In this article metrology-grade multichannel scanner from Data Proof is serviced. This is second scanner in my lab that will be used for second zener bank array system to monitor new ADR1001 zener standards I’m building. Model evaluated here is an old variant of Data Proof 160, which feature 16 channels with direct cable input.

The Data Proof low thermal scanner was originally designed to perform series opposition measurements of standard cells back in 70s. The design is based on the recommendation from NBS (now NIST) published in the NBS Technical Note 430. The document was published in 1967 for use with standard cells, but it became the accepted method of measuring zener based references in modern metrology uses today for calibration of secondary standards like the Fluke 732A or similar.

Data Proof 160A/320A manual w/schematics

Data Proof 160A, 164A, 320A 1999 Manual w/schematics

Newer Data Proof 160B/320B datasheet

Data Proof OhmREF datasheet

Data Proof VoltRef – software for voltage maintenance

Exterior

I’ve got 16-channel unit here with Option 1. A couple of options were available with the scanner. Option 1 indicate configuration with pre-attached eight conductor unshielded cables. These are basic telephone-looking cables without shielding using a solid bare copper conductor. The cable listed in the parts list is a Belden 9794. Option 2 had 16 pairs of low thermal binding posts made by Pomona 3770. The option 1 are more convenient as no additional cables are needed to connect voltage references. I have already reviewed few of these Data Proof scanners in this article. One of 160A is already running daily for many years at xDevs.com as part of in-house zener voltage bank array. Freshly arrived scanner was used in a rack and does not have decorative wooden sides. No polish or any pretty covers, just pure business and aluminum enclosure meant to be integrated into standard 19” rack.

Image 1: Data Proof 160A Standard Cell Scanner, 16 channel version

Other than few sticker residues it looks pretty clean and without any signs of corrosion anywhere.

Image 2: Side panel, full metal enclosure

Rear shows interface IEEE-488 connector, protection monitor terminal block, output on dual 5-way binding posts terminals and mains power inlet. This scanner was assigned manufacturer serial number 504. This scanner was part of govt lab with warranty expired on 3-10-1993. If scanner was bought with standard one year warranty that datestamps age of this scanner to 34 years at the moment of this article publication.

Image 3: Data Proof 160A Option 1, rear panel

Unlike my other scanner this one still has original labels for each of the 8-wire cables. Cables are solid core pure copper with PVC insulation and without any shielding. Data Proof has option 4 for special low leakage PTFE shielded cables, but for voltage scanner application this cable works already good enough. Cables are not very long, perhaps about 1.5 metre each.

Image 4: Cable labels very handy during connection

Cable ends seen better days with some insulation cracking apart. Nothing too concerning and easy to rectify by chopping some more cable.

Image 5: Data Proof 160A Option 1, hardwired rear cables seen better days

After power up this scanner consumed about 11 W from 120 V 60 Hz Chroma 61604 AC source but didn’t want to do anything. Buttons on any of the channels didn’t produce any relays clicking or LED activation on front panel. So this is a weekend repair project as result.

Internal design and construction details

Removing cover immediately revealed rotten and crumbling foam pads. I might clean that and replace later after electrical functionality restored.

Image 6: Thermal insulation foam on cover is in poor condition

Data Proof used five control boards for all the digital logic plus front panel with switches and LEDs and small power supply board with very simple half-wave rectifier, regulator and capacitors. Holder for two spare relays is already empty in this unit :)

Image 7: Digital boards and power supply module

Let’s take a look on all easily accessible digital and control boards first.

Image 8: Interface board 320-007A with chips dated 20 week 1990

DIP switch S1 here sets the scanner IEEE-488 bus address. This is older interface card variant that was replaced in later Data Proof units.

Image 9: Rear side of the interface board.

Area behind PROM chip has suspicious darkening of PCB, suggesting elevated temperatures.

Image 10: Controller board 320-001

Data Proof didn’t use some fancy microcontroller or programmable logic device in these old scanners, everything is done with discrete logic gate ICs. This means also rather specific IEEE-488 communication protocol with some specifics like need of delay or sending end of character symbol for channels programming. This scanner will not respond to fancy SCPI or *IDN? commands :-)

Image 11: Controller board soldering side

Next we have two of open boards for relay deactivation, one for each of 8 channels.

Image 12: Open board 370-005A

Again just some logic gates ICs, three resistors and two capacitors.

Image 13: Second 370-005A PCB, which is identical

Image 14: Rear of open board circuitry

And finally the close board that switches relays on:

Image 15: Close board 320-004

This one has two larger 74LS154N devices and some more 7407 gates.

Image 16: Rear of the close board

Image 17: Power supply board is mounted to opening in the inner wall next to mains 50/60 Hz power transformer

Original capacitors look rather poor and first target for replacement. This will be first item of attention in repair.

Image 18: Capacitors crying for replacement

Dataproof didn’t bother making PCB for the bottom side to interconnect PCBs, they just wired all the edge connectors wire PTFE cables.

