| TroelsM |
Hi.
Here's a small peak at my latest classD-fumblings.
I concentrated on the output-stage at first. It's build around a IR2110 and two FQB19n20L.
Here's a few pics of the output-waveforms before the filter. First a 115Vpp square and the rising/faling edge. Rise/fall-time is approx. 35nS (10%->90% of voltage-range). Looks promising.
Next on the to-do is the modulator-scheme. Most likely selvf-osc in some form. Any ideas? What's your favorite-scheme of the day?
Please click this link to see more info: http://kofoedmejer.squarespace.com/classd-amplifier/
Regards TroelsM |
|
|
| luka |
Hi
I have tried clocked version, works great. Ucd's are also great...really hard to say what to go for... ucd is more simple if you make it all discrete
Hope you will have bigger cap for upper gate drive, when you will go into hard clipping
Are those pics befere or after LC filter? |
|
|
| TroelsM |
Waveforms are taken before filter. Just edited my prev. post to reflect this.
Regards |
|
|
| luka |
Hi
Bit you to it :D Before LC you won't have really any problems, you should connect LC filter too, and see what you get there...
Do you have any rail-rail decoupling caps on board? |
|
|
| TroelsM |
Hi.
Each rail has a 330uF/100V e-cap and a smaller film-cap. There is no caps from rail to rail. Why? |
|
|
| lumanauw |
| Is there any good clocked classD schematic with feedback taken after LC filter? |
|
|
| luka |
Hi
@ TroelsM
You will need noise free rails as much as possible, this is why you should have rail-rail cap too, 2x100n or 1x220-330nF, try to find some good ones
@lumanauw
Yes there is as far as you ask me, made by fredos and then redesigned by alexclaire |
|
|
| TroelsM |
@Luka
Output filter was connected at the time, but the plot shows the pre-filter waveform.
The cap on the floating supply is not very big, but I will change it later. Right now its not the biggest problem.
Thanks for the help. Re-visit the tread later to see more.
Regards |
|
|
| TroelsM |
Doh!
Regards TroelsM |
|
|
| luka |
Hi
What are you trying to say?
It is better then my... |
|
|
| TroelsM |
I'm trying NOT to say that i destroyed the lowers-side Mosfet and IR2110 driver when GND at my scope-probe for a brief moment touched something (lowerside-gate perhaps?).
That's a bit embarrassing... Oh well, - it's probably not the last time I'll do that :-)
The amp is re-build with better cooling and working again, so after all it was only a minor setback.
Regards TroelsM |
|
|
| Eva |
Don't worry, this has happened to all us at least once or twice.
Once I blew the ground lead of a probe when I inadvertedly did a short between +415V DC and GND through the other probe in a board having a 2250uF capacitor bank. Later the alligator clip of the ground lead of another probe ended up perfectly bonded to the leg of a 2W resistor, it seems that temperature due to the short circuit was high enough to melt both metals :D |
|
|
| TroelsM |
Hi Eva
"once or twice".. ? I've done that more times than I want to admit :-)
Well now I would like to measure the currents in the output-stage and therefore I'll try to construct at simple current-probe from a two-part ferrite-core, some wire, a few passive components and a bit of coax with a bnc-plug.
Any ideas? I won't go for high accuracy, but just something that will let me see the currents. I'm guessing that linearity is more important than absolute accuracy. For switch-mode it would be nice to be able to measure from 100Khz to 100Mhz.
Regards TroelsM |
|
|
|
