Switching power supply problem for cheap LED meter?

Was that 200-500 counts out of 2000 or so, or do you mean 200-500VAC on the reading of the 450VAC? A lot of the DC-DC converters are minimalistic junk producing stray high frequency magnetic fields all over the place. Because of the high frequency, it does not take much of a field to corrupt nearby circuits. Doing things like twisting leads and experimenting with relative orientations of the components gives away a through the air problem without actually making measurements. Then again with high frequency noise, all the specs for power supply and common mode noise rejection are out the window too.

That supports a problem I had with a system last year. It was a +/- 15V analogue measuring system, (linear supplies), looking at things in a system powered from a 48V switcher (from a reputable mfr).

At switchon the analogue readings were all over the place, and I subsequently discovered that the *whole* of the +/- 15V circuitry had huge 40KHz spikes in it. It turned out that that reputable? 48V switcher was splurging 40KHz common mode spikes everywhere.

A toroidal common mode choke and capacitors to the 0V of the +/- 15V sent the spikes away.

On 4/27/2007 5:45 AM, The digits of Paul E. Schoen's hands composed the following:

Others posted some possible scenarios, but I also want to say do you have a ground potential issues? Is your 9V switcher isolated? Have you also looked for load stability? You say it is a 10W switcher. Some cheap switchers need to be in continuous mode (not DisconTin) mode for good regulation. At light load it may not be that way.

My 2 cents Hawker

On Fri, 27 Apr 2007 05:45:43 -0400, "Paul E. Schoen" wrote: ...

... Most (el cheapo) 199.9 mVDC meters need 9V independend power. Don't mess about, just provide it.

A simple circuit idea to create such floating power can be found here (bottom of the page)

Cheers,

Joop

Thanks for the info. That looks like a "flying capacitor" isolation supply, which may work well for very low current LCD meters, but maybe not for my LED type. Also, I still need to obtain my supply from a 120 VAC source.

I think the problem must be common mode noise that is perhaps capacitively coupled from the switching supply high voltage section through the transformer and to the output. I'm not going to try common mode chokes or other cures, because the linear supply works like a champ and is the lowest cost, simplest solution, and will not suffer from obsolescence of the switching supply.

Looking at the supply, the transformer seems to be wound with layers over a common EI core. This probably makes for better coupling and higher efficiency than coils wound side by side or on opposite ends of a toroid, but provides a direct capacitive path for the 180 volt peak transients to couple to the output and wreak havoc with the meter. I think it may actually be causing glitches in the meter's digital circuitry, causing incomplete conversions, and random displays varying several hundred counts.

Paul

The dual slope conversion obtains it's immunity to 60 or 50 Hz linear interference through selection of components syncronizing the conversion rate .

Hf noise generated by the switcher is not compensated and can affect the early terination of either conversion slope period unpredictably. You could probably get an indication of what you're dealing with on a scope and judge the effectiveness of any orientation change, added filtering or grounding/screening hardware.

RL

On Fri, 27 Apr 2007 05:45:43 -0400, "Paul E. Schoen" wrote:

I have read all the posts about your problem. All the responders seem to sugest exotic dc-dc conversion method, dual slope yada yada. I don't get the feeling that you understand dc-dc converter operation. Seems to me your looking for a BUCK type converter. One that throttles the output down to a lower level etc. 12v to 9v, 120V to 12V. But what I don't seem to hear about is your converter inductor current. This you need to set limits on so that it doesn't saturate the field, (loss of magnetic flux density). Also your period time is determined by some RC time constant. Anyway I have built a few converters in my day and never experenced your problem. The output remained at a dc level despite the load because of the output capacitor. The only problem of high loads is that it may exceded the supply wattage and output deteriates. You can easily make a good converter from Linfinity SGx842/43 pwm chips. The peak coil current should be at 1.5x your load current. Most PWM converter topology use FET to sink current through the coil. Just feedback the peak voltage across the sense resistor in series with the coil. Converters usually have a current sense line which stops the converter when voltage reaches 1V. Also there is a voltage feedback line that controls the output voltage. The voltage feedback maybe more than 1V, the ones I'm familar with stop the converter when 3V is present. Usually this is from the mid-point of a voltage divider. So output voltage is determined by the divider ratio. In any case you should not see the bouncing you described.

Yag1064