A vacuum chamber and its connected pump can show different pressures while gas is flowing. That difference is useful evidence. If the gas throughput is known independently and the conditions suit the model, the two readings can help estimate the effective conductance of the connection between them.
The difficulty is deciding what the estimate actually describes. It might represent a pipe, a pipe together with valves, or a more complicated path. A small pressure difference can also be poorly resolved even when both gauges display many digits. This guide develops those issues through original fictional calculations, including cases in which the same pair of readings supports several explanations.
Read The Gold Standard of Vacuum Science for the wider foundations. This specialist guide focuses on inference from pressure differences, rather than providing equipment selection or operating instructions. Every numerical setup is a teaching model, not a specification for a real vacuum installation.
eduKateSG · Vacuum Science
Find your next learning step
Use paired vacuum pressures, throughput, uncertainty bounds and original worked cases to infer connection conductance without overclaiming.
Chapter contents
Understand the question · Chapters 1–4
Work the first cases · Chapters 5–8
Test the limits · Chapters 9–13
Transfer and check · Chapters 14–18
Report and continue · Chapters 19–21
CHAPTER 1 OF 21 · Understand the question
1. Draw the two measurement locations first
Call the pressure measured at the chamber end pA and the pressure measured at the pump end pB. Define the connection being investigated as the region between those two locations. If the gauge ports sit beyond additional valves or fittings, those components may belong inside the measured region even if a sketch labels only a pipe.
This boundary matters because the pressure difference belongs to the entire intervening path. A calculation cannot isolate a hidden component simply because the analyst gives the result that component’s name. Start with the actual measurement locations and trace the gas route between them.
For a fictional example, suppose a chamber connects to a pump through a pipe and one valve. A gauge at the chamber and another at the pump inlet observe the combined path. Calling the inferred value “pipe conductance” would omit the valve’s possible contribution.
Record the gas, temperature assumptions, operating state and units beside the drawing. The purpose is to make the interpretation reproducible. A later comparison with a different gas, valve state or pressure range may describe a different effective connection, even when the hardware has not visibly changed.
Before any arithmetic, ask a student to point to the upstream location, downstream location and included components. If those cannot be identified, a precise conductance value would be premature.
CHAPTER 2 OF 21 · Understand the question
2. Keep throughput distinct from pumping speed
Use Q for pressure–volume throughput, expressed here in mbar L/s. Use C for conductance, expressed in L/s. At the stated operating condition, the relation is Q = C(pA − pB), so C = Q/(pA − pB).
Leybold’s conductance reference explains the physical relation and the importance of the flow regime. Pumping speed is also commonly expressed in L/s, but it describes removal at a specified location. Sharing units does not make conductance and pumping speed interchangeable.
The pressure difference measures the connection’s role. The local pressure at the pump enters Q = S pB when the simplified pump model is applicable and limiting-pressure effects are negligible. The chamber pressure enters Q = Seff pA for the corresponding effective speed seen by the chamber.
A useful dimensional check is that mbar L/s divided by mbar gives L/s. Dividing throughput by a pressure ratio would leave the wrong units. Dividing by pA alone would give effective pumping speed, not the conductance of the connecting path.
Keep those three calculations on separate lines. Their relationship becomes clearer when the same data are used consistently, rather than when one number is labelled “speed” everywhere.
CHAPTER 3 OF 21 · Understand the question
3. State the conditions behind the simple inference
The basic calculations assume a steady operating state, a defined gas and temperature context, valid comparable pressure measurements, and a known throughput passing through the connection. They also assume that the selected form of conductance is meaningful for the flow regime and operating range.
A particularly useful classroom case treats C as constant over a limited range, as can be appropriate in a molecular-flow model for fixed geometry, gas and temperature. It does not follow that the same numerical C applies throughout a complete pump-down from atmospheric pressure.
If there is a significant gas source between the two gauges, the throughput at one end may differ from the throughput at the other. If pressure is changing rapidly, gas accumulation and measurement response times may also matter. A two-reading calculation that ignores those features can produce a number without establishing the intended property.
The assumptions are not decorative qualifications placed after the answer. They determine whether the equation connects the observed quantities in the first place. Check them before choosing which data to insert.
