Michael D. Schrock, a city lawyer, said the settlement meant the city would not have to worry about possible litigation with Brekford and could move forward in selecting a new contractor.

Problematic from start

Brekford was supposed to represent a fresh start for Baltimore when officials awarded the camera contract to the firm over longtime incumbent Xerox in November 2012.

By then, city transportation officials had known for months about erroneous speed readings from a radar-equipped Xerox camera on West Cold Spring Lane. Days later, such problems began spilling into public view as a Sun investigation documented inaccurate citations at seven cameras across the city. In one case, a Mazda was ticketed for speeding while stopped a red light.

Around the time that the city switched vendors, a lobbyist for Xerox warned officials that Brekford lacked the experience to run a program as big as Baltimore's, with a resume limited to more modest operations in cities like Salisbury and Hagerstown.

Maurice Nelson, Brekford's managing director, insisted that the company was up to the challenge. "We're not promising perfection, but we're promising we'll strive for it," Nelson said last December.

Though Brekford formally took over the city's camera program in January, a bumpy handoff from the previous vendor kept it from issuing citations until late February. For red-light cameras, Brekford used the 81-camera network supplied by Xerox; it began rolling out its own speed cameras featuring so-called tracking radar.

On April 16, The Sun asked transportation officials why a Brekford speed camera on The Alameda had been programmed for a 25 mph speed limit when posted signs showed the actual limit to be 30 mph. The discrepancy meant some drivers were being ticketed despite not going at least 12 mph over the limit, the legal threshold for issuing an automated citation.

That afternoon, the city announced that the entire speed and red-light camera program had been halted indefinitely, citing the Alameda camera error and a problem with how payment options were listed on tickets.

The suspension signaled an end to the flow of badly needed revenue — speed cameras generated $50 million for the city over three years. Transportation officials said it also took away a public safety tool that helped deter speeding in areas around schools, even as critics pointed out that many cameras were blocks away. The shutdown brought a new public embarrassment months after the problems with Xerox had dominated headlines.

From April until last week, city officials said little publicly about the status of the camera program.

But out of public view, the city and Brekford had spent more than three months trying to fix a range of problems so enforcement could resume.

The testing was monitored by URS, hired under a $278,000 contract to provide "specialized monitoring services," ensure that "protocols and policies are properly implemented," and prepare monthly progress reports.

Documents released through the public records request show the consultant chronicled test results almost weekly from mid-July to late October. Apart from a draft report dated Aug. 16, most of the information is contained in grids and charts, with no accompanying text to provide further details.

From the start, the records show, there were problems.

During the latter half of July, for example, URS noted difficulty some days in turning the cameras on and off. Maintenance by Brekford caused "misalignment" with one camera, an issue the company said it would address. And Police Department supervisors charged with reviewing tickets for accuracy reported having trouble accessing an online citation queue.

In addition, the consultant wrote, Brekford's speed cameras had a "high controllable error rate," meaning tickets were invalid because of an equipment malfunction, a blurry license plate photograph or another reason considered within the company's control.

Of 1,676 potential tickets, one in five had to be discarded because of a controllable error, URS found.

Over the next three months, the controllable error rate fluctuated, dropping as low as 4 percent and rising as high as 30 percent, records show.

Some problems endured for weeks, URS found. For five weeks, the consultant noted that Brekford had not submitted certain data reports. In early September the reports were "finally submitted." But over the next seven weeks, the company either did not provide reports or they were late or incomplete.