space-rha.comShort courseChapter 3 · RHA requirements

Chapter 3 · Lesson 3 of 6 · about 10 minutes

Writing a TID and DDD requirement

A TID requirement is more than a dose. Behind the number are a choice of solar-particle model and statistical case, a local dose calculation, a margin with a reason, and a decision about what evidence will show that the part still performs. The number alone leaves all of those open.

A TID requirement is a chain of decisions.

TID

State the absorbed dose with its material, such as krad(Si), at a named location, for the declared lifetime and environment case. Without the location and the case, the number cannot be reproduced.

DDD / TNID

State the convention (a displacement damage dose or an equivalent fluence), the target material, and how it is tied to the response of the technology in question. A bare proton fluence, with no energy and no material, is not a complete damage requirement.

The evidence route is part of the requirement

A board-level TID test can be useful evidence for the application when the source, the dose rate, the dose reached, the operating modes and what the test can observe are all appropriate. It generally does not reveal every part parameter, the worst-case bias for each part, or the variation across the flight lot, so what it is expected to show should be decided when the requirement is written.

Turn the local exposure into a testable criterion

The simplest form of the criterion multiplies the declared local exposure by a design factor that the project has chosen and written down. That gives a test dose. It does not give the circuit limits that define failure, the evidence that the part's response applies to this application, or the conditions under which the test is run, and the criterion is not complete without them.

TID conditions. The requirement should state the bias during irradiation, the temperature, the dose rate, the sequence of measurements and anneals, and the electrical limits that define a failure. Where the technology is sensitive to dose rate, say how that is to be handled. A test dose on its own does not define the test.

DDD conditions. One common convention is 1 MeV neutron-equivalent fluence in Si. Whatever convention is used, state the NIEL or hardness-factor basis, the reference material and energy, the local spectrum it was derived from, and the evidence that the scaling holds for this technology.

SPE statistics. Choose a model for the mission cumulative solar-particle fluence and the statistic taken from it, and record the mission duration and the solar-cycle assumptions that went into it. A largest-event case and a peak-flux case answer different questions, so say which one the dose is for. SAPPHIRE, PSYCHIC and ESP are different models, and a percentile from one generally cannot be read across to another.

Board-level acceptance. Decide up front whether the board test adds to the part-level evidence or is offered as the primary evidence for a bounded application. Either way, specify the source, the dosimetry at the board, the dose rate and any ELDRS consideration, the powered and unpowered modes, which parameters are monitored, which samples and lots the result applies to, and the checks made after irradiation.

3D scene

Dose, multiplier and endpoint

The predicted local dose, the required level after the margin factor, and the demonstrated endpoint, as three columns. Change any of the three.

  1. The first column is the absorbed dose at the part: a number that is only reproducible with its material, its location, the declared lifetime and the environment case behind it. The columns here are illustrative heights, 12 krad(Si) in this teaching case, with no confidence level implied.

  2. The middle column multiplies the local dose by a factor the project has chosen and written down, giving a test dose of 24 krad(Si). The factor is a policy choice with a reason behind it (lesson 5); it does not define the test, which still needs the bias, the temperature, the dose rate, the sequence of measurements and anneals, and the electrical limits that define a failure.

  3. The third column is what the evidence shows: a part that stayed within its limits to 30 krad(Si) under stated conditions. The requirement is met with conditions attached, and those conditions (the lot, the bias, the dose rate, the parameters measured) are part of the evidence, not a footnote to it.

  4. Move the part to a location where the local dose is 25 krad(Si): the test dose becomes 50 and the demonstrated endpoint no longer covers it. The gap can be closed with more evidence, a different part, local shielding, or a smaller factor with a written reason; which of these is chosen is a project decision, and the requirement should record it.

The original space-rha.com environment and mission renderer (NASA imagery and spacecraft models, credits in the view), with the course scenes drawn by the same renderer. Trails, clocks and counts are illustrative; the scene computes no dose or rate.

Worked example: from the environment sheet to a TID requirement

The question. The 800 km, 98.6°, five-year mission of chapter 1 has 12,900 rad(Si) of trapped dose behind 100 mil of aluminum (solid sphere, AE-8 MAX and AP-8 MAX) and 505 rad(Si) per August 1972 event through the quiet cutoff. A bipolar regulator on a board behind a chassis wall is to be specified. What does the requirement say?

Steps.

  1. Replace the reference shielding with the part's own. Sector analysis of the box (chapter 2) gives, say, an effective 2.0 g/cm² for this part; the chapter 1 curve at that depth gives 336 rad(Si) per year of trapped dose, 1,682 rad(Si) in five years. The solar term comes from the ESP active-year spectrum through the same geometry; the 1972 event at this depth is 233 rad(Si).
  2. Add the project's solar term at its confidence level. With, for illustration, three active years of the mission at 95 % confidence, the ESP spectrum through SHIELDOSE-2 might add a few hundred rad(Si) at this depth; the exact figure is the open item of chapter 1, lesson 6. Call the local total 1,982 rad(Si).
  3. Apply the project's radiation design margin. NASA guidance uses a factor of 2 on the predicted local dose as a common policy where the environment and geometry carry ordinary uncertainty (NASA/TM-20210018053, chapter 5); the factor is a policy, not a statistic (lesson 5). The requirement becomes: the part shall meet its application limits after 3,963 rad(Si), rounded to 4,000 rad(Si).
  4. State the conditions that make the number verifiable: the dose-rate condition (a bipolar part needs a low-dose-rate or ELDRS-aware test, chapter 4, lesson 5), the bias during exposure, the parameters and limits that define failure in this circuit (output voltage, dropout and leakage for a regulator, not a data-sheet table), the lot and the sample size with its statistical basis, and the material of the unit (rad(Si)).

