|
Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets. |
|
Thread Tools |
24th Aug 2017, 8:13 am | #41 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Simple inverter cct.: comments, please
As for RFI the one I built was terrible. There was ringing around 1MHz with a peak of 224v. That's a pretty bad transmitter.
|
24th Aug 2017, 8:56 am | #42 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,803
|
Re: Simple inverter cct.: comments, please
There is a lot of very sensitive measuring equipment out there in the world with switch-mode power supplies in it - even the Agilent Noise Figure Analyser which is used to measure the noise created by ultra-low noise semiconductors... well below the thermal noise floor of room temperature. These things all use dramatic amounts of screening and filtering. There is an Icom IC7700 HF transceiver by my right elbow. It covers 30kHz to 60kHz without any gaps, so there is nowhere to hide its SMPS witching frequency, and that's a 500W switcher. Again it uses screened boxes for the PSU and many screened boxes for the RX, all separated with lots of filtering.
So it can be done, but that level of screening costs money only tolerable in expensive gear. As Argus says, there are other ways, but they are expensive in different ways, inevitably in efficiency. Efficient switches are either on (and dropping little voltage) or off (and leaking little current) They can't afford to hang about in the middle where power dissipation is high. Fast switching means lots of harmonic content. You don't want fast switching of either voltage or of current (everyone forgets the current!) An example of the low initial noise way would be an audio power amp driving a transformer and then some rectifiers. The ony quick switching in there is the current in the rectifiers. Some capacitors on the diodes is usual, but there is one very counter intuitive thing. Old slow rectifier diodes like the 1N400x series make a lot more noise than super fast ones. Theoretically the fast diodes ought to give more harmonic energy from their faster waveform, and they do, but this is overshadowed by the slow diodes being so slow they carry reverse current then turning off with a snap. A very harmonic laden snap. This problem exists with 50Hz mains supplies as well. Obviously a linear audio amplifier is going to give <<50% efficiency. So switching circuits with a resonator can be used, but watch out for fast-edged pulses of current in the switches, these generate harmonics just like fast voltage edges do, however, the harmonics of fast current changes are mostly magnetic fields and magnetic fields are a lot harder to screen than electric fields. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
24th Aug 2017, 10:08 am | #43 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Simple inverter cct.: comments, please
Indeed. The scope I bought has a switching power supply in it.
Not sure if this has been mentioned or not but the legendary Jim Williams wrote up Linear AN70 which covers very low noise switching regulators: http://cds.linear.com/docs/en/application-note/an70.pdf |
24th Aug 2017, 1:10 pm | #44 | |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Quote:
Attached are a few photos of a screening box made out of solid aluminium. This is for an inverter supplying an Ever Ready Model K radio - it sits within the frame aerial - and there is still a very slight whistle on very weak stations. RFI is definitely radiated, not conducted, as removing the inverter box from the frame aerial makes it quiet. (Incidentally, the inverter has 3 outputs: 90V HT, 1.35V LT, and variable grid bias). Board in the box: And the interior of the (lipped) lid. |
|
24th Aug 2017, 1:48 pm | #45 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Simple inverter cct.: comments, please
|
24th Aug 2017, 1:57 pm | #46 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Simple inverter cct.: comments, please
I had similar problems with a log detector I built. Turned out that the issue was the RFI leaking through a hole. I drilled it out and put a feed through capacitor in the hole and all was good.
My RF experiments in future will be: PCB box, SMA connectors, feed through caps for power/control signals. I assume your enclosure above is connected to ground somewhere? |
24th Aug 2017, 5:04 pm | #47 | ||
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Re: Simple inverter cct.: comments, please
Quote:
Quote:
Incidentally, reading Argus25's own efforts in the link he provides, for comparision my switching frequency is 20kHz; circuit topology is a flyback converter with complete energy transfer (boundary mode operation). So the rectifier diodes turn off very gracefully with current declining smoothly to zero; the switching transistor turns on also very benignly with current rising from zero smoothly. The sharp transition is the transistor turning OFF at maximum current, although even then the self-capacitance of the transformer allows collector voltage to rise at its own rate till the output diodes turn on. I did add RC snubbers but found they hurt efficiency with very little benefit! Output has voltage regulation by varying transistor 'on' time. Efficiency is just over 80%. |
||
24th Aug 2017, 6:52 pm | #48 |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Simple inverter cct.: comments, please
Kalee 20: I see wires emanating from a metal enclosure. Do those wires have adequate RF by-passing to common ground? If not, that's probably where the problem lies. Plus, an RF gasket on the lid of the box to the flange of the box case may also be required.
