HI BOB, Thanks for the articles on output impedance. They reminded me of the time back in the early 1970s when I was working as a part-time engineering technician (while going to school) for a company that made professional audio tape recorders.
I had been working part-time in production test until one fateful day when there was some sort of upheaval in the engineering department. The VP of engineering and senior engineer both quit. The junior engineer was promoted to senior engineer, the engineering technician was promoted to junior engineer, and I became the new engineering technician.
For the most part, my job was documentation, etc. However, one day I noticed some sort of commotion regarding a new machine that was soon to be released into production. Apparently, a decision was made that the maximum audio output level previously specified at +16 dBm was not sufficient and the spec was changed to +20 dBm. The new senior and junior engineer were working on the problem and I had plenty of work to do with my ammonia-belching blueprint machine, so I didn’t get involved.
I was off for a couple of days and returned to find that they were still trying to solve the output level issue. The two guys really looked the worse for wear. I offered to help. (My sinuses were so burnt from the ammonia, I could see the problem with an open mind.)
Here is what I found: a 24-V power supply powering all op amps; op amps biased at 12 V; a 741-type op amp used as an output stage, ac-coupled driving an audio transformer wired for 600 O: 600 O; an HP voltmeter/distortion analyzer connected across the secondary of the transformer; and a 600-O termination at the input of the normally high-input impedance of the HP voltmeter.
I ran the circuit and verified the +16-dBm output. I then asked what they had tried to get the additional 4 db. They told me that the output transformer could be configured for 150-O: 600-O operation and that should give them 6 dB more output level, and that is what they have been working on.
So I rewired the transformer for 150:600 and tested the circuit. I confirmed what they saw. They couldn’t even get the +16 dBm with this configuration. I asked what they else they had done. They had spent two days trying 20 pieces of the same transformer, convinced that they were defective!
I grabbed a piece of perfboard, a complementary pair of transistors, a couple of diodes, and a few resistors and built a class AB output stage. I broke the feedback around the 741 and placed it around my output stage. I didn’t notice that while I was working on the solution, the president of the company had been taking notice. Within 30 minutes, not only had I solved the problem, but I was able to divorce myself from the blueprint machine.
Funny—and sad. I’m glad someone realized that a 741 can’t drive ±37 mA into a 150-O load. And that a transformer can’t put out more mW than you put into it!
I have seen a resistor as high as 30k in the negative feedback of a unity-gain op-amp follower. Do you have any insight as to why it might be there? (It is often a good idea to put a resistor there to offset a similar R that is the impedance of the signal source at the + input. Many op amps have an IB+ that’s about the same as the IB–, so the I × R’s tend to cancel. However, it is still a good idea to put a CF across that 30k = RF, such as 3 or 30 or preferably 300 or 1000 or 10,000 pf. The right value to put in requires a little engineering... /rap) I cannot find any articles to suggest this is a good thing. (Oh, yeah, we have done this for many years. I could find you two if I had to. See AN-30, etc., from over 30 years ago. Also AN-3, AN-4, AN-20, and several others. /rap) I believe it may cause the op amp to become unstable. Any thoughts would be greatly appreciated. I am stumped.
The answer is simple, as above. Sometimes even 300k or 3M of RF may be a good idea, so the total impedance looking out the + and – inputs is the same. But adding some pf across the RF is almost always a good idea. It is possible that in some layouts, the printed-circuit board wiring strays add 2 or 3 pf, which is enough. But you ought to engineer this, not just wish for this.
What is a Trarlington (“What’s All This Current-Source Stuff, Anyhow?” June 12, 2008, p. 60)?
For pity’s sake! Just because you can’t look it up on Google, it doesn’t mean it doesn’t exist! A Trarlington is a sort of triple Darlington, a cascade of three transistors.
Comments invited! [email protected] —or: Mail Stop D2597A, National Semiconductor P.O. Box 58090, Santa Clara, CA 95052-8090