Result. "U1 shall remain within its application limits (table X) after a total ionizing dose of 4,000 rad(Si), demonstrated by a low-dose-rate test per MIL-STD-883 Method 1019 condition D, under the bias of table Y, on N samples of the flight lot, with the acceptance statistic of section Z." Every number in that sentence traces to a line of the environment sheet, the sector analysis, the margin policy and the application analysis, which is what lesson 6 of this chapter asks for.

Exercise

Draft a cumulative-effects criterion for one part, with the right units and with conditions that a test engineer could set up.

Keep three numbers separate in the draft: the exposure, the required multiplier, and the endpoint the test actually demonstrated.

Going deeper

The failure criterion and its uncertainty

A data-sheet violation, a loss of circuit margin and a loss of mission function are three different endpoints, and a test report should say which one it used. The highest dose at which the part passed is a lower bound on where that endpoint lies, under the conditions tested, and it is not the dose at which the part failed.

RDM convention. The NASA guideline defines the margin as the failure level divided by the specification level. ECSS defines it as the sensitivity divided by the received level. Write out the numerator and the denominator you are using, and keep the minimum margin the project requires separate from the ratio a test demonstrated.

Scaling limits. NIEL scaling can fail for particular technologies, particle energies, and defect and anneal conditions, so the requirement should not assume it holds without evidence. Combined TID and DDD, the temperature history and long-term annealing can each require evidence beyond the scaled fluence.

What the probability means. The confidence level attached to a model fluence is the probability that the fluence will not be exceeded. It is not, by itself, a confidence in the absorbed dose, in the survival of the part or in the performance of the system. Record the model and its version, how the statistic is defined, how the active years of the solar cycle were treated, the transport assumptions and the model used for the spread of part response.

Source equivalence. Co-60 gamma gives a penetrating TID exposure, once the dose rate and the geometry have been checked. Protons deposit dose too, but they also produce displacement damage and single events at the same time. With electrons, the energy and the geometry of the assembly decide how much of the part is actually exposed. Low-energy X rays need an explicit assessment of penetration and of dose enhancement at material interfaces. Match the source to the mechanism you are trying to exercise; a single rad(Si) number does not make two sources equivalent.

Takeaway

Specify the SPE statistic, the reason for the margin and the evidence route, as well as the dose.

Check yourself

Five parts passed 30 krad with no parametric failure. What is the measured failure dose?

Unknown; it is above 30 krad for those five parts. A passing dose is a lower bound on the failure dose of the sample, and the lot statement needs the failure doses or a tolerance argument (chapter 8), not the passing level.

What does the design margin on dose protect against, and what does it not protect against?

It generally covers the uncertainty of the environment model, the shielding analysis and part-to-part variation at the stated dose rate. It does not by itself cover a dose-rate effect (ELDRS) in a bipolar part, a different lot, or a parameter the test did not measure; each of those needs its own evidence.

The environment sheet of the worked example gives 12,900 rad(Si) behind 100 mil of aluminum for the five-year mission. A colleague multiplies it by the project's factor of 2 and writes the product as the regulator's requirement. What did that skip?

The reference shielding has to be replaced by the part's own before anything is multiplied: sector analysis gives an effective 2.0 g/cm² for this part, where the trapped dose is 336 rad(Si) per year and 1,682 rad(Si) in five years, and the solar term is recomputed through the same geometry (233 rad(Si) for the 1972 event at this depth, a few hundred rad(Si) for the project's confidence-level case), for a local total of 1,982 rad(Si). The factor of 2 then applies to the local dose, giving 3,963 rad(Si), rounded to 4,000. Even then the number is a test dose and not a requirement until the dose-rate condition, the bias, the parameters and limits that define failure, the lot and sample size, and the material of the unit are stated with it.

The solar-particle fluence in the requirement is taken at 95 % confidence, and the project reports the part as having a 95 % probability of surviving the mission dose. What is wrong?

The confidence level attached to a model fluence is the probability that the fluence will not be exceeded. It is not, by itself, a confidence in the absorbed dose, in the survival of the part or in the performance of the system, each of which carries uncertainties of its own (the transport, the spread of part response, the definition of failure). The record should give the model and its version, how the statistic is defined, how the active years of the solar cycle were treated and the transport assumptions; a largest-event case and a peak-flux case answer different questions, and a percentile from SAPPHIRE, PSYCHIC or ESP generally cannot be read across to another model.

A displacement-damage requirement is written as a proton fluence per cm² with nothing else. What is missing before a test engineer can act on it?

A bare proton fluence, with no energy and no material, is not a complete damage requirement. It needs the convention (a displacement damage dose or an equivalent fluence, such as 1 MeV neutron-equivalent fluence in Si), the NIEL or hardness-factor basis, the reference material and energy, the local spectrum it was derived from and the evidence that the scaling holds for this technology. NIEL scaling can fail for particular technologies, particle energies, and defect and anneal conditions, and combined TID and DDD, the temperature history and long-term annealing can each require evidence beyond the scaled fluence.

Further reading