On QRS amplitude and other errors produced by direct-writing electrocardiographs