6000W By IRS2092

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
PCB for irs2092

Hello guys,

I'd like to ask to someone if please is possible to receive pcb files for the famous 6kw class d power amplifier.... :D

please send me a copy at my mail fabiodord#libero.it

My intent it's refurbish a couple of old burned amplifier with a "lite" good class d power amplifier.

These amplifiers has a good power supply stage but all the power amp are burned and probably with the same cost I can replace the old AB class with the class D....

Thanks in advance
 
First of all, Happy New year to all of you :)
Nice to see some of the old guys and girls still being around, Eva, Pafi, Choco, Fredos .... I was begining to fear that class-d had become something with only single chip solutions on this forum:rolleyes:

I have been working on a small prototype for some time (have had all to much to do at work to find time for class-d :( )
It's a "motherboard" for the switching part, based on 74HC86 for deadtime,IR2110 and just as a start IRF640. The intention is to make this switching part work well, and then being able to try out different modulator designs, by adding/changing a small board.
I have got it all to work in SODA mode now. The morar is running on +-30V at the moment. .... looks prommesing.

Anyway, Eva could you explain a litle more on this:

- Add diodes from IR chip outputs to their rails to prevent latchup when the outputs are forced to exceed the rails.
- Add a diode from VS to COM to prevent the IR chip from failing due to excess negative spikes on VS. These diodes should be SMD to keep things small and low inductance.

Your sggestion/advice is to connect 4 diodes in all (in SE mode) ... see my drawing. Is that correct? What diodes are you using for this? Guess these dont have to rated very high Id wise.

The Vs to COM is ofcourse only necessary when using the IRS2092.

- Add LC filters to supply rail inputs to prevent the HF ripple due to high current switching from reaching wirings.
What you are saying is just to as a couple of inductors, one to each rail ... see my drawing. Could you get by with just a couple of beads??
I'm guessing you are using beads in your design (not just in the rails). I'm not at all good at magnetics, so could you tell which beads you are using ... material, size ....
I guess that if these have too high peramability or are too small for the current used, they go into saturation and dosen't work at all .......

(Yes I know the output coil is to close to the input board on my PCB, but it dosen't seem to make any problems with the current quite low voltage PSU .... it will be changed on the next board ..... but I'm still not sure how to position it most correctly .... does anyone have a good drawing of how the leakage fields look like for a toroid coil?)

Best New Year wishes Baldin :cool:
 

Attachments

  • Eva 001.jpg
    Eva 001.jpg
    97.6 KB · Views: 2,929
  • IMG_7051 (Small).JPG
    IMG_7051 (Small).JPG
    69.4 KB · Views: 2,782
Last edited:
Diodes should be:
from VSS to LO
from LO to VSS+12V
from VS to HO
from HO to VB

Anyway IR2110 and the likes seem to be happy with as much as 5V inductive override, there is an application note on IFR site describing parasiting diodes inside the chip and how can an inductive spike cause any trouble.

You may well end up not needing any of these diodes with proper layout and not-hyper-fast di/dt ;)
 
Deadtime and gate drive

From what I read in your post Eva, you are only introducing delay between driver and gate. Is that correct?

I'm using a 74HC86 at the input of the IR2110 driver, but would like to get rid of it all together, as it takes up space / adds complexity.
I was thinking of just skipping the 74HC86 but still use two separate RCDs on the two input pins .....
Any suggestions here???
 

Attachments

  • Deadtime circuit 001.jpg
    Deadtime circuit 001.jpg
    54.7 KB · Views: 2,608
you are right, fets gate would not go above vcc in order to turn on

Hi Luka and Workhorse
.... yes, yes, I actually got it when pointed out by darkfenriz ;)
I just drew up what I thought I read the frst time ...... drawings tend to explain things much better, than many words ;)


Thanks darkfenriz for the reference to the Design Tip ...... here is a link if anyone finds it hard to find: http://www.irf.com/technical-info/designtp/dt97-3.pdf

Baldin :)
 
From what I read in your post Eva, you are only introducing delay between driver and gate. Is that correct?

I'm using a 74HC86 at the input of the IR2110 driver, but would like to get rid of it all together, as it takes up space / adds complexity.
I was thinking of just skipping the 74HC86 but still use two separate RCDs on the two input pins .....
Any suggestions here???

