ngVLA Electronics Memo No.

3
Advanced Stirling Pulse Tube Cryocooler and
Variable Speed Gifford McMahon Cryocooler
Trade Study

Denis Urbain

28 August 2020

Abstract
The purpose of this memo is to evaluate two potential cryogenic systems for the ngVLA: the advanced
cryocooler and the variable speed GM cryocooler. After a description of both technologies to highlight
their key features, we explore the challenges presented by both systems, their technical readiness, and
their impact on the rest of the telescope design. Finally, a table compares both systems and a
recommendation is made for the ngVLA cryogenic system.

1 Introduction
The Next Generation Very Large Array (ngVLA) will have 263 antennas with baselines in excess of 1000
km and a broader frequency coverage than the current VLA telescope. Having operated the VLA for 40
years, NRAO knows that the running cost of the current cryogenic equipment used to cool the eight
radio receivers per antenna would be prohibitively expensive on the scale of the ngVLA and that new
techniques and technologies must be investigated.
The first step was to reduce the number of cryocoolers per antenna by broadening the frequency
bandwidth of the radio receivers and integrating five of the six remaining frequency bands inside a single
Dewar, thus reducing the number of cryocoolers per antenna to just two. The second step was to
research new or upgraded technology that will improve the reliability of the cryogenic equipment and
reduce the power consumption level such that the operation cost is kept within a factor of three times
the current budget. The NRAO engineering team, with help from outside experts, has selected two types
of cryocoolers that could meet the ngVLA requirements. The first comes from the space industry and is
a flex bearing Stirling pulse tube cryocooler (called “advanced cryocooler” in the following text) [RD02].
The second one is a variable-speed Gifford McMahon (GM) cryocooler [RD03].

Page 1 of 22
2 Advanced Cryocooler
This type of cryocooler was developed for the space industry where high reliability, light weight, and low
power consumption are essential and cost is not the primary concern. These cryocoolers are very
compact, and the compressor and cold head are tightly integrated, as seen in Figure 1.

Figure 1. Cryocooler with integrated compressor and cold head for space application (Northrop Grumman).

The reliability of the compressor stage comes from the flexure bearing technology that eliminates rubbing
contact in linear motors. Figure 2 shows an example of flexure bearing design.

Figure 2. Example of flexure bearing design.

Page 2 of 22
This type of bearing uses a thin metal sheet cut with a very specific spiral pattern that is flexible in the
axial direction but radially stiff. Figure 3 shows a flexure bearing configuration for a linear motor application
[RD04].

Figure 3. Flexure bearing arrangement for linear motors.

The original development was carried out by Oxford University and Rutherford Appleton Laboratory in
the late 1980s. Because the movement of the spring is only in the axial direction, a shaft connected to a
couple of these flexure bearing assemblies has perfect translation movement (see Figure 4).

Figure 4. Cross section of a flexure bearing linear motor piston drive.

Page 3 of 22
If the shaft is connected to a piston and driven by a linear motor, the piston can move inside a cylinder
without touching the walls. The lack of contact between surfaces eliminates the mechanical wear and the
need for lubrication. Very tight mechanical tolerances between the piston and the cylinder wall limit blow
by and allow gas compression (clearance seal). Because no lubrication is required, no oil/gas separation
circuitry is needed. When designed correctly, the flexure of the bearing does not introduce metal fatigue
and the bearing operates in the infinite life regime. Figure 5 shows the cross-section of a dual piston
compressor configuration used in the advanced cryocoolers. Having the movement of two pistons
synchronized and in opposite directions allows the system to be well balanced, limiting the vibrations
[RD08]. An active counterweight system could be implemented to further suppress the vibrations for
sensitive applications.

Figure 5. Flexure bearing dual piston compressor setup.

