclosure for it next.

A little expensive IMO, but given the trouble getting the precision
resistors to DIY with the LTZ1000A, probably worth it.


ESI SR104 Standard Resistor for sale

ESI SR104 Standard Resistor for sale at 1500 EUR in the Netherlands
the ultimate 10k standard.




Jay Walling via volt-nuts | 21 Jan 12:52 2015

EDC 522 calibrators going for cheap on Ebay

Hi all,

Just thought I'd give a heads up on this.
There are two EDC 522 DC voltage calibrators on Ebay right now going for cheap ($250 to $300)

I just bought and received one from "pepperellgeek" for $200. Works fine and I'm very happy with it.

No affiliation with either seller.

Mark Sims | 20 Jan 06:16 2015

HP3458A cal ram data dumper program

John Miles has included my  HP-3458A cal ram data dumper program in the latest release of his excellent GPIB
Toolkit package.  You can download  it from the KE5FX web site.  The released version produces the formatted
listing of the various cal RAM values in its output (it is the same version of the  data dumper  that I put up on