Image 19: Wiring harness for edge connectors

All wires are secured into harnesses and very stiff after all those decades.

Image 20: Pretty cabling work

I’ve took a photo for PSU cables wiring so I’ll know where each wire goes after board service.

Image 21: Power supply wire connection locations

Repair of power supply board 320-006A

Here’s the condition of power regulator board as it is, before any repairs.

Image 22: Power supply PCBA 320-006A with original capacitors

Flux residue on solder joins is original here :)

Image 23: Power supply regulator 7805.

Newer Data Proof 160A scanners have updated power supply board with two linear regulators. Here we still have old variant with only one. Original capacitors are axial type made by Sprague (today part of Vishay conglomerate):

• 1 × 4000 µF 15 VDC
• 3 × 100 µF 16 VDC

To repair I’ll replace all four caps with modern high-temperature capacitors of larger capacitance. This is linear regulator, so I don’t think some extra capacitance will hurt anything.

Image 24: Rear of the power supply PCBA

And top of the board view with new capacitors installed. I’ve chosen Nichicon UBY type which are radial capacitors. Smaller capacitors were replaced with 1000 µF 50 V and larger one was replaced to 10000 µF 35 V can. Small wire was added to extend the negative leg of larger cap as well.

Image 25: Serviced board ready for installation back into unit

Relay boards cleanup and repair

Repairing power supply reduced ripple quite a lot and measured output voltages were now +5.00 V and +15 V.

But scanner still didn’t want to actuate any of the relays unless PROTECT DEFEAT button for either A or B branch were hold down. Time to get inside of the thermal box with thick aluminum walls and clean up all relays on all four cards.

Image 26: Thermal shield cover removed

The heart of the scanner is based on a special relay that has no stationary contacts. The relay was designed to use the traces of a pcb as its contacts, avoiding any solder metal in signal path. I believe the relays had their origin in older POTS switches but cannot confirm this is true. Each channel has two possible output paths ways, arm A or arm B. Only one of the channels can be selected in any of the arm, but same channel can be used to connect to both arms. Control of each arm is fully independent and any A relay can be activated with any B relay combination. This is very powerful feature that makes these scanners much more than simple two-pole port extension.

Diagram 1: Functional design of single channel in Data Proof

The model of the relay is a patented PRINTACT Executone 12BW3LD. This model is a dual latching style and both sets of coil contacts can be seen from the bottom with small springs wound around them. The contact configuration is special 3PDT which is not common these days.

Images 27-28: Custom relay brush contact and label

Each of four card holds eight relays, each relay is 3PDT type representing this custom Printact type. This is rather cool relay type that allows no solder metal to be involved for signal switching. This was probably done for ease of field repair and replacements in telephone industry but here it finds a useful case in switching nanovolt signals where different metals would cause parasitic thermal EMF voltages.

Relay is held mechanically by metal spring clip. Relay coil contacts are made out of gold-plated springs and actual contacts are embedded on the base of the body.

Image 29: One of relay cards 320-002. This is unguarded variant

To remove relay one needs to unclip the spring and relay falls right out of the board, revealing both contacts and PCB surface with contacts out of traces. This makes cleaning contact pads trivial and easy. A drawback of such relay type is that contact interface is not hermetic and it may corrode or get leaky over time in harsh environments or poor storage conditions. I had to remove all signal cables to remove the relay card PCBs out of the frame.

Image 30: Rear of the PCB, with date code from week 40 of 1991

Card with all relays removed is pretty easy to clean by rubbing pencil eraser which is gently abrasive and cleans gold plated PCB pads very nice and without damage.

Image 31: All relays removed with PCB traces exposed for cleaning

3rd pair of contacts in each relay is used as daisy chain relay status signal feedback to digital brains. If any of the relays along the whole instrument is misbehaving and not closing that contact set the digital controller will not send the coil drive voltages to relays. But to allow easy troubleshooting designers left two momentary action buttons on the side of relay box to defeat this fault protection mechanism.

If you press and hold the button for specific arm, like A or B and then hit two buttons on front panel to activate the channel, controller will send pulses to toggle relay regardless of protection contact state. During my testing this was very handy but I wished I had third arm to pressing three buttons at same time :) Few times had to press button with nose too during testing. You don’t need to have output and intercard wiring attached to relay cards to test each channel operation.

Image 32: All cards are disconnected from wiring.

After removing all four cards and cleaning all 32 relays I put everything back together and powered up the scanner. With bit of fiddling for few hours I got B arm to operate and set channels for all cards except the last (for channels 13-16). Bending contacts and fiddling few stubborn relays for two more hours on this card got everything into operational state as well. So there was hope for this scanner after all.