In the advanced cryocoolers, the pistons opposing each other move at approximately 70 Hz. By
comparison, a GM cryocooler runs at approximately 1.2 Hz and requires valves to separate the
compression and expansion phases of the thermal cycle.
The refrigerator portion of the cryocooler is a multistage pulse tube that, by design, has no moving parts
and relies on a pressure oscillation generated by compression of Helium gas that goes through a
regenerator to cool. For simplicity, Figure 6 compares the single-stage Stirling pulse tube cryocooler with
the GM cryocooler. Unlike the latter, the Stirling-based cryocooler does not need valves to separate the
compression and expansion phase of the thermal cycle.

Figure 6. Single-stage Stirling pulse tube cryocooler (left) and GM cryocooler (right).

Page 4 of 22
Figure 7 shows that the pressure waves generated by the piston displacement create inside the pulse tube
some hot and cold spots. Figure 8 details the Stirling cycle, showing how heat is removed from the “cold
block” and rejected to “room temperature” in the warm part of the tube. The cold part is used to cool
the sensitive electronics and the warm part(s) require external cooling to dissipate the generated heat. In
the case of ngVLA, a Glycol cooling circuit will remove the dissipated heat from the warm section(s) and
cool the compressor. The portion of the electrical power required by the Glycol cooler to cool the
cryocoolers needs to be added to the overall power consumption of the cryogenics subsystem for a fair
comparison with the GM cryocoolers option.

Figure 7. Single-stage Stirling pulse tube schematic.

Figure 8. Pulse tube Stirling cooler refrigeration cycle.

Summary
The advanced cryocooler technology features high reliability and no maintenance. The flexure bearings
control the movement of the pistons with precision mechanical tolerance seals that eliminate mechanical
wear due to friction. The absence of oil lubrication greatly simplifies the compressor design and prevents
any risk of contamination of the cold stages. The cryocooler has no moving parts that can fail, and the
mode of operation that follows the Stirling thermal cycle is efficient and does not require a valve to isolate
the compression phase from the expansion phase. The overall design is very compact, is lightweight, and
has low power consumption, though it also presents a limited cooling capacity.
In our application, the thermal loads are higher than encountered in space missions and could exceed the
cooling capacity of a single cryocooler. Having to use a second cryocooler would double the cost and
make the mechanical design more complex and expensive. On the other hand, developing a new higher-

Page 5 of 22
capacity advanced cryocooler would compromise the technical readiness of this technology, and would
require an important financial commitment for the project early on, to pay for NRE.
The cooling capacity variations with cryocooler orientation impose additional restrictions on mechanical
design that could increase the overall cost. The external cooler required to dissipate generated heat
degrades the power budget and compromises the overall reliability. Finally, the mechanical design and
materials must be revised to lower the construction cost and simplify the manufacturability.

3 GM Cryocoolers
Standard GM Cryocooler
The standard GM cryocooler is currently in operation at the VLA and has been a part of radio astronomy
history since the late 1970s. Unlike the advanced cryocoolers, this type of cryocooler requires
maintenance because the compressor is lubricated and has mechanical wear from contact and friction.
The cold head has a displacer that moves inside a cylinder with surface contact and no lubrication because
of the very low temperature during operation. The displacer seals, bearings, and O-rings present in the
cold head need to be replaced periodically.
The Helium gas is mixed with oil for lubrication and cooling of the compressor, but both must be separated
before the compressed gas reaches the supply line and cold head. The separation is done in three steps
(see Figure 9): the bulk oil separator, the final/mist separator, and the charcoal trap adsorber. The oil
captured by the first two stages is recycled back to the compressor capsule; the oil trapped by the
adsorber is not. The adsorber is a disposable part of the compressor that must be exchanged every 30,000
hours to avoid saturation and possible contamination of Helium lines and cold heads.
The compressors are heavy pieces of equipment that consume several kW and dissipate most of this
power as heat that needs to be removed. For ngVLA, the plan is to have the compressor outside on a
platform above the azimuth bearing and cable wrap of the antenna. The heat generated by the compressor
is transferred to the ambient air via a heat exchanger and a cooling fan.

Figure 9. Compressor Helium and oil separation circuit; image taken from [RD09].

