To: Electron User Magazine
TAPE LOAD RELIABILITY
I have noticed a significant number of letters in E.U. relating to DATA? errors on tape loading and I would like to share my experience of this problem.
As stated in your replies to the letters, tape head cleaning and azimuth adjustment are both very important, but a design error in the Electron was the cause of my problem.
When I bought the computer this summer (from a reputable department store) it was very cheap because it had been sent back due to tape loading problems. The cassette loading was unreliable and data errors were experienced every few blocks and the problem became worse as the computer warmed up. I knew that the computer was at fault because the problem was the same on all of the cassette decks I tried including my Nakamichi BX-2.
As a professional electronics engineer I could not resist opening up the computer to investigate the fault. The cassette input circuit worked perfectly, but I changes the op-amp just in case it made a difference but it didn't. My thoughts then turned to the ULA and then the cause of the problem struck me. Acorn were driving the ULA input with a 1.2 volts peak to peak square wave from the cassette input circuit. This signal level is determined by the clipping diodes Q1 and Q2. 1.2 volts is not enough to drive any digital integrated circuit reliably. Bipolar circuits usually have switching thresholds of 0.8 volts for low and 2.0 volts for a high level, which is a difference of 1.2 volts leaving no noise margin at all. Acorn have got away with this in the past because they have used a bipolar ULA, but my computer has an issue 6 circuit board with a CMOS ULA. CMOS input switching thresholds are usually at 30% and 70% of the supply voltage making a 2.0 volt difference. The 1.2 volt signal from the cassette circuit is not enough in this case.
To fix the fault I replaced the transistors used as diodes Q1 and Q2 with green Light Emitting Diodes. These give a 3.6 volt peak to peak signal to the ULA, enough to switch a bipolar or a CMOS circuit reliably, but not so much as to overload the ULA.
Since carrying out this modification I have never had a single data error in several months of use.
The BBC Micro has the same cassette input circuit and friends of mine who have these computers, and who teach using them at school, also experience occasional data errors from the cassette and believe that this is just an inherent problem with cassettes. Many other problems could cause these errors, but I suspect that Acorn's marginal signal level is responsible in most cases, giving many users the idea that cassette data recording is inherently unreliable.
I would recommend anyone who has data error problems to swop Q1 and Q2 for LEDs and see the difference it makes!