A review by Jeffrey Borinsky FIEE C.Eng.
It’s standards converter time again! Vintage TV enthusiasts now have a choice of two standards converters. The well established Domino has now been joined by the Aurora. Inevitably I will compare the two products in this review. If you’re about to skip to my conclusions I can say now that both are excellent products, each with its own mix of strengths.
I reviewed the Domino in 2002. It was, and still is, a very good design that has sold well.
The new product is the Aurora multistandard converter, designed by Darryl Hock who lives and works near Detroit. This can not only produce 405 line pictures but a whole range of other standards from 819 all the way down to 30 lines. It can also accept PAL, NTSC or SECAM inputs. It does not have a modulator. I have used a prototype unit for this review but I have also seen a production unit. I will highlight any important differences.
I feel I need to make this point in all reviews of converters. Prices of £300 to £400 may seem high but the economics of small scale production make it inevitable. TVs, computers and other consumer products are only cheap because they are made in huge numbers. PCBs and components in small quantities are a lot more expensive than in bulk. So is assembly, especially for fine pitch surface mount devices as used in Aurora. I reckon that Darryl is making very little profit on each unit.
If you have more time than money there is now a design that could realistically be built by a skilled enthusiast. Darius in Germany has designed and built an analogue converter. The vital line memories are charge coupled delay lines that are normally used as part of the dropout compensator in a VCR. If you are interested you can contact Darius by email at:
At least one copy has been built successfully by Peter Scott, a UK enthusiast. You could probably buy all the bits for not much over £100 and some might be in your junk box anyway. There are no PCBs available for this design. Strictly a Veroboard job. I have seen it working and the results are better than you might expect given the simple methods employed. Its main failing is very limited bandwidth. This is inherently limited to about 1.8MHz (at the 405 line output) by the clock frequency used in the delay lines. Fine detail above this frequency is severely aliased which is very obvious on test card but less of a problem on real pictures.
After this digression it’s time to get back to the Aurora.
My first impression of the Aurora was “It’s tiny!” About 4”x3”x1.5”. The power supply is an external “wall wart”. This guarantees electrical safety and is an appropriate choice. The one supplied with Aurora will work on any mains voltage and has interchangeable clip-on adapters for UK, US and several others.
The technology under the lid is very modern. All the digital processing is in a Xilinx programmable logic device. This is flanked by some large SRAMs acting as framestores, a Philips SAA7113 multistandard decoder, a flash memory to store the Xilinx code, more flash to store test patterns and a fairly small number of other parts. A hex switch is used to select the output standard. The Xilinx can easily be reprogrammed; more about this later.
PERFORMANCE AT 405
It is hard to fault the performance of the Aurora. It has excellent high frequency response, performs well with all sorts of input signals and has excellent interpolation. For 405 line outputs Aurora interpolates 3 lines from the same field. This gives interpolation quality which is theoretically better than anything except the BBC CO6/509 which used four lines. In practice it is difficult to see any improvement over the 2 line interpolators used in the Domino. Other standards use a mixture of vertical and horizontal interpolation as appropriate.
As with any framestore based converter (Domino, Pineapple) the output will provide stable and continuous syncs regardless of input. This is kind to your vintage TVs.
The original 405 standard did not have equalising pulses. Although this should not be a problem with good receiver design the fact remains that many sets suffered from poor interlace. Production versions of the Aurora have a switch. You can turn EQ pulses on for best interlace or off for complete authenticity.
An unusual feature is a S-Video or YC input. If you have really high quality video sources, including monochrome ones, using this input will maximise bandwidth and minimise colour related artefacts.
The review sample had phono connectors for video. I would have preferred BNCs. Buyers are now offered the choice of phono or BNC.
The built-in test pattern generator can capture frames of still video to use as test patterns. Production units come supplied with test patterns including Test Card C. These are acquired via the analogue video input. I would have like to have seen some digitally derived test patterns pre-programmed in the flash memory but this is really me being a bit too fussy.
I loaned the review unit to David Boynes for further assessment. David’s achievements include the first 625 to 405 line converter to be designed and built by an amateur. He also has a fine collection of pre and post war receivers. I was with David when he displayed the converter output on a 23” receiver. He and I both thought the picture quality was superb, limited mainly by the quality of the vintage material that we viewed. We felt that very few viewers in the heyday of 405 would have seen such good pictures.
30 AND 32 LINE STANDARDS
Peter Smith has investigated performance on the NBTVA 32 line standard. Peter has built replica Baird Televisors and made extensive studies of mechanical television. He designed his own dedicated 30/32 line converter in 1989 using simple discrete logic. He used this for comparison in his tests.