Page 6 of 22
In a GM system, the cold head has between one and three displacer stages and could reach 4 K for the
superconducting market (MRI). The cold head has contact seals on the displacer assembly (to avoid Helium
blow-by) that wear out and need to be replaced to maintain the cooling capacity. Figure 10 shows the
cross-section of a two-stage GM cold head with the displacer seals circled in red [RD10].

Figure 10. Cross section of two-stage GM cryocooler.

A cold head valve actuated by the motor synchronizes movement of the displacer with the connection to
the compressor’s high and low pressure sides. Figure 11 shows the GM thermal cycle and synchronization
of the displacer movement with the valve, and the connection to the high and low pressure areas.

Figure 11. GM cycle.

Page 7 of 22
The GM cryocoolers have many moving parts and need to be refurbished regularly to restore their cooling
capacity. The maintenance periodicity directly relates to the displacer speed. At the VLA, for example, the
smaller cold heads that run at 200 rpm will need service three times more often than the larger model,
which runs at 72 rpm [RD11].

Summary
The GM cryocoolers have demonstrated a long service life, and they can be serviced and rebuilt almost
infinitely. They are commercially available from several manufacturers at a reasonable cost and with a wide
range of temperature and cooling capacity. In general, the GM cryocoolers are power hungry because
their thermal cycle is less efficient than the Sterling cycle, and the compressor and the cold head are not
always well matched in size. The compressor is very often oversized, and a portion of the high-pressure
Helium generated is bypassed internally and does not contribute to the cooling process. The required
lubrication makes them susceptible to oil contamination and imposes a regular maintenance schedule.

Limitations of the Standard GM Cryocooler

The graph in Figure 12 shows the Helium circuit supply and return pressures versus the flow, for a
Sumitomo FA-40 compressor that runs at 40 Hz. To create this graph, the cold head has been replaced
by an adjustable needle valve. When the valve is closed, there is no Helium flowing through the circuit; all
the pressurized Helium generated by the compressor capsule is going through the Internal Relief Valve
(IRV); see Figure 9 for location of the valve. The IRV allows passage of Helium from the high pressure side
to the low pressure side, preventing compressor over-pressure. As the needle valve opens slowly, Helium
starts flowing through the circuit and the differential pressure begins to drop.

Figure 12. Flow and pressures for a Sumitomo FA-40 compressor at 40 Hz.

Page 8 of 22
The flow of pressurized Helium produced by the compressor is divided between the needle valve and the
IRV. When the flow reaches 28–29 scfm, there is an inflection in the pressure curves and the pressures
start to change more rapidly with the flow. This knee in the curve occurs when the IRV closes completely
and all flow produced by the compressor is going entirely through the needle valve. To be in that regime
at 40 Hz, the differential pressure (supply pressure minus return pressure) has to be below 210 psi.
For example, using a FA-40 Sumitomo compressor that runs at 40 Hz connected to one Trillium CS350
cold head that runs at nominal power frequency 60 Hz, the supply pressure is 313.4 psi and the return
pressure is 92.1 psi, giving a differential pressure of 221.3 psi (Table 1).
Compressor Supply Return Differential Helium Flow Power Power
Frequency Pressure Pressure Pressure Cold head consumption Factor
(Hz) (psi) (psi) (psi) (scfm) (kW)
35 309.5 99.1 210.4 25.6 3.395 0.854
40 313.4 92.1 221.3 27.1 3.951 0.861
45 318.1 90.9 227.2 28.2 4.354 0.878
50 324.8 88.5 236.3 28.7 4.782 0.891
55 324.05 84.88 239.17 28.7 5.278 0.908
60 327.33 86.04 241.29 28.9 5.886 0.915
Table 1. Cold head flow and power consumption for various compressor operating frequencies.

Table 1 lists the supply and return pressure and the flow for a single Trillium 350 cold head that runs at
60 Hz and a compressor that operates at frequencies between 35 Hz and 60 Hz. The power consumed
by the compressor and the power factor are recorded at the same time.
When the operating frequency increases, the operating point moves to the left away from the knee in the
curve and more flow is diverted through the IRV, and at the same time, the power consumption goes up.
In our example, the Trillium 350 cold head has very similar flow for 35Hz and 60Hz compressor operating
frequency; the cooling capacities are very similar but the power consumption almost doubles.
Figure 13 shows the load maps for the Trillium 350 cold head at eight different compressor running
frequencies.