For this guide, treat every worked case as an explicitly idealised model unless a section changes an assumption. Real interpretation requires appropriate technical expertise, instrument information and a system-specific method. The learning objective is to recognise which evidence the simple inference needs and which claims remain outside it.
CHAPTER 4 OF 21 · Understand the question
4. Calculate one conductance from independently known throughput
Suppose the fictional throughput is Q = 0.12 mbar L/s. The chamber pressure is pA = 0.009 mbar, while the pump-end pressure is pB = 0.003 mbar. The measured difference is therefore 0.006 mbar.
Conductance is C = 0.12/0.006 = 20 L/s. This is the effective conductance of the defined connection under the stated conditions. It is not automatically the catalogue value of any one fitting inside that connection.
Calculate the corresponding chamber-effective speed: Seff = Q/pA = 0.12/0.009 = approximately 13.33 L/s. The simplified local pumping speed is S = Q/pB = 0.12/0.003 = 40 L/s. The three results differ because the denominators describe different physical relationships.
| Quantity | Calculation | Fictional result |
|---|---|---|
| Connection pressure difference | 0.009 − 0.003 | 0.006 mbar |
| Connection conductance | 0.12 / 0.006 | 20 L/s |
| Local pump speed | 0.12 / 0.003 | 40 L/s |
| Chamber-effective speed | 0.12 / 0.009 | 13.33 L/s, approximately |
An immediate check is that the chamber-effective speed is lower than either C or S in this positive finite series model. If a calculation produces Seff larger than both, review the equation, measurement boundary and assumptions before accepting the answer.
CHAPTER 5 OF 21 · Work the first cases
5. Derive the series relation from the same readings
Starting with Q = C(pA − pB), write pA = Q/C + pB. Under the simple pump relation pB = Q/S, this becomes pA = Q(1/C + 1/S). Since Seff = Q/pA, the reciprocal relation is 1/Seff = 1/C + 1/S.
For the previous example, 1/20 + 1/40 = 3/40 s/L. Its reciprocal is 40/3 L/s, or approximately 13.33 L/s. This agrees with the independently calculated Q/pA.
The derivation explains why the pump’s local speed does not simply pass unchanged through a restrictive connection. The total pressure required at the chamber includes the drop across the connection and the pressure associated with the pump’s finite removal speed.
Keep the domain restrictions visible. If C and S are positive and finite, Seff = CS/(C + S), which is less than each. If one becomes very large, the other still limits the result. These are mathematical consequences of this model, not a universal description of every vacuum system.
Ask a learner to explain the two terms before substituting numbers. Being able to recite the reciprocal formula is less valuable than recognising which measured region contributes each term and why both belong in the chamber’s response.
CHAPTER 6 OF 21 · Work the first cases
6. Estimate conductance when throughput is inferred from a pump model
Sometimes Q is not independently measured. An analyst instead uses a known local speed S and the pump-end pressure to infer Q = S pB. Substitution gives C = S pB/(pA − pB).
Take S = 50 L/s, pB = 0.002 mbar and pA = 0.007 mbar. The inferred throughput is 0.10 mbar L/s. The pressure difference is 0.005 mbar, yielding C = 20 L/s.
This calculation is conditional on the pump model and the speed value being appropriate for the actual gas and operating point. It is not an independent measurement of C in the same sense as a calculation using an independently established Q. If the assumed S is wrong by a factor, the inferred C inherits that factor when the pressures are held fixed.
For example, an actual local speed of 40 L/s with the same readings would imply Q = 0.08 mbar L/s and C = 16 L/s. The two pressure readings alone do not decide between those speed assumptions.
A clear report distinguishes “inferred using this assumed pump speed” from “estimated using an independent throughput.” That distinction allows a reviewer to identify which additional evidence would most improve the estimate. It also prevents a circular argument in which an assumed pump property is later claimed to have been verified by the same calculation.
CHAPTER 7 OF 21 · Work the first cases
7. Use a pressure ratio without losing the absolute information
Let r = pA/pB. Under the same simple model, r = 1 + S/C, so C = S/(r − 1). This can be a convenient algebraic form when a pump speed is supplied.
