| ??? 12/02/03 18:37 Read: times |
#59803 - No tricks! Responding to: ???'s previous message |
Damjan wrote:
Another idea is to put 100pF across buil-in transistor of MC34063 and across the coil. Do you thik of any better idea? and: I tried to add 1N5819 in series with inductor. Hallo Damjan, first, don't try to introduce any kind of unproved tricks!! This can result in total desaster!!! Ripple is typical for all switchers, and the try to avoid them by additional diodes will NOT work. The only way to reduce ripple is a. by the use of additional ripple filter at output b. or by the use of a design with different topology. But not one achieved by any tricks, but by proper design! Ready DC/DC converters often use very appropriate topology to achieve best performance. E.g. very high switching frequency, very small dimensions, partially or fully shielding, use of state-of-the art MOS-transistors and -very important- very special inductivities with very special ferrite material. Many companies produce them, e.g. 'Traco power', 'Newport components', etc. I assume, it's impossible to top these ready made DC/DC converters by a discrete design. But this does NOT mean, that it is impossible to build a good one with discretes like MC34063. At page 7 of this document you will find a demo-PCB: http://www.promax.dk/download/...063-ST.pdf Well, that's not a brilliant layout, but better than nothing. It's important to use the concept of star point grounding. This means, all parts of your application must have their ground potential connected to this star ground point. In the demo-PCB of above link you will find this star ground point directly at GND pins of connector (on the right of PCB). To this point every ground must be connected to: Ground of input voltage, ground of 5V regulator, ground of optional ripple filter and any shielding. If star point grounding is omitted, you can get big problems: Introduce of drastical ground noise, emitting of interference, increased susceptibility to interference etc. At figure 9 of: http://www.smdicjmf.com/utc/icpdf3/MC34063.pdf you will find information about optional ripple filter. I would recommend you to use 100µ/16V aluminium electrolytic in combination with 'BL02RN2' ferrite bead from Murata, or similar. Unavoidable equivalent series resistance of electrolytic capacitor will prevent ringing. Corner frequency is about 1 / 2 / pi / SQRT (2.2µH x 100µF) = 10.7kHz. So, this ripple filter makes only sense, if switching frequency is much higher than 10.7kHz!! I have a bad feeling since I saw this ripple on the OUTPUT of 5volt regulator. Well, this looks serious!! Normally, there should not be any relevant ripple at output of voltage regulator. If there IS high ripple, then there IS a mistake! Low voltage drop regulators are not easy to handle. Reason is, they are intrinsic unstable. Only in combination with suited decoupling capacitor at output they become stable!!!!! This behaviour is totally different from that of classic linear voltage regulators. If you don't use suited decoupling capacitor at output you will get problems with instability! This might be the reason, why you see tremendous ripple at output of low drop voltage regulator. Unfortunately, I didn't find datasheet of your regulator (4940V5). But if you have a look at datasheet of LM2940, which you willl find here: http://cache.national.com/ds/LM/LM2940.pdf then you will see from figure 1 and according text, that output decoupling capacitor must fullfill some very rigid requirements: Equivalent series resistance (ESR) must be in a very certain range, and this for all temperatures of application! The use of 22µF tantal electrolytic at output is a must. Keep in mind, that ceramics capacitors can show too low ESR for this application. Do only use this 22µF/10V tantal. You can also see from figure 1, that just the case of low output currents can be troublesome. This is an often observed detail, that low drop voltage regulators show instability at very low output currents, which disappears, when load current rises!! But this will only be observed, when having poor decoupling at output. With adequate decoupling they will work very well. After talking so much about decoupling capacitor at output, of course, don't forget the one at the input! By the way, if input voltage of 5V regulator is 7.5V, then you could also use standard 5V regulator, like LM7805. Of course, it depends on load current. Have a look into according datasheet. Good luck, Kai |
| Topic | Author | Date |
| powering 8052 with MC34063 | 01/01/70 00:00 | |
| RE: powering 8052 with MC34063 | 01/01/70 00:00 | |
| RE: powering 8052 with MC34063 | 01/01/70 00:00 | |
| RE: powering 8052 with MC34063 | 01/01/70 00:00 | |
| RE: powering 8052 with MC34063 | 01/01/70 00:00 | |
| More details? | 01/01/70 00:00 | |
| RE: More details? | 01/01/70 00:00 | |
| RE: More details? | 01/01/70 00:00 | |
| RE: More details? | 01/01/70 00:00 | |
| RE: More details? | 01/01/70 00:00 | |
| No tricks! | 01/01/70 00:00 | |
RE: No tricks! | 01/01/70 00:00 |