60Hz , 55Hz , 50Hz , 48Hz , 45Hz , 42Hz , 40Hz , 35Hz

Figure 13. Trillium 350 cold head load maps for compressor frequency 35–55 Hz.

Page 9 of 22
To generate the maps, thermal loads are applied to the first and second stages using heat resistors and a
dual-channel programmable power supply. The load on the first stage varies from 0 to 15 Watts in 5-Watt
steps while the load on the second stage varies from 0 to 4 Watts in 2-Watt steps. These load maps are
a common 2D representation of a refrigerator cooling capacity.
This measurement shows a small variation in cooling capacity of the cold head with the compressor
operation frequency (pink is 35 Hz and orange is 60 Hz), and confirms that as the frequency goes up, most
of the added flow is bypassed by the IRV and does not contribute to the cooling.
To run a single Trillium 350 cold head, the Sumitomo FA-40 is oversized, but having the ability to run the
compressor at lower speed allows the flow to be adjusted to maintain cooling capacity while reducing
power consumption significantly.

Variable-Speed GM Cryocooler
Variable-speed operation of the GM cryocooler addresses some of the limitations of the standard system.
In a standard GM cryocooler, the compressor is always oversized to ensure the required Helium supply.
The excess flow is bypassed internally by the IRV and not used for cooling. The cold head is also selected
with excess cooling capacity to meet the desired cooling time rather than the steady state thermal load
[RD03]. If a precise temperature is necessary, additional heat is applied and adjusted through a control
loop. Having the ability to change the speed of the compressor and cold head allows users to adjust the
cooling capacity to meet their needs, limiting the bypassed Helium flow and cryocooler cooling power.
When the cryocooler runs slower, the wear and tear is reduced and the interval between maintenance is
extended, improving reliability. This was demonstrated using maintenance data from the VLA and is
documented in [RD11].
Adjusting the speed of the compressor changes the amount of high-pressure Helium produced and limits
the amount of Helium bypassed internally that doesn’t contribute to the cooling. The compressor runs
cooler and the heat rejection is lowered, so the oil separation is more efficient, reducing the amount of
oil that could reach the adsorber. The variable speed of the GM cryocooler does not change the efficiency
of the thermal cycle but optimizes the cold head’s cooling capacity to meet the thermal load and adjusts
the production of pressurized Helium to eliminate the bypassed flow through the IRV. The reliability is
improved and the interval between maintenance extended.

4 Advanced Cryocooler Impact on Dewar Design

The advanced cryocooler uses a pulse tube type cryocooler with the advantage of having no moving parts
but also the inconvenience of a cooling capacity that varies with orientation. Pulse tube cryocoolers have
their optimum cooling performance in the vertical direction and see a degradation in performance when
tilted to the side. When the telescope moves in elevation, the cryocooler orientation changes and its
cooling capacity varies. To limit this effect, the cryocooler would be placed at a 45° angle from the vertical
position. In this orientation, as the antenna moves through the full range of elevation, the cryocooler angle
would remain within ±45°, limiting the cooling capacity variation within a reasonable range.
The initial thermal analysis of the ngVLA Dewar-A and Dewar-B [RD12] has shown that the thermal load
exceeded the capacity of the two-stage advanced cryocooler currently available. Developing a new
cryocooler with larger cooling capacity would require NRAO to invest a large amount of money up front
for NRE, and Northrop Grumman to commit an engineering team to the project. This effort would be
expensive and time consuming, and would present a risk for the project. In addition, the long-term
reliability of the new cooler would not be demonstrated by many years of continuous operation.
One solution proposed by NG was to add an intermediate stage at 110 K that would be cooled by a
separate single-stage advanced cryocooler. Having this intermediate stage drastically reduces the radiative

Page 10 of 22
load on the 50 K stage but has only a marginal effect on the 15 K stage. This solution has a strong impact
on the design of Dewar A and Dewar B, with a second radiation shield at 110 K required and additional
space required for the second cryocooler.