I'm using some logic before the IR driver too, both for preventing too short pulses in either direction and for splitting the H/L signals. I use RCD directly at the inputs of the IR drivers to fine tune pulse widths and get almost no dead time equally on all transitions, which in my circuit requires pulse extension rather than shortening (IR2110 delays are not always matched, it depends on supply voltages, but they seem to be very consistent from part to part). The inputs of IR2110 are schmitt triggers and seem to have quite consistent thresholds so no additional buffering is needed.

Concerning the diodes, HO should be clamped to VB and VS with schottkys, LO should be similarly clamped to COM and Vcc, and finally VS should be clamped to COM. For HO and LO clamping 1A 20V schottkys are ok, I use low inductance SOD-323 packages. For the later purpose I use a bigger 3A 200V schottky in SMB package (consider figuring out the part numbers as some homweork :D ).

These diodes may not be always required, it depends on di/dt and layout parasitics. My layout is good, but I also use insane turn-off di/dt close to 4KA/us resulting in negative Vs spikes of more than 5V at high output currents and also negative spikes on HO and LO, so the diodes are almost required for reliability. At some point in the development I added buffer transistors to the outputs of the IR2110, which greatly reduces the need for diodes on HO and LO because the chances for negative or positive spikes are reduced, but I dediced to keep the diodes.

On your schematic the value of C2 is way too high and the connection seems quite strange, I have not tried to simulate it in detail but I think it's calling for trouble.

Concerning XOR gates, you may find 74AHC1G86 (74LVC1G86) single ones in SOT23-6 (and smaller ones too) to be quite useful. 74LVC1G57 and 74LVC1G58 are also quite versatile. LVC seems to have higher output current than AHC.
 
Last edited:
Hi Eva

Hmmm, had not seen this reply before now? ....

Thanks for the tips. Well it's an awful lot of diodes to find room for, but I think I might be experiensing the lack of these at the moment. Funny things are happening when either I remove the load or I try to start up without (shoot through instability).
Have tried to mount a 10k resistor from com to Vs as suggesed in IR DT 99-7, which at least prevent shoot through when starting up with no load ... but it the IF2110 dosen't start switching either ..... I'm scratching my head, and it must be something with the working of the IR2110 in the un loaded state ....

Home work: 1A/20V could be On's NSR1020MW2T1G and the 3A/200V could be MBRS3201T3G ;)

Haven't any schottkys at the moment, but will try with some ES1D

As to the size of C2. Yes it's too big, but was only for the first trials to make sure there was enough dead time.
Now running with 100pF, on FETs without heat sink. With a little alu on the FETs I could probably go down to 10-33pF.
The connection is right I belive, and it's running without any problem. You culd use two caps going to GND, but this connection saves one and works as well. :)
 
Concerning the diodes, HO should be clamped to VB and VS with schottkys, LO should be similarly clamped to COM and Vcc, and finally VS should be clamped to COM. For HO and LO clamping 1A 20V schottkys are ok, I use low inductance SOD-323 packages. For the later purpose I use a bigger 3A 200V schottky in SMB package (consider figuring out the part numbers as some homweork :D ).

These diodes may not be always required, it depends on di/dt and layout parasitics. My layout is good, but I also use insane turn-off di/dt close to 4KA/us resulting in negative Vs spikes of more than 5V at high output currents and also negative spikes on HO and LO, so the diodes are almost required for reliability. At some point in the development I added buffer transistors to the outputs of the IR2110, which greatly reduces the need for diodes on HO and LO because the chances for negative or positive spikes are reduced, but I dediced to keep the diodes.

.

The new series of IRS2110 & IRS2113 gate drivers address some of these failure modes and therefore they now come with INBUILT CLAMPING diodes except VS to COM diode. ;)


http://www.irf.com/technical-info/appnotes/an-1100.pdf
 
Last edited:
Clamping the output

I just tried to clamp the output (just after the filter) with 2 diodes (ES1D).
I will still try all the other clamps, but I was actually wondering what would happen to the stored energy in the filter, when the load is just removed. It might not be much in idel mode or just playing softly, but otherwise a large current would give rise to a very high voltage on the output ....

It did help; removing the load, it still keep oscillating, but nor stable, and there are still some shoot through ....

Eva and Workhorse, do you also use clamping dirodes on the output??
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.