If r = 3 and S = 40 L/s, the inferred conductance is 40/(3 − 1) = 20 L/s. The result agrees with the earlier case in which 0.009/0.003 = 3.
However, a ratio alone does not establish throughput. Pressures of 0.009 and 0.003 mbar have the same ratio as 0.09 and 0.03 mbar. If the same constant-speed model applied, the second pair would correspond to ten times the throughput.
Furthermore, the model may not apply equally across those pressure ranges. A constant ratio is therefore evidence to interpret, not permission to disregard the absolute values. Preserve the original readings and units alongside any ratio summary.
A useful exercise is to ask what remains unknown when only r = 3 is provided. Without S, no numerical C follows. Without an absolute pressure and a suitable speed or throughput measurement, no numerical Q follows. The equation makes the missing information visible instead of encouraging the reader to invent a convenient scale.
CHAPTER 8 OF 21 · Work the first cases
8. Interpret a very small pressure difference cautiously
When pA and pB are close, the denominator in C = Q/(pA − pB) is small. Small changes in either reading can then cause a large change in the inferred conductance.
Suppose Q = 0.12 mbar L/s, pA = 0.0032 mbar and pB = 0.0030 mbar. Using the displayed central values gives C = 0.12/0.0002 = 600 L/s. That large result is not automatically a precisely known material property.
Imagine each pressure has a stated absolute bound of ±0.0001 mbar. Conservative subtraction gives a possible difference from 0 to 0.0004 mbar. Under those bounds, the data do not provide a finite upper bound for C through this simple ratio. The lower difference could approach zero.
With exact Q and a positive difference no greater than 0.0004 mbar, the model supports C at least 300 L/s, but not a sharply bounded value of 600 L/s. A report should not turn unresolved subtraction into confident precision.
This is an example of an ill-conditioned inference. The gauges can be useful instruments and the arithmetic can be exact, yet the particular difference being estimated can remain poorly resolved. The right response is to examine measurement suitability and the question being asked, not merely to display more decimal places in the final quotient.
CHAPTER 9 OF 21 · Test the limits
9. Bound the result when both pressure and throughput are uncertain
Consider the earlier 20 L/s case, but now treat Q as lying between 0.11 and 0.13 mbar L/s. Let pA lie between 0.0088 and 0.0092 mbar, and pB between 0.0028 and 0.0032 mbar.
The conservative positive pressure-difference interval is 0.0056 to 0.0064 mbar. With no additional dependence information, a lower bound for C is 0.11/0.0064 = approximately 17.19 L/s. An upper bound is 0.13/0.0056 = approximately 23.21 L/s.
These are interval bounds under the specified input bounds. They are not a confidence interval unless a suitable statistical model has been supplied. Describing them as “95% confidence” would add a probability statement that the dataset does not contain.
The calculation also illustrates why pairing all low values together is not always conservative. To minimise a positive quotient, use the smallest numerator and largest denominator. To maximise it, use the largest numerator and smallest positive denominator.
If the readings share calibration errors or other dependence, the simple independent-bound treatment may be overly broad or incomplete for a particular purpose. Preserve what is known about those relationships. The essential teaching point is that uncertainty in a difference and uncertainty in a ratio should be propagated through the actual equation rather than attached as an arbitrary percentage afterward.
CHAPTER 10 OF 21 · Test the limits
10. Distinguish common scaling error from differential error
Suppose both gauges report every true pressure multiplied by the same positive factor k. Their measured pressure difference is also multiplied by k. If Q is independently known, the conductance calculated from that difference is divided by k.
If Q is instead inferred as S times the same scaled pump-end reading, both numerator and denominator are multiplied by k. In that particular calculation, the common factor cancels. This cancellation depends on the assumed common scale and the shared measurement chain; it does not mean calibration is generally irrelevant.
An additive offset behaves differently. If the same offset is added to both pressures, it cancels in their difference, but it does not cancel in the inferred throughput S pB. Different offsets in the two gauges can directly distort the difference.
For a simple numerical check, add 0.001 mbar to both pressures in the original example. Their difference remains 0.006 mbar. With independently known Q = 0.12, C remains 20 L/s. But using S = 40 and the offset pump reading of 0.004 would infer Q = 0.16 and C approximately 26.67 L/s.