For the tests Peter used Test Card C as a source and a Tektronix 602 XY display. This CRT based display makes it possible to assess the converter without any artefacts that might be caused by a mechanical display. TV engineers will recognise the vectorscope graticule which is within the CRT and irrelevant to this report.
Linearity and grey scale were excellent but resolution along the line was lacking. This was not unexpected as the Aurora specification states only 60 pixels on the 32 line standard. This is apparent in the photos showing results from the Aurora and Peter’s own converter. For a more realistic test, a 10kHz low pass filter in the output of each converter mimics the best bandwidth that was available to Baird in the 1930s. This reduces the difference between the two converters but the lack of resolution in the Aurora is still very apparent. At the time of writing, Darryl is planning updated firmware which will double the resolution on the mechanical standards. This will be available by the time you read this review. Existing units can easily be updated as noted below.
The only facility Peter found missing was the ability to pan a 3:7 aspect ratio 30 line picture over any slice of 625 line input. This is a very useful facility when converting material from 625 or fine tuning a camera shot when the camera is out of reach.
Unfortunately the unit could only be evaluated on the NBTVA 32 line standard as the firmware update for Baird 30 line and TeKaDe standards could not be fitted in time for these tests. I would expect performance on these standards to be very similar to 32 line.
I have not been able to review all the other standards, simply due to lack of appropriate receivers. If any are found to be incorrect it will not be a problem for Darryl to update the firmware and issue a new file which can be used to reprogram existing units.
Any framestore based converter will delay the video by a relatively large amount. Typically 20-40ms. If the accompanying sound is not delayed you can sometimes see lipsync errors. These are all too frequent on today’s TV broadcasts and are due to the large amount of framestore based equipment now used in TV stations. The Aurora has an audio delay facility that allows the sound to be retimed with the picture. I’m not convinced that this is a particularly useful facility but it did not add much cost to the design and you don’t have to use it. The audio output is potentially more useful for mechanical standards where it is used to send a synchronising signal.
The converter contains high speed digital circuitry which is a potent source of RF interference. The plastic case does not inspire confidence but the use of a multilayer PCB with continuous ground plane should help. I do not have facilities for proper EMC tests but I did not notice any obvious problems while testing.
POOR QUALITY INPUTS
While I cannot test the unit with all possible poor quality signals I can state that the input AGC copes well with low amplitude down to at least –6dB. Note that the input video and sync amplitudes must be in the correct 7:3 ratio since the AGC measures the sync amplitude. Slightly noisy off air and ordinary VHS replay are fine too. It is always possible that really bad VHS replay could cause tearing or other effects but I have not seen this happen.
SOME MINOR PROBLEMS
This is a very minor criticism and is of no real importance to users. In professional practice the black level of a video signal is at 0V with the sync tips at –300mV. This cannot be achieved without split +/- voltage supplies. The original video output circuit was a rather odd partially AC coupled affair. This has now been changed to a simple DC coupled output leaving sync tips at 0V and black level at +300mV.
Please note that the absolute DC offset of the signal will not cause the displayed picture to have incorrect black level because the signal will always be AC coupled and DC restored or clamped in a monitor or modulator.
Standards selection switch
The prototype had a rather awkward and flimsy rotary hex switch which had to be adjusted with a screwdriver. This has been replaced in production by a more robust part with a small knob. If you change standards frequently I am sure that this will be the first part to fail. Fortunately these hex switches are fairly standard and not too difficult to replace.
If Darryl designs any updates for the Aurora the resulting code can be programmed into existing units. All that’s needed is a PC and a simple adapter cable attached to a parallel port. I will be able to offer a programming service in the EU for no more than the cost of return postage.
There are some standards that could not be accommodated in the standard Xilinx code. These require special code for the particular standard concerned. Darryl has already demonstrated CBS sequential colour. 405 line NTSC may well be possible.
The Aurora has no modulator. This is a disadvantage for the average 405 enthusiast who will have to build or buy one. It would have been very difficult to include a modulator that would work on every standard that the Aurora can produce.
How do you choose between Domino and Aurora, both fine products at comparable prices. The Aurora is a beautifully engineered product and is difficult to fault. The main omission is a modulator. If you have any interest in vintage standards apart from 405 then you have just one choice.
If your sole interest is UK 405 line TV then the Domino converter provides a most effective solution that will do the whole job straight out of the box.
The Aurora is only available directly from its designer:
There are further pictures on the web site. The full Aurora manual can also be downloaded from the web site
Price $599 including delivery to the UK. Prices in other currencies will vary with exchange rates. You can pay by Paypal or credit card. There is also, in theory, a liability to import duty and VAT. My review unit escaped this problem though I have heard of others being subject to £20 and £60. There are no guarantees with UK Customs and Excise!