Basically I am a fan of 3D-P2P constructions. But my handmade schematics and handmade routing plans were somehow annoying. And – frankly speaking- a relict of the last century. OK, even if learning is hard in my age... No way around starting my first babysteps in CAD. I decided to use the cost free Eagle Light. And my current class D project was exactly right for that, so I went for a 2 layer PCB with ground plane structure.
Schematic first. It is an self resonant hysteresis design. A fast comparator (LM160) with inverting/noninverting outputs to generate also the signals for a full bridge. Each halfbridge is using a IRS 20954 driver and two IRFB52N15D MOSFETs.
The missing connection in the schematic between upper source and lower drain will be a wire, because I need to do current measurements in this wire for optimized dead time adjustment. Each PCB is just one half bridge with comparator, but without output filter. For a full bridge the comparator is only needed on one PCB and delivering the signals for both half bridges. So far my theoretical plans... Reality will show....
Schematic first. It is an self resonant hysteresis design. A fast comparator (LM160) with inverting/noninverting outputs to generate also the signals for a full bridge. Each halfbridge is using a IRS 20954 driver and two IRFB52N15D MOSFETs.
The missing connection in the schematic between upper source and lower drain will be a wire, because I need to do current measurements in this wire for optimized dead time adjustment. Each PCB is just one half bridge with comparator, but without output filter. For a full bridge the comparator is only needed on one PCB and delivering the signals for both half bridges. So far my theoretical plans... Reality will show....
Attachments
Today I received the PCB samples and was quite pleased about the quality.
I got samples with 35 micrometers copper thickness and with 70 micrometers.
Somehow I feel more comfortable with thicker copper especially for my 1kW full bridge, with max. currents around 50A...
Component side:
I got samples with 35 micrometers copper thickness and with 70 micrometers.
Somehow I feel more comfortable with thicker copper especially for my 1kW full bridge, with max. currents around 50A...
Component side:
Attachments
Uhm, this is a promising project... Go on with it. I wish I could have double-sided hole-plated PCBs with solder mask and silkscreen as yours delivered quick and cheap here 
Currently I have to print, expose, reveal, etch and drill my own double sided boards 
(no plating 
)
BTW: My new 2500W low-frequency class D amplifier powered with 416V directly from the output of a PFC is working and almost ready. I may post some pictures soon for you to see. I have been really scared sometimes during the development because the capacitor bank at the output of the PFC stores around 200 joules
Currently I have to print, expose, reveal, etch and drill my own double sided boards (no plating )BTW: My new 2500W low-frequency class D amplifier powered with 416V directly from the output of a PFC is working and almost ready. I may post some pictures soon for you to see. I have been really scared sometimes during the development because the capacitor bank at the output of the PFC stores around 200 joules
.
Eva said:Uhm, this is a promising project... Go on with it. I wish I could have double-sided hole-plated PCBs with solder mask and silkscreen as yours delivered quick and cheap here
BTW: My new 2500W low-frequency class D amplifier powered with 416V directly from the output of a PFC is working and almost ready. I may post some pictures soon for you to see. I have been really scared sometimes during the development because the capacitor bank at the output of the PFC stores around 200 joules
...200 joules, so you have something like 2200uF at 416V..
Oh, yes... you designed your inrush current limiter for obvious reasons....and just to mention... don't let your screw driver drop into the running circuit.... or fry your necklace... it's a good habit to be slightly scared, when handling such energies.
My PCB... fast? ...delivery time is 2 weeks. And it is not cheap at all, even in China. But living in China is cheap, so I can spend more money on my hobby. And honestly I start to enjoy professional PCBs, even if expensive. It is fantastic: You play around at the computer with some CAD software, make some schematic, some virtual pictures..., then let the machine generate some mysterious data files and send them out (OK, I first had a look to them with a gerber viewer). Like from wizzard's hand some weeks later you get the real thing!!! And it is really exactly what you wanted to have! I love it!
Aren't there PCB companies in Spain? Or is it mostly a matter of money? How much would it be ?
First halfbridge module
I assembled the first half bridge module.
For my experiments I used a mostly U-shaped aluminium plate. Outline 100mm x 130mm. Finally the PCB allwos easy assembly on a 160x100 plate which would offer some space for the filter and also enable to insert the module into Euro-Rack systems.
I assembled the first half bridge module.
For my experiments I used a mostly U-shaped aluminium plate. Outline 100mm x 130mm. Finally the PCB allwos easy assembly on a 160x100 plate which would offer some space for the filter and also enable to insert the module into Euro-Rack systems.
Attachments
Something has happened ! 
Don't ask why.
I just assembled and powered up.
... and it seems to work...
I adjusted the dead time according my former experiments with my bread board, which showed that you can allow quite some overlap/crossconduction.
Also thermal stability of the crossconduction peaks was unexpected good.
I adjusted this module to overlap peaks, which are nicely wiped out by the inductive currents of the filter choke during idle operation.
Have a look to the zip file....
Don't ask why.
I just assembled and powered up.
... and it seems to work...
I adjusted the dead time according my former experiments with my bread board, which showed that you can allow quite some overlap/crossconduction.
Also thermal stability of the crossconduction peaks was unexpected good.
I adjusted this module to overlap peaks, which are nicely wiped out by the inductive currents of the filter choke during idle operation.
Have a look to the zip file....
...also first power tests are fine.
The attached screen shot shows the output voltage across a 0.75 Ohms load (just a long copper wire...).
The output filter was 42uH & 22uF. Something which would be OK for a sub woofer.
At about 50A the over current protection is shutting down the system. The screen shot show operation close to shut down. This halfbridge is delivering about 800W into 0.75 Ohms. Means the full bridge can deliver 1600W into 1.5 Ohms (limited by max. current).
Or up to over 2kW into 2 Ohms, if you have a stiff +/-60V power supply... (My future palns are something like a +/-55V SMPS).
So far so good... I will have to assemble the second halfbridge.... combine both to a full bridge and perform some more measurements and listening tests...
The attached screen shot shows the output voltage across a 0.75 Ohms load (just a long copper wire...).
The output filter was 42uH & 22uF. Something which would be OK for a sub woofer.
At about 50A the over current protection is shutting down the system. The screen shot show operation close to shut down. This halfbridge is delivering about 800W into 0.75 Ohms. Means the full bridge can deliver 1600W into 1.5 Ohms (limited by max. current).
Or up to over 2kW into 2 Ohms, if you have a stiff +/-60V power supply... (My future palns are something like a +/-55V SMPS).
So far so good... I will have to assemble the second halfbridge.... combine both to a full bridge and perform some more measurements and listening tests...
Attachments
darkfenriz said:
Hi Dark !
Yes it is a simple hysteresis self resonator with a fast comparator (LM160). This comparator is directly driving the IRS20954. And because of the circuit simplicity I called it Rookie. I am going to call all my simple straight forward designs Rookie_xxx.
For more sophisticated designs (may be adaptive dead time.... or in analogue may be cascoded input stages and VAS, or special error corrections...) I will have to find a different naming.
Have a look to my first post. ...pick the data sheets of both ICs and the circuit will start to make sense.
Hi Markus 2006,
....well, my real name is also Markus...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.