These contrasting cases show why “the errors cancel” needs a derivation. State which errors, which equation and which shared factors are involved. A plausible cancellation in one measurement arrangement may fail in another.
CHAPTER 11 OF 21 · Test the limits
11. Test whether a constant-conductance model fits several states
One operating point can always produce a quotient when its denominator is positive. Several independently measured states can test whether one constant C describes the connection over the chosen range.
Suppose three fictional measurements give pressure differences of 0.002, 0.004 and 0.006 mbar, with corresponding throughputs of 0.04, 0.08 and 0.12 mbar L/s. Each yields C = 20 L/s. This consistency supports the constant-C model over those measured states, subject to measurement limitations.
Now suppose the throughputs are 0.04, 0.07 and 0.12 instead. The inferred conductances are 20, 17.5 and 20 L/s. Whether the middle result is inconsistent depends on the actual uncertainty and conditions. Do not declare a physical change from a small discrepancy without examining those details.
Plot or tabulate Q against the pressure difference. In the ideal constant-C model with no offset, the relationship passes through the origin with slope C. A persistent intercept can prompt investigation of offsets, additional gas sources, model scope or data treatment. It does not identify one cause by itself.
A useful report states the measured range over which agreement was observed. It should not extrapolate a locally successful model to a different gas, geometry or flow regime without additional justification.
CHAPTER 12 OF 21 · Test the limits
12. Recognise what two endpoint gauges cannot identify
Two endpoint readings can describe the combined pressure drop of a path. They do not normally identify every component’s contribution. For two positive series conductances, 1/Ctotal = 1/C1 + 1/C2.
If Ctotal = 20 L/s, the pair C1 = 40 and C2 = 40 L/s is one solution. The pair C1 = 30 and C2 = 60 L/s is another. Both have reciprocal sum 1/20 s/L.
Consequently, an endpoint estimate of 20 L/s cannot establish whether two components are equally restrictive. It cannot prove that a named valve is the problem while the pipe is satisfactory. Those conclusions require additional measurements, validated component information or another system-specific investigation.
This is an identifiability limit rather than an arithmetic weakness. More precise endpoint readings can narrow the total estimate while leaving the internal split unresolved. Precision and uniqueness are different properties of an inference.
For a transfer exercise, ask students to construct a third positive pair with total 20 L/s. Choosing C1 = 25 gives 1/C2 = 1/20 − 1/25 = 1/100, so C2 = 100 L/s. The growing list of valid internal explanations makes the boundary of the endpoint evidence unmistakable.
CHAPTER 13 OF 21 · Test the limits
13. Compare a connection change with a pump change
Use the idealised baseline S = 40 L/s and C = 20 L/s. The effective chamber speed is approximately 13.33 L/s. Doubling the local pump speed to 80 L/s while leaving C unchanged gives Seff = 80 × 20/100 = 16 L/s.
Doubling the conductance to 40 L/s while leaving S at 40 gives Seff = 40 × 40/80 = 20 L/s. Under these deliberately simplified assumptions, the connection change produces the larger improvement in chamber-effective speed.
This comparison is a model exercise, not an instruction to alter an installed system. Actual changes involve geometry, gas compatibility, operating regime, safety, serviceability and cost. The calculation’s purpose is to challenge the automatic assumption that a larger nominal pump always yields the best proportional result.
At fixed throughput Q = 0.12 mbar L/s, the baseline chamber pressure is 0.009 mbar. The pump-speed change gives 0.12/16 = 0.0075 mbar. The conductance change gives 0.12/20 = 0.006 mbar.
The fixed-throughput condition belongs in the conclusion. If a process change also alters gas load, these pressure comparisons no longer isolate pumping or conductance alone. A good worked example names which variables are held constant before describing improvement.
CHAPTER 14 OF 21 · Transfer and check
14. Do not infer a leak from a pressure difference alone
A pressure difference across a connection indicates the relationship between throughput and conductance under the selected model. It does not establish the origin of the gas. Process gas, surface release and ingress can contribute differently to a real system’s gas load.
Imagine two fictional systems with the same Q, C and S. One receives a controlled process input. Another has the same total throughput from a different combination of sources. Their two endpoint pressures can be identical even though the operational explanations differ.