Summary of Impact
• 45° orientation recommended for both pulse tube cryocoolers:
o More complex design because of the 45° orientation.
• Additional intermediate stage at 110 K:
o Possible larger Dewar volume required to fit the additional radiation shield.
• Second advanced cryocooler:
o Two mechanical interfaces with the Dewar.
o Possible larger volume required to accommodate two advanced cryocoolers.
o Larger Dewars might impact the design of the protective enclosure and could be a
problem for the smaller 6m antenna.
o Might restrict access to other components and make maintenance more difficult.
• Tight mechanical integration of advanced cryocoolers and Dewar:
o Unlike the GM system, where the cryocooler and Dewar interface is simple and allows
the cold head to be replaced on the antenna, the advanced cryocooler is tightly integrated
with the Dewar and does not allow any antenna maintenance.
• Integration of compressor and cryocooler:
o Power dissipation at the cryocooler Dewar interface increases the amount of heat
generated inside the protective enclosure.
o Connection to the chiller is required to remove dissipated heat from the Front End
enclosure.

Page 11 of 22
5 Dewar-A and Dewar-B Thermal Study Results
At this stage of the project, only paper designs are required. The Front End group decided on two Dewar
concepts. Dewar-A has the lower 1.2–3.5 GHz frequency band, while the five higher bands covering 3.5–
116 GHz with a gap for the oxygen line are integrated in Dewar-B. A thermal analysis was conducted by
an external expert (Callisto, France) on both Dewars to estimate the thermal loads and help with the
selection of the cryocoolers [RD13, RD14]. These studies compared a two-stage GM cryocooler against
a dual advanced cryocooler with the additional 110 K radiation shield. The results of the thermal studies
are summarized in the tables below. The 50K temperature was arbitrary chosen for the thermal study but
in practice a temperature around 80K is acceptable.
Dewar-A @ +20°C Ambient Temperature GM Cryocooler Advanced Cryocooler
Heat Lift in Watts Heat Lift in Watts
110 K stage n/a 10.01
50 K stage 9.88 0.43
20 K stage 3.08 2.37

Dewar-B @ +40°C Ambient Temperature GM Cryocooler Advanced Cryocooler

Heat Lift in Watts Heat Lift in Watts
110 K stage n/a 19.3
50 K stage 11.8 4.1
20 K stage 3.5 3.5

The Table below lists the cooling capacities of the GM cryocooler and advanced cryocoolers considered
for this trade study. The information for NGAS coolers comes from [RD02] and [RD16], while the
information for the Trillium comes from a load map done at NRAO at 40Hz compressor frequency.

Trillium 350 NGAS HEC Single Stage NGAS LT-RSP2

Cooling Capacity Cooling Capacity Cooling Capacity
in Watts in Watts in Watts
110K stage n/a 20 n/a
60K stage 10 n/a 5
20K stage 5 n/a 2.6

Notes
The load on the 110 K stage for the Dewar-B could be reduced by adding more layers of MLI, but is still
within the capability of the single-stage advanced cryocooler.
The addition of the 110 K radiation shield reduces the load on the 50 K stage significantly because most
of the thermal load comes from radiative loading of the Dewar walls. However, the impact on the 15 K
stage loading is negligible because most of the heat load comes from radiation through the vacuum
windows that are not impacted by the additional radiation shield.
Some multilayer Infrared (IR) filters cooled by the 50 K stage load are placed between each window and
feed horn to intercept some of the load, but their impact on the incoming radio signal has to be as low as
possible to maintain good receiver sensitivity. This requires a delicate balance between IR attenuation, RF
transmission, and the physical temperature of the vacuum window. If the vacuum window gets too cold,
condensation could form on the outside surface under certain environmental conditions and the signal
would be attenuated, having a strong impact on the receiver sensitivity. This problem is well described in
Callisto’s report [RD14].

