An Author Replies...
I have some more to add to my article \["Shed Some Pounds With This AC/DC Transformerless Power Supply," Nov. 22, 1999, p. 109\] in response to comments published in the April 17 issue.
- The use of a single-phase rectifier. I doubt that people will use it in their final application. The key feature of this article isn't the single-phase or full-bridge rectifier. It's about transformerless ac-dc conversion. I expect a less experienced user to start with a single-phase, and after the prototype works, use a full-bridge. My original manuscript included both rectifier circuits. But the two circuits are the same except for the ac-dc block. Because of page space limitations, we eliminated one and put in a paragraph on page 116, "Say you replace the half-bridge rectifier in .... Figure 3 with a full-bridge."
I didn't specify all components, such as the diode D1. In my prototype, the diode D1 is part of a full-bridge rectifier rated 400 V, 35 A continuous and 400 A surge, which is an overkill. But its single-piece price is $2.99 and it measures 1.13 by 1.13 by 0.3 in. Both its price and size are very small compared to this circuit's overall cost and size. As an application engineer, I answer customer questions every day and I only specified here those components that users have had difficulties with. I didn't get many inquiries regarding the selection of bridge diodes. Still, Electronic Design printed my name, phone, fax, and e-mail address in the article for readers to contact me.
- Use of 180-V clamp diode D8. I didn't address the tolerance of the 115 V ac. Instead, I gave an explanation on how to choose the right transzorb, D8, on page 114; "It had to be above the 163 V needed to operate the SA14, and equal to or below the SA14's rated voltage of 200 V." I also explained how the 163 V was derived from the 115 V ac with no tolerance considered. I respect your comment about 10% tolerance for the 115 V ac. I'm still concerned that the tolerance can be more than 10% in some countries and less than 10% in other installations. A user knows what his or her application is for. It's just equipment used in the house or a product sold all over the world. In that case, he or she needs to consider not just the tolerance, but also the nominal voltage of 115 V ac, 120 V ac, or both. Then, the user can specify the right transzorb. The 180-V D8 is just one example.
Also, the purpose of D8 isn't to absorb a continuous 20-A current with a continuous over-voltage. As stated on page 114, it's there to absorb energy from high-voltage spikes kicked back from inductive loads of the SA14.
- Output ripple and size of smoothing capacitor. The feedback control circuit was designed for maximum frequency bandwidth, as most users asked for that. I used larger smooth capacitors to reduce line-frequency ripple. That's one way, but maybe not the best. Would you kindly share your expertise? I appreciate your constructive criticism.
Yu Jen Wong
Apex Microtechnology Corp.
...And So Does A Letter Writer
Thank you for allowing me to make additional comments on Mr. Wong's article.
- The problem with the full-bridge, as you said in the article, is the output is no longer ground referenced. So, you must have a floating load and reference signal generator. As a "transformerless power supply," this may be a problem.
In view of the poor power factor at maximum rms line current, you could not get 20 A in the load at full output voltage, for example 160 V into 8 Ω. But you would still need a big rectifier.
- You cannot select a standard 5% part which fits between the 200-V rating of the SA14-chip and the maximum peak line voltage.
In the event of high line voltage, there's only the 20-A fuse to limit the current into D8.
- Basic closed-loop control theory tells you that the output disturbance, due to your dc-bus ripple, should be attenuated by the loop gain of your control loop—that's if the phase delay at the frequency of interest isn't too significant. As the phase delay becomes more significant, you will get less disturbance rejection for the same loop gain. If the phase ever gets around to 180° with G>1, you have an oscillator. Because of your PWM frequency, I expected you to select an output filter and loop gain that would provide you with enough gain at 50/60 Hz for low line-frequency ripple at the output. Offline smps have good line-frequency rejection. You don't see ones advertised with 10% ripple on the 5-V rail.
Power Design Manager
Control Techniques, U.K.
A Trip Down Memory Lane
I enjoyed your article on the DuMont 425 O-scope \["Digital Readout Scope," Feb. 7, p. 66\]. I owned one in the '70s, and can tell you it was a piece of work! I bought it cheap because it didn't work. Not surprisingly, the thing had, if my memory serves me correctly, 104 tubes. The delay line alone had six twin triodes. The machine was probably 50% bigger than a contemporary Tektronix, drew 1800 W and had a 1-GHz sampling unit. It must have weighed 120 pounds. It had two handles for a reason!
Fixing it involved replacing about 20 tubes, mostly 6DJ8, and removing a solid-state power supply with which some bozo had "improved" the beast.
Probably the most interesting before-its-time feature was the "computer interface." On the scope were two huge connectors comprising hundreds of pins, which primarily connected the digital readout switches to a computer. That was '60s data logging!
Director of Engineering