Benchmarks and overall system integration

To verify good operation I’ve shorted A LO and B LO outputs with a copper wire and connected HP 34420A nanovolt meter set to 1 mV range. Here’s one of first configurations to test:

Diagram 2: Configuration for internal zero EMF relay short pathway

And visually scanner configured to both arms shorted via internal link on relay board.

Image 33: Interchannel short for testing

As result setting both arms to same channel shorts the signal path to nanovoltmeter and lets us measure residual thermal EMF of the scanner relay itself. After running initial script that goes over each channel on both arms, A15 relay failed again and I had to actually steal spare relay from my other Data Proof 160A after all. Here’s the photo of bad relay on the left that was either shorting poles together when I bent the contacts too much, or not making contact at all. Angle of contacts has to be set rather precisely for these relays to operate properly.

Image 36: Replacement relay 12BW3LD from other DataProof spare box

Scanner then was reassembled and connected to modified xDevs.com TECkit script to measure channels combinations by HP 34420A with short attached.

Residual EMF was analyzed in next two plots for clarity. First plot shows interchannel shorts by DataProof itself. Horizontal scale is number of scans done. Each measurement is taken by nanovoltmeter at NPLC 10 with 5 seconds delay after any channel switch operation. Average of 3 samples is used as result. Each scan of all channels took about 10 minutes. Results:

Image 37: Residual EMF for internal scanner channel short results

All channels show impressive spread well inside 7 nV levels, no problems detected here. Next test would be to measure zero EMF between two different channels on opposite arms. For this A and B set to desired ports and those ports are shorted together externally at the end of the cable. This test represents actual use case for measuring zener standards attached to scanner.

Image 34: External short test between A12 and B9 channels

This is same configuration as used in series-opposition voltage standard scans. According to datasheet from Data Proof scanner should meet typical residual EMF below 20 nV with maximum limit at 50 nV for this test.

Image 35: And I can get 20 nV after some settling time. Some other channels show 30-40 nV as well

Diagram 3: Configuration for external zero EMF with cable positives shorted

Results based on same timeframe over multiple hours.

Image 19: External EMF for some groups of channels

Between channels combination with series-opposition ports is worse as expected but nothing bad. Combinations against channel 5, 9, 8 and 16 were tested in this run. Worst channel combination here was CH14 to CH16, barely touching 30 nV of offset. Next worst is CH10 to CH9 riding around typical limit of 20 nV. Rest are well below 20 nV which would translate to uncertainty of measurement between two 10 V zener standards way down in the noise. Error of 20 nV translate to 0.002 µV/V for 10 V transfer which is 20 times less than noise generated by standard like Fluke 732C or similar. These errors are only significant contributors when scanner is used with quantum voltage standard like PJVS.

I’ll test some more channels and then install this scanner into xDevs UBank experimental array where we can monitor and measure various DC voltage standards and DIY builds relative to known good stable standard. Below Diagram 4 shows how DataProof scanner with two arms is used to compare two voltage standards with help of nanovoltmeter.

Diagram 4: Configuration diagram for actual comparison of two voltage standards

Use of nanovoltmeter increase sensitivity of the comparator to levels much below the noise floor of even fancy DMM like HP 3458A. If nanovoltmeter is not available, one could use common 6½-digit DMM with lowest 100 mV range and obtain still great results. Because with Data Proof this is differential measurement effects from nanovoltmeter temperature changes, non-linearity or gain error are cancelled out and uncertainty of comparison is well below 0.5 µV/V, essentially only limited by calibration uncertainty and performance of known standard unit.

For comparison even fanciest 25000 dollars 8½-digit DMM can only demonstrate uncertainty about 0.6 µV/V 24 hours after fresh calibration on quantum standard or only 2.85 µV/V for typical 1 year specification in tightly controlled ±1 °C laboratory environment.

Conclusion and summary

I’m happy to report successful repair of this Data Proof and expansion of zener experiments at xDevs.com. This project wouldn’t be the same without support of my friend Chuck who is developing the new generation voltage standard that delivers performance a magnitude better than 40-year old designs with LTFLU-1 and LTZ1000 chips we are used to see in calibration labs today.

Feel free to contact us for questions and comments. Discussion about this article and related stuff is also welcome at our own IRC-chat server: xdevs.com (port 4808, channel: #xDevs.com) if you like more real-time interaction with electronics engineering enthusiasts like the author of this article. There is also public xdevs forum where you can discuss many metrology questions with our small community.

Projects like this are born from passion and a desire to share how things work. Education is the foundation of a healthy society - especially important in today's volatile world. xDevs began as a personal project notepad in Kherson, Ukraine back in 2008 and has grown with support of passionate readers just like you. There are no (and never will be) any ads, sponsors or shareholders behind xDevs.com, just a commitment to inspire and help learning. If you are in a position to help others like us, please consider supporting xDevs.com’s home-country Ukraine in its defense of freedom to speak, freedom to live in peace and freedom to choose their way. You can use official site to support Ukraine – United24 or Help99. Every cent counts.