Support for downloading binary files in using ajax is not great, it is very much still under development as working drafts.

Simple download method:

You can have the browser download the requested file simply by using the code below, and this is supported in all browsers, and will obviously trigger the WebApi request just the same.

$scope.downloadFile = function(downloadPath) { 
    window.open(downloadPath, '_blank', '');  
}

Ajax binary download method:

Using ajax to download the binary file can be done in some browsers and below is an implementation that will work in the latest flavours of Chrome, Internet Explorer, FireFox and Safari.

It uses an arraybuffer response type, which is then converted into a JavaScript blob, which is then either presented to save using the saveBlob method - though this is only currently present in Internet Explorer - or turned into a blob data URL which is opened by the browser, triggering the download dialog if the mime type is supported for viewing in the browser.

Internet Explorer 11 Support (Fixed)

Note: Internet Explorer 11 did not like using the msSaveBlob function if it had been aliased - perhaps a security feature, but more likely a flaw, So using var saveBlob = navigator.msSaveBlob || navigator.webkitSaveBlob ... etc. to determine the available saveBlob support caused an exception; hence why the code below now tests for navigator.msSaveBlob separately. Thanks? Microsoft

// Based on an implementation here: web.student.tuwien.ac.at/~e0427417/jsdownload.html
$scope.downloadFile = function(httpPath) {
    // Use an arraybuffer
    $http.get(httpPath, { responseType: 'arraybuffer' })
    .success( function(data, status, headers) {

        var octetStreamMime = 'application/octet-stream';
        var success = false;

        // Get the headers
        headers = headers();

        // Get the filename from the x-filename header or default to "download.bin"
        var filename = headers['x-filename'] || 'download.bin';

        // Determine the content type from the header or default to "application/octet-stream"
        var contentType = headers['content-type'] || octetStreamMime;

        try
        {
            // Try using msSaveBlob if supported
            console.log("Trying saveBlob method ...");
            var blob = new Blob([data], { type: contentType });
            if(navigator.msSaveBlob)
                navigator.msSaveBlob(blob, filename);
            else {
                // Try using other saveBlob implementations, if available
                var saveBlob = navigator.webkitSaveBlob || navigator.mozSaveBlob || navigator.saveBlob;
                if(saveBlob === undefined) throw "Not supported";
                saveBlob(blob, filename);
            }
            console.log("saveBlob succeeded");
            success = true;
        } catch(ex)
        {
            console.log("saveBlob method failed with the following exception:");
            console.log(ex);
        }

        if(!success)
        {
            // Get the blob url creator
            var urlCreator = window.URL || window.webkitURL || window.mozURL || window.msURL;
            if(urlCreator)
            {
                // Try to use a download link
                var link = document.createElement('a');
                if('download' in link)
                {
                    // Try to simulate a click
                    try
                    {
                        // Prepare a blob URL
                        console.log("Trying download link method with simulated click ...");
                        var blob = new Blob([data], { type: contentType });
                        var url = urlCreator.createObjectURL(blob);
                        link.setAttribute('href', url);

                        // Set the download attribute (Supported in Chrome 14+ / Firefox 20+)
                        link.setAttribute("download", filename);

                        // Simulate clicking the download link
                        var event = document.createEvent('MouseEvents');
                        event.initMouseEvent('click', true, true, window, 1, 0, 0, 0, 0, false, false, false, false, 0, null);
                        link.dispatchEvent(event);
                        console.log("Download link method with simulated click succeeded");
                        success = true;

                    } catch(ex) {
                        console.log("Download link method with simulated click failed with the following exception:");
                        console.log(ex);
                    }
                }

                if(!success)
                {
                    // Fallback to window.location method
                    try
                    {
                        // Prepare a blob URL
                        // Use application/octet-stream when using window.location to force download
                        console.log("Trying download link method with window.location ...");
                        var blob = new Blob([data], { type: octetStreamMime });
                        var url = urlCreator.createObjectURL(blob);
                        window.location = url;
                        console.log("Download link method with window.location succeeded");
                        success = true;
                    } catch(ex) {
                        console.log("Download link method with window.location failed with the following exception:");
                        console.log(ex);
                    }
                }

            }
        }

        if(!success)
        {
            // Fallback to window.open method
            console.log("No methods worked for saving the arraybuffer, using last resort window.open");
            window.open(httpPath, '_blank', '');
        }
    })
    .error(function(data, status) {
        console.log("Request failed with status: " + status);

        // Optionally write the error out to scope
        $scope.errorDetails = "Request failed with status: " + status;
    });
};

Usage:

var downloadPath = "/files/instructions.pdf";
$scope.downloadFile(downloadPath);

Notes:

You should modify your WebApi method to return the following headers:

• I have used the x-filename header to send the filename. This is a custom header for convenience, you could however extract the filename from the content-disposition header using regular expressions.

• You should set the content-type mime header for your response too, so the browser knows the data format.

I hope this helps.