"LLM not configured" appears in the Generate AI Report menu


The AI Reporting feature requires an API key from a supported LLM provider before it can send requests. Open AI signal report, then Settings, and configure your provider, API key, and model. Use the Test Connection button to verify that your settings are correct. Once configured, the menu item will show the active model name and the analysis submenu will become available.

Test Connection fails


There are several common causes for connection test failures.


Your API key may be incorrect or expired. Double-check that you copied the complete key, including any prefix (such as "sk-ant-" for Anthropic or "sk-" for OpenAI). Some providers invalidate keys that have not been used for an extended period.


Your account may not have billing enabled or may have insufficient credits. Most providers require a payment method on file before API requests will be accepted. Visit your provider's account dashboard to check your billing status and credit balance.


Your network may be blocking outbound HTTPS connections to the provider's API endpoint. This can happen in corporate or institutional networks with restrictive firewalls. Contact your network administrator if you suspect this is the case. The endpoints used are api.anthropic.com for Anthropic, api.openai.com for OpenAI, and generativelanguage.googleapis.com for Google Gemini.


If you are using a Custom provider, verify that the Base URL is correct and that the endpoint is reachable from your machine.

Analysis takes a long time or times out


LLM responses typically arrive within 5 to 30 seconds depending on the model, the analysis mode, and the size of the signal's metric report. Longer signals produce larger metric reports with more data points, which may increase processing time slightly.


If requests consistently take more than 30 seconds or time out, try switching to a faster model. Claude Haiku, GPT-4o mini, and Gemini Flash are the fastest options among the supported models. Premium models like Claude Sonnet, GPT-4o, and Gemini Pro provide higher quality analysis but may take longer to respond.


Timeouts can also be caused by network latency or temporary overload on the provider's servers. If a single request times out, try again after a short wait.

The analysis seems inaccurate or too generic


Several factors can affect the quality of AI-generated analyses.


Model choice matters. Budget models (such as GPT-4o mini or Gemini Flash) are fast and affordable but may produce less detailed or less nuanced interpretations. If the analysis seems shallow, try a more capable model such as Claude Sonnet, GPT-5, or Gemini Pro.


The analysis mode should match the signal type. Domain-specific modes (Vibration Analysis, Audio Signal Analysis, Electrical / Power Signal Analysis, Biomedical Signal Analysis, and Predictive Maintenance Assessment) are designed for particular types of signals and will produce the best results when applied to those types. If you apply a domain-specific mode to the wrong signal type, the AI will typically note the mismatch and provide a more general assessment, but the results will not be as useful as they would be with the correct mode. When in doubt, start with the General Signal Description mode, which works well for all signal types.


Keep in mind that the AI analyzes only the computed metrics, not the raw waveform. If important signal characteristics are not captured by the available metrics, they will not appear in the analysis.

The browser does not open after analysis


SIGVIEW saves the analysis report as an HTML file in your system's temporary folder and opens it using your default web browser. If the browser does not open, verify that you have a default web browser configured in Windows. You can check this in Windows Settings under Apps, then Default apps.


If the browser opens but displays nothing, the HTML file may not have been created successfully. Check that you have write access to your system's temporary folder (typically C:\Users\YourName\AppData\Local\Temp).


Old analysis report files in the temporary folder are cleaned up automatically when SIGVIEW starts, so they will not accumulate over time.

The analysis prompt does not copy to the clipboard


When using the Copy Analysis Prompt to Clipboard function, ensure that no other application is locking the Windows clipboard. Some clipboard manager tools or remote desktop sessions can occasionally interfere with clipboard operations. Try closing other clipboard-related applications and attempt the copy again.


The prompt text can be quite long (typically 2000 to 3000 characters). All standard LLM chat interfaces can handle prompts of this size without issues.

API costs seem higher than expected


The cost per analysis depends on the model used and the size of the signal report. A typical analysis costs between $0.005 and $0.03 USD. Reports from longer signals or signals with more metrics will use slightly more input tokens and cost slightly more.


To minimize costs, use budget models (Claude Haiku, GPT-4o mini, or Gemini Flash) for routine analyses and reserve premium models for signals that require the highest quality interpretation.


You can monitor your API usage and spending through your provider's dashboard: console.anthropic.com for Anthropic, platform.openai.com for OpenAI, or aistudio.google.com for Google.

Error messages from the LLM provider


If SIGVIEW receives an error from the LLM provider, the error message is displayed in a dialog box. Common errors include:


"Invalid API key" or "Authentication failed": Your API key is incorrect, expired, or has been revoked. Generate a new key from your provider's dashboard and update it in Settings.


"Rate limit exceeded": You are sending too many requests in a short period. Wait a moment and try again. If this happens frequently, check your provider's rate limits for your account tier.


"Model not found": The model ID in your settings does not match any model available from the selected provider. Check for typos in the model name, or select a different model from the dropdown in Settings.


"Insufficient quota" or "Billing not active": Your account has run out of credits or does not have a payment method configured. Add credits or a payment method through your provider's billing page.