The inverse problem is therefore underdetermined with respect to source identity. Calling the calculated throughput a “leak rate” adds a claim that was not established by the endpoint relationship. Source investigation requires a suitable method and system context.
The vacuum science root connects to the existing engineering guides that address evidence, chambers and service. Those are more appropriate pathways when the question concerns an actual installation’s behaviour. Keep this article’s algebra within its educational role.
For students, ask what observation would discriminate between competing source explanations. The answer should identify an additional piece of evidence or an appropriately controlled comparison, rather than simply repeating the same pressure calculation. A number can describe a total without identifying the mechanism that produced it.
CHAPTER 15 OF 21 · Transfer and check
15. Treat changing readings as a different problem
The worked steady-state relation assumes that the relevant gas flow and pressure measurements describe a sufficiently stable state. During a transient, gas can accumulate or be depleted within the defined region. Gauges may also respond on different timescales.
Suppose one instrument reports a chamber reading at time t while another reports an averaged pump reading covering an earlier interval. Subtracting them as though they were simultaneous can create an apparent pressure difference that does not belong to any single system state.
Record time alignment and averaging as part of measurement interpretation. A smoothly displayed trend can conceal different response settings or delayed data. More frequent logging does not by itself correct a systematic timing mismatch.
For a classroom model, provide two labelled time series and ask learners which pairs can legitimately be compared. If no time alignment or response information is supplied, the honest result may be that a steady conductance estimate has not yet been established.
Do not take an instantaneous quotient at every displayed point and assume the resulting curve is a changing material property. It could reflect transient storage, measurement lag, model limitations or actual physical change. The next step is to decide which model and evidence are appropriate, not to assign a cause from the curve’s appearance alone.
A fictional connection carries Q = 0.18 mbar L/s. Its chamber-end pressure is 0.012 mbar and pump-end pressure 0.003 mbar. Find the connection conductance, local pump speed and chamber-effective speed under the simple model.
The pressure difference is 0.009 mbar, giving C = 20 L/s. The local speed is S = 0.18/0.003 = 60 L/s. The chamber-effective speed is Seff = 0.18/0.012 = 15 L/s. Check that 1/20 + 1/60 = 1/15.
Now suppose only the two pressures are known. Can the numerical conductance still be found? No. Many throughputs and conductances can produce the same pressure difference. If Q were 0.09 rather than 0.18, C would be 10 L/s.
For a second change, keep Q = 0.18 and let the measured difference be no more than 0.001 mbar but not resolved away from zero. The central-value temptation is to report a very large C. A more careful interpretation asks whether a finite upper bound is supported at all.
The learner should be able to explain why these are different evidence situations. Knowing the equation is only the first step; deciding whether its inputs are actually established is the part that transfers to unfamiliar measurements.
CHAPTER 17 OF 21 · Transfer and check
17. Write a result that another reader can audit
A concise defensible result names the measured boundary, gas and operating context, pressure locations, units, throughput source, model assumptions and uncertainty treatment. It then states the inferred quantity without extending it to hidden components or source diagnoses.
For the first example, a suitable teaching conclusion is: “Under the stated steady constant-conductance model, independently supplied throughput 0.12 mbar L/s and endpoint pressures 0.009 and 0.003 mbar imply a combined connection conductance of 20 L/s.” This sentence says what was inferred and how.
If throughput came from an assumed pump speed, say so. If pressure bounds support only a one-sided statement, retain that limitation. If the measurement region includes a valve and pipe, name the combined path rather than attributing the result to one component.
Keep the original data table with the result. A report that preserves only “20 L/s” cannot be checked later when a gauge correction, unit issue or boundary question arises. Traceable inputs are part of the explanation.
A useful review prompt is to ask what would change the conclusion. A revised throughput, different gauge location or invalid steady-state assumption could do so. Making those dependencies explicit helps readers distinguish a supported local estimate from a permanent equipment label.
CHAPTER 18 OF 21 · Transfer and check
18. Check units before comparing two estimates
Suppose one report gives conductance as 20 L/s and another as 0.020 m³/s. These values are equal because one cubic metre is one thousand litres. A missing conversion can make identical results appear different by three orders of magnitude.
