RESONANCE EFFECTS FOR EVERYONE TO SHARE

[gotoluc]

Dear Luc,

How long will it take you to build the crystal oscillator?

Cheers
Thane

Hi Thane,

thanks for dropping in

I think that idea I had will not work as I can now see that the frequency would need to be dynamic (change as loads changes)

I think I'll just leave it to the talented in the art of electronic to figure it out

Luc

[b0rg13]

Hi everyone,

Good news

Test 9: http://www.youtube.com/watch?v=5TcKmArOXsw

Luc

hi Gotoluc, thats a great vid, thanks for sharing.,and merry xmas to you and all, peace.

[TinselKoala]

Hi TinselKoala,

thanks for the positive comment ... after doing the video test 9 I did hook up a DC ma meter (just for fun) to see what the ma drain was and it was 24ma with no loaded coils and 21ma with the loaded coils.

As I have said before and Groundloop also agrees!... I do not think a digital or even worse an analogue meter can give an accurate reading when the DC is switched on and off over 350,000 times a second but if you think it is fine then the numbers above are the readings.

As you also know, (being first to replicate Groundloops circuit) the switch side of the circuit consumes about that many ma with no load attached from a 45vdc input voltage. Have you also noticed that if you raise the input voltage, still with no load, it keeps going up in draw of amps. When mine gets to the 100vdc range the IR2103's are too hot to keep my finger on them . Even at 60vdc mine get to 130 degrease Fahrenheit. However I have not yet changed the resistors and added the capacitors that Groundloop has recommended in order to help with the heat issue. Hopefully these changes will help with the energy waste?

I'll report the changes once this has been done.

Luc

You are right--before I only looked at the voltage ripple at the bridge input, and it wasn't too bad. But just now I hooked up a current-viewing resistor, in series with a 22.8-volt LiPo battery at full charge, running the logic side off a separate 11.4-v LiPo.
The scope shows a pretty pronounced ripple on the input current to the bridge. So I agree, the DMMs will probably be confused. But that's not the case for the Simpson, usually. At those ripple freqs, with milliamps as opposed to microamps, the Simpson analog meter is actually pretty accurate at averaging things out. You pay for this in response time, sensitivity, and overall accuracy. But a lot of folks still prefer good analog meters like the Simpson for this kind of work.

I have not yet gone to higher input voltages. But I did purchase a nice toroidal power transformer yesterday that has 39-0-39 secondary, so I'll be there soon.

I noticed for the first time some heating in my components, running the 24 volts into a dead short ( the TBC primary) for about half an hour. Nothing severe, but I suppose I should change the diodes, since I now have 15 of the fancy motorola units.
My SOIC IR 2104 chips haven't gotten noticably warm yet. But the mosfets, the diodes, and the high-current wiring do get warm.

I think the heating at no load comes from "shoot through" where the mosfets aren't exactly synchronized at the freqs we're working with. A couple years ago I designed and built a more complex H-bridge around an Intersil single-chip controller that had fine tuning functions so you could really get the Hi and Lo side MOSFETS clipping along in lockstep. It worked pretty good, but I had to pack it away to move across country, and I still haven't unearthed it. That's why I was glad to see this simpler circuit.
Anyway, if the switching times can be synchronized better, there will be less power wasted in the MOSFETS. This involves, in our circuits, good lead placement, even lead lengths, matched decoupling capacitors, and stuff like that there. It's pretty hard to do without an optimized printed circuit board, even at 350 kilohertz.

[wattsup]

@gotoluc

I am still somewhat confused but when was that news right....

I am thinking that you should maybe take both outputs from the coils, rectify it and connect to a good sized dead battery to see if it will charge and how fast. Right now, with those two resistor loads, it is hard to see the actual energy developed or to appreciate the effect on the output end. If it is a dead battery, maybe it should not interfere with the resonance until it is getting well charged up.

[Groundloop]

@TinselKoala,

One way to reduce the wiring heat is to use a 100uF (voltage must be higher than the input voltage) across
the plus and minus of the high side of the switch input power. The capacitor will average the current and reduce
the heat in the wires. The hexfet transistor do get hot at high input voltages when switching current. This is normal for all hexfets. Use a heat sink. Now the drivers, they will also get a little hot when switching at high
frequencies. This is also normal for  the IR2103. The driver IC's must charge the gates of the hexfets and then
discharge the same gates at high frequencies.

It is a surprise to me that the switch is using power with no load at the output.
The IR2103 has a built in timing to prevent that the hexfets is on at the same time.
But as you say, when running at high frequency then the design starts to be important.
But for a simple switch as this I think it is doing a good job. :-)

Groundloop.