Page 12 of 22
6 Comparison Tables
Option Power Consumption

Variable-Speed GM Both Dewar-A and Dewar-B use trillium 350 cold heads that draw about
Cryocooler 200 W of electrical power. The compressor uses about 4 kW for a total
of 4.4 kW at 40 Hz.
Advanced Cryocooler Both Dewar-A and Dewar-B need two advanced cryocoolers to get cold:
a single stage to cool the 110 K radiation shield and a dual stage to cool
the 50 K and 15 K stages. The estimated power consumption is 160 W
for the single stage and 700 W for the dual stage. In addition to the
cryocoolers, some power must be added for the portion of the chiller
used to remove the dissipated heat from the cryocoolers. An estimated
220 W per advanced cryocooler combination/Dewar is necessary. So the
total power consumption comes to 2.16 kW, approximately half the
power consumed by GM system.
Preferred Option The advanced cryocooler will require half the power of the GM system
or less depending on the compressor operating frequency.

Option Maintenance

Variable-Speed GM The GM cryocooler requires regular maintenance to prevent drop in

Cryocooler efficiency and catastrophic failure. Every 30,000 hours the compressor
adsorber needs to be replaced, and the cryocooler usually needs a rebuild
every 20,000 hours on average.
Advanced Cryocooler This type of cryocooler was designed for space and military applications,
so no maintenance is required. It is a completely sealed system with no
lubrication, therefore no oil management is necessary and there is no risk
of contamination. The only drawback is the system cannot be repaired; it
must be replaced in the event of a failure.
Preferred Option The advanced cryocooler does not require any maintenance, whereas the
GM system requires regular service.

Option Reliability

Variable-Speed GM If the maintenance is done properly and in a timely manner, the GM

Cryocooler cryocooler is reliable. Even if the system breaks, it could be completely
rebuilt.
Advanced Cryocooler The advanced cryocooler is extremely reliable, but the chiller required to
remove the dissipated heat is much less reliable.
Favorable to The advanced cryocooler is extremely reliable, but it requires an external
chiller to remove the dissipated heat at the Dewar interface. The glycol
chiller we are contemplating using for ngVLA has a reliability comparable
to the Scroll compressor used by the GM system.

Page 13 of 22
Option Cooling Capacity

Variable-Speed GM The selected GM cryocooler has plenty of capacity to cover the

Cryocooler estimated thermal heat loads of Dewar-A and Dewar-B. The variable-
speed operation allows the cooling capacity to be adjusted to match the
loads while minimizing power consumption and maintenance.
Advanced Cryocooler The advanced cryocoolers are very compact and lightweight because they
were designed for space application where volume and mass cost a
premium. They have a limited cooling capacity which explains why two
cryocoolers need to be used in parallel to cool Dewar-A or Dewar-B. At
this time, it is still uncertain that even a dual system will have enough
capacity to cool the Dewar-B.
Preferred Option Larger cooling capacity for GM cryocooler could be adjusted with the
variable speed option. With the Dewar-B it seems that we are exceeding
the cooling capacity of the available advanced cryocoolers.

Option Volume and Mass

Variable-Speed GM The mass and volume of the GM system is significantly higher mainly due
Cryocooler to the compressor weighing over 140 kg and measuring 948 x 1016 x 391
mm. The cold heads are about 15 kg each. When you add the Helium
lines that run from the compressor all the way to the Front End
enclosure, the total mass of the GM system exceeds 250 kg.
Advanced Cryocooler Each advanced cryocooler has a mass of 10 kg or less, so the overall mass
of the cryogenic equipment is reduced by approximately 80%. However,
the mass located inside the Front End enclosure on the antenna feed arm
is comparable for both systems. Being located on the separate platform
behind the main dish, the GM compressor has very limited impact on the
design of the antenna feed arm.
Preferred Option The advanced cryocooler has a smaller overall mass, but the mass located
on the antenna feed arm is comparable. Because the compressor and
cryocoolers are integrated, the volume required by the advanced
cryocoolers inside the Front End enclosure might be larger.

Option Mechanical Interface

Variable-Speed GM The mechanical interface of the GM system is very simple with a sleeve
Cryocooler insert that is permanently attached to the Dewar and allows the cold
head to be replaced on the antenna.
Advanced Cryocooler The advanced cryocooler interface is more complex because of the 45°
angle mount and the fact that two cryocoolers are needed per Dewar.
The geometry of the advanced cryocooler is also more complex and we
have to avoid mechanical interference between the two cryocoolers.
Preferred Option Simpler mechanical interface for the GM cryocooler.

Page 14 of 22
Option Sensitivity to Motion or Orientation

Variable-Speed GM The GM system can be operated in any orientation and the cooling
Cryocooler capacity variation is not significant. For example, ALMA placed the
cryocooler parallel to the antenna elevation axis to minimize the
temperature variation with elevation.
Advanced Cryocooler The pulse tube cooling capacity is sensitive to orientation with the
vertical position being optimum. The Stirling pulse tube is less sensitive
because of the higher frequency of operation and the absence of valves
but it is recommended not to exceed ±45° from the vertical position. To
meet this requirement, both advanced cryocoolers need to be mounted
with a 45° offset position.
Preferred Option The GM cryocooler has the advantage because it is less sensitive to
motion and orientation. The temperature stability of the cryocooler
translates directly into the gain stability of the low noise amplifiers it
cools and the ngVLA requirement might be too tight to be met with the
advanced cryocooler.

Option Heat Rejection and Management

Variable-Speed GM The heat removed by the cold head is transported away from the Dewar
Cryocooler by the Helium gas. The compressor is air cooled and the heat produced
by the compression of the Helium is transferred to the oil that is cooled
by the heat exchanger and the fan.
Advanced Cryocooler In the advanced cryocooler, the compressor and cold head are tightly
integrated and the heat generated needs to be moved away from the
Dewar interface by an external Glycol circuit.
Preferred Option The GM system is simpler and does not rely on an external chiller.

Option Cost Estimate

Variable-Speed GM The GM system has other commercial applications, for example, in the
Cryocooler medical field to cool MRI scanners and in the semiconductor industry to
evacuate sputter machines. Therefore, the GM system benefits from the
economy of scale. The cost is approximately $60,000 per antenna.
Advanced Cryocooler The advanced cryocoolers have only been manufactured in very small
quantities for space missions. The fabrication process will have to be
adapted for a larger production run. Because of the limited cooling
capacity, each Dewar requires two coolers, which almost doubles the
cost. The current cost estimate is approximately $400,000 per antenna.
Preferred Option The initial cost of the advanced cryocooler is so much higher than the
GM system that it would take longer than the expected lifetime of the
telescope to recover the construction cost difference [RD15].

Page 15 of 22
7 Summary Table
Variable-Speed GM Advanced Cryocooler
Cryocooler
Power Consumption 0 1
Maintenance 0 1
Reliability 1 1
Cooling Capacity 1 0
Volume and Mass 1 1
Mechanical Interface 1 0
Sensitivity to Motion and Orientation 1 0
Heat Rejection and Management 1 0
Cost 1 0
Total 7 4

8 Conclusion
The low power consumption and absence of required maintenance make the advanced cryocooler very
attractive for ngVLA. However, its limited cooling capacity requires the combination of two cryocoolers
to cool the Dewar-A, and is not sufficient to cool the Dewar-B in the current configuration. More work
needs to be done on the IR filtering for the Dewar-B to reduce the thermal load on the 15 K stage to a
level compatible with the advance cryocooler cooling capacity.
The impact on the design of the Dewar may be significant and will add to the fabrication cost of the Front
End. If we add the fact that an external chiller is required to remove the heat generated by the cooler, the
reliability of the complete system becomes more or less equivalent to that of the GM system.
Given that the advance cryocoolers are designed for space missions, the unit construction cost is very
high and is not optimized for large production. Northrop Grumman attempted to cut costs by using lower-
cost materials and reducing the required testing, but the initial cost is still too high to be competitive with
the GM system. This is even when the cost savings due to lower power consumption and maintenance
over a 30-year life cycle are taken into account.
At this stage of development in the ngVLA project, the potential benefits offered by the advanced
cryocooler are outweighed by the very high initial cost, the limited cooling capacity, and the added
complexity to the Dewar design.
A cryogenic system based around variable-speed compressors and VFD GM cryocoolers is therefore the
preferred option for the ngVLA project.

9 Reference Documents
Ref. No. Title Authors, Publication
RD01 Flexure bearing support, with particular United States patent 5,522,214 June 4, 1996
application to Stirling machines
RD02 Advanced Cryocoolers for ngVLA Larry D’Addario, Caltech: ngVLA Optics
Workshop, Pasadena, CA, June 20, 2018
RD03 Improved Power Efficiency for Cryogenics D. Urbain, W. Grammer, G. Peck, J. Jackson,
at the Very Large Array S. Durand, International Cryocooler
Conference, San Diego, CA

Page 16 of 22
RD04 Development of the LSF 95XX 2nd J.C. Mullie, P.C. Bruins, T. Benschop, M.
generation flexure bearing coolers Meijers THALES Cryogenics B.V. SPIE
Conference 2005
RD05 Development Progress of Long Life Twin L. Duband, A, Ravex and P. Rolland
Piston Pressure Oscillator Advances in Cryogenic Engineering, Vol. 39
RD06 Modeling an Analysis of Deformation on a Vennapusa V. Sivareddy
Flexure Bearing in Linear Compressor IJMERR, Vol. 3, No.1, January 2014
RD07 AIM-Space Cryocooler Program M. Mai, I. Ruhlich, A. Schreiter, S. Zehner
AIM Infrarot-Module Gmbh Heilbronn,
Germany
RD08 Refrigeration Systems for Achieving Ronald G. Ross, Jr., Jet Propulsion
Cryogenics Temperatures Laboratory, California Institute of
Technology, Pasadena, CA 91109
RD09 FA-40H and FA-40L Air-Cooled Helium Sumitomo (SHI) Cryogenics of America, Inc.
Compressor Operating Manual 1833 Vultee Street, Allentown, PA 18103
RD10 https://www.arscryo.com/cryocooler- Advanced Research Systems, 7476 industrial
principles-of-operation Park Way, Macungie, PA 18062
RD11 MTBF Report on EVLA Cryogenics James Gregg, Socorro NM, March 29, 2016
NRAO Private Communication
RD12 ngVLA Front-End Thermal Study Initial Remi Rayet, Callisto, 12 Av. De Border
Analysis Report (11/07/2018) Blanche, Villefranche de Lauragais F-31290,
France
RD13 ngVLA Front-End Receivers Thermal Remi Rayet, Callisto
Study Analysis Report with 110K
Intermediate Cooling Stage (05/12/2018)
RD14 ngVLA Front-end Receivers Thermal Study Remi Rayet, Antonella Simone, Callisto
Dewar-B Update (14/02/2020)
RD15 Cost Comparison between the Stirling NRAO Internal Document
Pulse Tube Advance Cryocoolers Concept # 020.30.10.00.00-0005-REP
and the Reference Design Gifford
McMahon Cryogenic System
RD16 High Efficiency Cryocooler Performance D. Durand, T. Nguyen. E. Tward
Northrop Grumman Aeronautics Systems

10 Acronyms
EVLA Extended VLA
GM Gifford McMahon
ngVLA Next generation VLA
MLI Multi-Layer Insulation
MRI Magnetic Resonance Imaging
NRAO National Radio Astronomy Observatory
NGAS Northrop Grumman Aeronautics Systems
NRE Non-Recurring Engineering
VFD Variable Frequency Drive
VLA Very Large Array

Page 17 of 22
11 Appendix

Page 18 of 22
Page 19 of 22
Page 20 of 22
Page 21 of 22
Page 22 of 22