I once worked for a company that specialized in manufacturing, supplying and servicing Tempest-approved digital hardware to certain users. Generally, there were two ways of meeting that spec.: containment and source-suppression. My above remark comes into the former category, but some additional work in the latter one might also help. Al. Last edited by Skywave; 24th Aug 2017 at 6:58 pm. Reason: Add para. two. |
24th Aug 2017, 8:04 pm | #49 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,803
|
Re: Simple inverter cct.: comments, please
This looks like a job for feedthroughcapacitorman!
But they're getting rather hard to find nowadays. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
24th Aug 2017, 10:23 pm | #50 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Simple inverter cct.: comments, please
Without wishing to violate any rules, you can get a respectable amount of them from eBay seller gw0egh. 25 for £7.99 including delivery. Rated 1nF 50v.
Rapid sell them still as do digikey but they are crazy expensive! |
25th Aug 2017, 12:08 am | #51 |
Dekatron
Join Date: Nov 2011
Location: Surrey, UK.
Posts: 4,385
|
Re: Simple inverter cct.: comments, please
There are also ex-Soviet stock available from a few auction site sources in e.g. 15nF 250V and 4n7 400V- at the moment, some are almost astonishingly inexpensive in small quantities but may not be around for ever. They are relatively large though, ISTR the 15nf ones I used were M6 X 0.75 fitting, the 4n7 M8 x 1.0.
|
25th Aug 2017, 8:48 am | #52 | |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Re: Simple inverter cct.: comments, please
Quote:
The wires are RF bypassed on the board itself to 0V, with surface-mount capacitors close to the on-board 'D' connector. The leads coming out just come direct from the mating connector. I agree that feed-through capacitors here would be a good idea, though I was hoping that it would not be necessary. Like I say, it's only with the caboodle sitting in the middle of the frame aerial, that there is any hint that there's switching going on! I do have ideas to reduce the susceptibility of the radio by replacing the frame aerial with a ferrite-rod aerial that the converter obviously doesn't sit inside, but not got round to trying that yet. |
|
25th Aug 2017, 11:46 am | #53 | ||
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Simple inverter cct.: comments, please
Quote:
Quote:
Al. |
||
25th Aug 2017, 1:53 pm | #54 |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,061
|
Re: Simple inverter cct.: comments, please
Well I can't argue with that! Something I must bear in mind for the future... Thanks Al!
|
25th Aug 2017, 2:02 pm | #55 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Simple inverter cct.: comments, please
One radio I have is a baby Beethoven. The frame antenna is built into the cabinet too. Also there is an interesting quirk; there is an inter-stage audio transformer (iron cored) with a large number of secondary turns, feeding an audio valve's grid. Of all the battery valve radios I have, this one has been the bigger challenge to operate with any switching supply. Even with the RFI above 150 kHz eliminated, the transformer acts as a pure magnetic field sensor (analogous to the tele-coils in hearing aids) and I have not yet successfully been able to completely shield any type of switching supply, though I have not tried a very thick walled enclosure, but mu-metal failed.
|
25th Aug 2017, 3:45 pm | #56 | |
Rest in Peace
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
|
Re: Simple inverter cct.: comments, please
Quote:
Al. |
|
25th Aug 2017, 10:06 pm | #57 |
No Longer a Member
Join Date: Oct 2016
Location: Maroochydore, Queensland, Australia.
Posts: 2,679
|
Re: Simple inverter cct.: comments, please
I can also report another thing I tried, that didn't work. Since the step up ratio of the supply needed only needed to be 8 times and the current was low, I came up with the notion of doing it without any transformers or inductors, so as to have minimal radiated magnetic fields. So I simply made an oscillator (I think it was about 1kHz) driving a complimentary pair of TO-220 mosfets and fed that into a voltage multiplier. The unit had excessive interference output and was no better than any I had made with transformers! The wiring itself was radiating. I tried to suppress it with series resistors, but it became too lossy with poor regulation. It was after that I went to the idea of a low fundamental frequency with a slowed rise time so as to not need any shielding.
|
27th Aug 2017, 11:33 am | #58 |
Dekatron
Join Date: Jun 2016
Location: London, UK.
Posts: 3,687
|
Re: Simple inverter cct.: comments, please
|