Can Someone Please Switch Off The Internet?
While doing some research on the 50-year-old famous voltage regulator IC µA723 I made an unexpected discovery.
Surfing the Internet often results in unexpected discoveries. The one I made today came up while doing some research on the 50-year-old famous voltage regulator IC µA723 developed at Fairchild Linear Integrated Circuits. A well-known search engine gave me a link to a page describing good design practices for the µA723 with at the end a link to a video about a “famous but misdesigned Elektor circuit”. This, as you will understand, attracted my attention.
This chip designed at the end of the sixties of the previous century when the Beatles released the White Album (unnecessary emphasis to make it sound really old) had specifications similar or better than many modern-day linear voltage regulators. The Texas Instruments catalog still lists the part in the metal-can (TO-100) version (other versions are available too). Although claimed to be in active production, one can’t help wondering if this is new-old stock (NOS) found while cleaning out a forgotten storage room after the acquisition of National that had taken over Fairchild in 1987 or so.
Doublechecking my sources, I did an online search for the 1978 article. Thanks to those people who scan & upload everything ever printed, it is easy to find. Google being Google, I found the Dutch version of the article first. To my surprise, it doesn’t talk about a laboratory supply and makes no claims of excellency; it even mentions a weakness in terms of a ripple voltage of 10 mV (nothing about temperature stability, though).
Can someone please switch off the Internet?
The designer of the µA723 did a pretty decent job...
The circuit in question was published in the Elektor July/August 1978 edition; the video about the article appeared more than 40 years later in May 2019. Entitled Fawlty Powers :-), it is quite interesting and shows some of the impressive capabilities of the µA723: 40 V input range, line and load regulation of 0.02% and 0.03% respectively, and a temperature coefficient of only 0.002 %/°C (i.e. 20 ppm/°C).This chip designed at the end of the sixties of the previous century when the Beatles released the White Album (unnecessary emphasis to make it sound really old) had specifications similar or better than many modern-day linear voltage regulators. The Texas Instruments catalog still lists the part in the metal-can (TO-100) version (other versions are available too). Although claimed to be in active production, one can’t help wondering if this is new-old stock (NOS) found while cleaning out a forgotten storage room after the acquisition of National that had taken over Fairchild in 1987 or so.
... but Elektor messed up.
The Elektor article describes a trick to make the µA723 produce a variable voltage all the way down to 0 V, something it can’t do normally. The video explains in detail what is wrong with the Elektor design and concludes that the presented laboratory supply does not have the excellent line and load regulation as claimed in the article.Doublechecking my sources, I did an online search for the 1978 article. Thanks to those people who scan & upload everything ever printed, it is easy to find. Google being Google, I found the Dutch version of the article first. To my surprise, it doesn’t talk about a laboratory supply and makes no claims of excellency; it even mentions a weakness in terms of a ripple voltage of 10 mV (nothing about temperature stability, though).
Mut(il)ated in translation
Knowing how things work internally at Elektor, chances are pretty high the article was written in Dutch before being translated to English. Things are not only lost in translation; sometimes whole paragraphs appear out of nothing. That seems to have been the case here as well. The English text contains about 50% of the Dutch text, the beginning and end having been replaced by someone. Probably to make the augmented translation fit in the space reserved for the article, an important detail was left out in the process: the adjustment procedure for P1 requires a temporary one-watt load resistor of 1 kΩ to be connected to the power supply output.Incredible
Fifty years after the introduction of an IC, I find myself publishing a correction for a (bad) translation of a 40-year-old article about said IC because someone scanned it and put it online where it got picked up and was discussed in forums, encouraging yet another person to spend several hours on recording and editing a 20-minute video basically pointing out that the article used the word “excellent” where it shouldn’t have. And now you know about it too (sorry for that).Can someone please switch off the Internet?
