Electronics > Metrology

Teardown: Valhalla 2707A

(1/4) > >>

Sometimes you have to take a chance. Such as the other day where a Valhalla 2707A programmable precision dc voltage standard was for sale on eBay. Apparently a model similar to the 2701C but no information is to find on the entire Internet about the 2707A. No specifications, no YouTube videos, no nothing at all. Should I bid $500 or not? 

You have probably already guessed the outcome. I won (with no competing bids) the Valhalla 2707A which is the subject of this mini teardown. 500$ plus another 300$ tax and shipping. Did I overlook something? Suggestions whether this is a bargain or not is welcome.

After nearly two weeks the unit arrived from the US and frankly it's stupid not to turn it on before you take it apart. So I did. The unit turned on without funny sounds and odors and a short test revealed that the unit was completely operational.

This is the unit. Besides the model number it looks exactly like the 2701C which is still being sold.

All voltage ranges work, and compared with my Keysight 34470A the 2707A agrees within 21 ppm. Excellent.  However, no "120 mA current source" was installed. What a pity.  :'(
Now turn it off and take it apart.

Now teardown time.

The picture shows the main board and it appears to be in excellent condition. Judging by the date codes on the components, it's from around 1999 or a few years later. The unit is clean and in good condition. No questionable repairs.

After the initial euphoria of the unit being in complete and in decent condition the search for the LM199/399 reference began. Where the heck is it?  Checking Master Shahriar's tear down of the 2701C points me in the right direction.


But :wtf: - no reference is installed! This can't be true. The unit outputs voltage close to specification!  Zooming into the PCB where the reference is supposing to be, reveals wire leads that is soldered from the bottom of the PCB. What’s going on?

Turning the unit upside down provides the answer.

An aluminum box probably enclosing a voltage reference mounted beneath the main board. Very exciting. Kind of a Chinese box.  Details on the following pictures.

Let´s crack it open as Dave would have said. 

But you will have to wait until the next post.

Ready?  Aluminum box opened here

Isolation removed on the next picture:

Now let's remove the three screws and flip the PCB around:

What a voltage reference. Two identical LTZ1000 (non A versions) on one PCB, loaded with Vishay precision resistors and an Analog Devices AD708B dual precision amplifier.  All thermally isolated from ambient with polystyrene (or similar) plates. Note the date codes on the LTZ1000.

PCB is marked Alphanetics Inc. An engineering company still around. If anybody know this company, please let me know.

Now it's reverse engineering time. But you will have to wait until the next post. :)

This is too cool.
so someone fill me in; why two LTZ100s?

Here follows a crude schematic of the reference.

Two identical references whose outputs are averaged to reduce noise. Each LTZ1000 reference section is built almost exactly as prescribed in the LTZ1000 data sheet. Both LTZ1000’s are properly raised a millimeter above the PCB. Nice attention to details!

70K resistors are made up of 50K in series with 20K, both 0.01% Vishay resistors. The "120 ohm" resistor is two 0.01% 250 Ohm Vishay resistors in parallel resulting in 125 ohm which is the only deviation from the LTZ1000 reference circuit in the datasheet, I have found.

The important temperature set point resistors 13K/1K resistor pair is interestingly an ordinary 13K/1% resistor that is marked with a small number as if it was selected. The 1K resistor is a Vishay Mill specified 50 ppm resistor marked RV60C1001B.  Question is if this pair is TCR matched as it should be. Otherwise there is room for improvement.
Lowering the operation temperature could also be considered.

The presence of a low noise high stability reference in this unit indicates that the rest of the hardware actually can benefit from not only a single LTZ1000 reference but two with reduced noise.   This comes as a surprise as the noise specs for the LM399 based 2701C isn’t all that great as I recall.

But this is of cause wideband noise that tends to be averaged during DMM calibration. The dual LTZ1000 reference will probably greatly reduce noise below 10 Hz compared to a single LM399 which is of great importance when used as a DMM calibrator or voltage transfer standard.

With respect to stability it’s interesting if the PWM based voltage generation now becomes the weak link. The crystal might be an ordinary 30 ppm type.  Absolute precision is unimportant but most ordinary fork type crystals have temperature coefficients of about -0.05 ppm/C. Aging is often better that 3ppm/year. However, Valhalla could have selected a crystal with much better specs in the model.  Would an OCXO be a worthy upgrade? Maybe.


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod