Little is known in the West about the production, care and feeding of commercially-available passive hydrogen masers -- only what the predominant Russian manufacturers have deigned to disclose to their customers, which isn't much. When it comes to high-performance time and frequency standards, though, passive H-masers provide stability second only to their active cousins which are even harder to come by. Consequently they are all worth preserving in good operating condition whenever feasible.

This unusual eBay find is an example of the VCH-1006 maser from Vremya-CH in Nizhny Novgorod, probably an early-production model around 20-25 years old. It was still in working condition when it landed on my bench but was starting to show its age. This page will describe measurements, repair notes, and general observations on this particular unit.

VCH-1006 user guide
VCH-1006 maintenance and service manual (such as it is)

VCH-1006 sell sheet published by Oscilloquarz SA, dated 2006 and labeled OSA-VCH 1006, specifying 2E-12 @ t=1s
OSA-3700 sell sheet OEM version marketed by Oscilloquartz SA circa 2009 with 7E-13 @ t=1s advertised stability
Russian-language sheet from Vremya (2014?) with 2E-12 @ t=1s specification matching the OSA-VCH 1006 brochure above
Most-recent specifications downloaded from vremya-ch.com (March 2024, 7E-13 @ t=1s)

Assorted historical references and research papers

.CSV data logger Runs under Windows, includes C source code. (Requires DB-9 straight-through cable between PC and maser)
.CSV data plotter Python source based on pandas and matplotlib

Schematic draft

Hydride temperature adjustment trimmer

The unit's hydrogen pressure was reported at the minimum 1.5-atmosphere level when the unit arrived. Adjusting the 'Hydride' trimmer to increase the temperature of the lanthanum nickel hydride canister brought the pressure up to the 2- to 2.5-atmosphere range. This may have improved the second-harmonic level a bit, but parameters were changing so frequently at the time due to various factors, including recovery from long-term storage and the dissociator issue described below, that it was hard to be sure. (There is also a significant change in operating parameters that occurs when the maser's top cover is removed, further complicating the tweaking process. This effect is accompanied by small frequency excursions.)

Subsequent temperature increases didn't appear to raise the H2 pressure much, suggesting that the hydrogen will soon need topping up. The pressure sensor could also be suffering from embrittlement or other age-related effects that degrade its accuracy. It's possible that the intended setting for the 'Hydride' trimpot is at a point that maintains the H2 pressure level near the 1.5-atmosphere minimum.

There appears to be little or no upside to running the maser at higher H2 pressure, assuming the purifier loop acts to maintain the same low-pressure setpoint regardless. (Which is not a sure thing; see July 2024 update below.) In any case lowering the ~2 atm pressure back to 1.6 atm had only small-scale effects on most parameters (.CSV file here.)

Toggle switch on oven control PCB

The mysterious toggle switch originally arrived in the downward 'М' position (i.e., towards the front panel as seen in the photos). Flipping it to 'Б' caused a transient that affected nearly all parameters, similar to what happens when the top cover is removed but significantly more pronounced. Lock was lost for about a minute while the operating parameters returned to their previous values. The transient had to be removed from the data set for this screenshot, as it corrupted the RS-232C output from the controller.

Ultimately, the switch's only noticeable effect was that the maser's output frequency was about 1.58E-8 higher (0.158 Hz at 10 MHz) with the switch in its upward position. Despite this substantial frequency error, stability was unaffected at taus up to at least 100s. Flipping the switch back down restored the previous operating frequency with little or no apparent retrace error once the transient settled. The switch may control some sort of degaussing, axial field homogeneity, or post-selection drift field (neck coil) function, and/or may cause the maser to lock to a spectral line at a different Zeeman sublevel, although the literature indicates that the Zeeman resonances at 1420 MHz are much more closely-spaced.

The repaired dissociator yields a beautiful view of the entire Balmer series through 100-micron slits:

(Not to be confused with the less Nobel-worthy Ballmer series)

The dissociator bulb shows some visible cloudiness in its PTFE coating, but this evidently isn't harmful:

• The 2nd harmonic tracks the voltage reported by the optical sensor in the HFO assembly very closely. The brighter the source discharge, the stronger the recovered signal. This may not be true shortly after lock is established following a power cycle, though.

• The current drawn by the dissociator HFO is now reasonably stable, but it took a bit of a dive near the 160-hour mark when I poked a fiber probe through the metal slots in the housing at rear to capture the hydrogen spectrum. There was no effect on the 2H level or discharge intensity.

• H2 pressure from the hydride canister showed signs of falling back after the 'Hydride' trimmer was adjusted earlier, but the longer-term trend is less certain. It seems to be recovering to some extent.

• Vacuum quality seems good, with consistent ion pump current readings below 2 uA.

The result is trustworthy enough below t=1500s, though. It can be seen that the VCH-1006 is outperforming the 2E-12 @ t=1s spec, but it falls short of the more aggressive 7E-13 @ t=1s spec from some of the other sales literature. On one hand, if a way can be found to increase the 2nd harmonic level, I'd expect the short-term performance to improve, perhaps substantially.

(See June 22, 2024 entry below for more on this. After replacing some faulty electrolytics, it turns out that increasing the purifier current can allow the maser to approach the original Vremya factory specs. This may not be prudent, though, given the uncertain age of the H2 supply, ion pumps and getters.)

In fact, the VCH-1006 isn't good enough to assess the short-term performance of the Rb1 and Rb2 units by itself. A three-cornered hat measurement is informative: while still susceptible to environmental influence as mentioned above, it shows that Rb1 and Rb2 can approach the maser's best factory specs at taus below 1000s. Only the maser, however, can tell us how the HP 5065As are behaving beyond that.

The slight ripple near t=1s is something I've seen before in measurements made with Tom Van Baak's Kvarz CH1-76 PHM. Not clear what's going on there, or if it's the same phenomenon at all.

• Large-scale phase excursions on 5/10/100 MHz outputs
  • These were not OCXO jumps or OFine adjustments, but appear to be 33-us phase steps executed by the DDS prior to the final internal cleanup stage
  • Steps occurred (at random?) every hour or so and take about one minute to recover with slight overshoot
  
  
• During excursions the serial output tended to exhibit large-scale parameter errors
  • Voltage parameters such as p27_V1, p15_V, n15_V reported far out of range, as well as Ion_uA and Batt_V (even though there's no battery installed)
  • Effect is reminiscent of intermittent parameter corruption seen prior to HFO repair and during other operational hiccups
    
  
  
• Sinusoidal phase artifacts occurring at various times between excursions
  • Typically 150-200 ps pk-pk as observed in 100 MHz output
  • Occurs in bursts of typically 10-12 cycles about 600s apart
  • Correlated with ripple observed in cavity oven voltages as well as other parameters.

While both capacitors showed ESR readings near 100 ohms, the input capacitor was primarily responsible for the failures. This was verified by temporarily removing both replacement capacitors to restore the fault condition, then installing a replacement on the input side while the output capacitor remained disconnected:

Additionally, a bad 470 uF 16V electrolytic was found on the current regulator PCB for the H2 purifier element. Replacing this capacitor increased the purifier current by a factor of roughly 2x and improved the 2nd harmonic level by a factor of about 3x, lowering the ADEV at t=1s from 1.4E-12 to 9.5E-13. Pending some benchmarking, it may make sense to reduce the purifier current somewhat to conserve the H2 supply. This was done about 3 weeks later, when hydrogen pressure errors began to occur (below).

Regardless of the root cause, at 9.4 +/- 0.002 volts there seemed to be a lot of headroom before reaching the 21V H2_V limit, so I cranked up the 'Hydride' trimpot on the oven control board to establish a higher setpoint near 1.7 atmospheres. This adjustment had multiple effects, some unexpected:

Because the purifier setpoint is based on current (temperature) rather than H2 volume, increasing the H2 oven pressure should(?) result in additional gas flowing to the dissociator. Paradoxically, not only did this fail to improve the maser's short-term stability, but ADEV actually increased from approximately 9E-13 @ t=1s to near 1.5E-12. Although the ion pump current didn't change much, noise on the IF and second-harmonic signals increased substantially.

To counteract this undesirable behavior and forestall the point at which I'll have to figure out how to recharge the LaNiHx storage canister (and possibly deal with the ion pumps), I decided to lower the purifier current. As the traces above show, this took care of the excess noise. At about 1.2E-12 @ t=1s, the maser's short-term stability is now roughly what it was when I bought it several months ago, which is fine. 2H_V is now reported at near 1.2V on the front panel/1.3V on the graph.

July 24: Traces from an entire two-week period show that the parameter noise/spikes eventually resumed after lowering the purifier setpoint (data). No further error messages have occurred, however, and performance looks solid:

Notes and open questions for H2 controls:

• Pin 1 of the associated connector drives the canister heating element, with pin 2 grounded
• It looks like pin 4 is a high-Z input from the pressure sensor
• Pin 3 supplies a fixed 16V bias from the floating Zener, not a typical power rail
• There are four more wires on pins 5-8, but they are unconnected to anything
• If the loop with the 'Hydride' trimpot drives the heating element voltage based on input from the pressure sensor, when/why does it need to be adjusted? Shouldn't it maintain a constant pressure until the H2 supply runs out entirely?
• Why did additional H2 flow cause SNR to get worse? Why did HFO_V return to baseline, HFO_A go down, and discharge light intensity remain largely unchanged?
• What causes the correlated noise and jumps on the H2_V, IF_V, and HFO_A parameters?
• Why does H2_V fall back so quickly after the oven setpoint is raised? For that matter, why did it return to the original baseline after the purifier setpoint was lowered? (Presumably because it now takes less canister heat to maintain the same H2 pressure upstream of the purifier.)
• How does the CPU read the pressure sensor? Need to trace the wiring more carefully

Open questions in general:

• Pin 1 of the 4-pin connector on the FLLP board is likely to be the varactor tuning voltage (~11.06). What is pin 2 for?
• What does the toggle switch on the heater control board do?
• What's the window comparator on the heater control board for?
• What's on the PCB above the backplane connectors? Appears to be some type of proprietary remote control interface that's unused in this application
• Need to draw schematic for the purifier interface board. It has its own setpoint control, so what are the other wires for?