Pressure units also belong in the check. If throughput is supplied in Pa m³/s and pressures in Pa, their quotient is m³/s. Do not insert a pressure difference expressed in mbar into that equation without a consistent conversion. The algebra does not know that two differently labelled numbers refer to the same physical quantity.
For the original example, Q = 0.12 mbar L/s is 0.012 Pa m³/s, and Δp = 0.006 mbar is 0.6 Pa. Dividing gives 0.020 m³/s, or 20 L/s. Reconstructing the result in a second coherent unit system is a useful independent arithmetic check.
Keep precision sensible. A unit conversion does not create additional measurement accuracy. Writing 20.000000 L/s because a spreadsheet can display it would misrepresent the evidence if the inputs are only weakly constrained.
Before comparing results over time, check the units, gas, measurement locations and model assumptions together. A numerical discrepancy can arise from a reporting convention, while identical numbers can conceal different physical conditions. Both possibilities deserve attention before a change is attributed to the connection.
CHAPTER 19 OF 21 · Report and continue
19. Decide which missing measurement would be most informative
Consider three evidence gaps. In the first, two pressures are known but throughput is absent. In the second, throughput is known but the pressure difference is unresolved. In the third, total conductance is well estimated but the analyst wants to identify one internal component.
These gaps call for different additional evidence. An independently established throughput addresses the first. More suitable differential-pressure evidence may address the second. Additional internal information or a validated component-specific investigation is needed for the third. Simply repeating the original pair of readings does not necessarily solve any of them.
This is a useful exercise in planning a measurement before collecting it. Ask what range of answers remains compatible with the current data, then identify which observation would make those answers disagree. The purpose is not to accumulate a larger file of readings but to reduce the uncertainty relevant to the actual claim.
For example, if the total conductance is exactly 20 L/s in the teaching model, endpoint precision cannot choose between internal pairs 40 and 40, 30 and 60, or 25 and 100 L/s. That uncertainty is structural, so it needs a different kind of information.
A good conclusion can therefore include a concrete next evidence need. Naming it shows understanding of the result’s practical limit and prevents an attractive numerical estimate from being mistaken for a complete diagnosis.
CHAPTER 20 OF 21 · Report and continue
20. Questions about conductance inferred from pressures
Do two pressure readings determine conductance?
Not alone. The basic equation also needs throughput, or an additional justified relationship such as an appropriate local pumping-speed model. Two pressures provide a difference, not the missing flow scale.
Can the chamber and pump have different pressures?
Yes, in the flowing model used here. The connection has a pressure difference associated with finite conductance. Always distinguish the actual measurement locations rather than treating “the vacuum pressure” as one location-free property.
Does nearly equal pressure prove infinite conductance?
No. The difference may be small or poorly resolved. A quotient with a denominator near zero can have a weakly constrained upper value. Measurement bounds and model assumptions determine what can be stated.
Can endpoint data identify a restrictive valve?
They can estimate the combined path under suitable conditions, but generally cannot uniquely divide that total among unseen series components. More precise endpoints do not automatically solve an internal identifiability problem.
Is this a method for selecting or operating a vacuum system?
It is a mathematical and evidence-reading guide. Real equipment selection, changes and operation require appropriate technical expertise, manufacturer information and system-specific controls. The examples are not operating specifications.
CHAPTER 21 OF 21 · Report and continue
21. Connect the calculation with the wider vacuum curriculum
Return to Vacuum Science for pressure, molecular behaviour, surfaces, gauges and the grouped subject pathways. Continue into Vacuum Engineering and Vacuum Systems Engineering for their distinct engineering contexts.
For technical reference, use Leybold’s conductance fundamentals, its pump-size discussion, and Kurt J. Lesker’s pumping speed and conductance explanation. The references support the physical framework; all numerical comparisons in this guide are original educational constructions.
The broader connection to Learning That Transfers into Life is the discipline of preserving measurement locations, assumptions and uncertainty through a calculation. Take a new pair of readings, draw their boundary and list what is independently known. That is the first useful step toward a result whose meaning survives inspection.
Find the wider science and systems routes in How X Works.