Badly in need of help on output inductor design for ZXCD1000 based 500 watts amp

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

Though new to Class D amplifier design; i had completed a design for 150 watts.

Using a standard, off-the-shelf 20uH gapped toroid PG0058 from pulse. Now I consider myself confident enough to port it to a 500 watts output power. :D

The problem is finding an easily available gapped (or something similar toroid, RM or EE core) capable of handling 20 amps :hot: current without saturation.

With level of discussions everyday happening here at diyaudio.com Can somebody please guide me to an informative link or share horizons to success...:rolleyes: All suggestions are most welcome... :)



_______________________________________

Regards
Chaturvedi

The good news is that nothing is compulsory.
The bad news is that everything is prohibited
 

Attachments

  • p554.zip
    94.8 KB · Views: 258
Razor_Edge said:
I am attaching the circuit diagram for reference.

I would really appreciate and modification and suggestion from DIYers...:)


Hi,

While I can't help you with the inductor question, and I know others will, I think your driver and output stage will fall apart at 500 watts.

It seems most wouldn't consider using a comp. output stage past 200 watts.

I dont' know anything about the IC you're using either, but it seems strange to me that it's driving the MOSFETs directly with no buffers.

I think for a 500 watt amp you're looking at upgrading to a dual N channel output stage driven discretly or via a good driver IC.

Best Regards,
Chris
 
I dont' know anything about the IC you're using either, but it seems strange to me that it's driving the MOSFETs directly with no buffers.

I think for a 500 watt amp you're looking at upgrading to a dual N channel output stage driven discretly or via a good driver IC.



Firstly: The datasheet for this IC states that it can drive a Mosfet for a Ciss upto 2200pf with a similar configuration. The datasheet can be downloaded from link below.

same as defined in the datasheet for IR2010 driver. :smash:

I would like your critics on this part. (where am i lagging... :rolleyes: )



Anyways, thanks for the valuable feedback Chris... :)


Best Regards

http://www.zetex.com/audio/audio05-1-2a.asp
 
Razor_Edge said:
I dont' know anything about the IC you're using either, but it seems strange to me that it's driving the MOSFETs directly with no buffers.

I think for a 500 watt amp you're looking at upgrading to a dual N channel output stage driven discretly or via a good driver IC.



Firstly: The datasheet for this IC states that it can drive a Mosfet for a Ciss upto 2200pf with a similar configuration. The datasheet can be downloaded from link below.

same as defined in the datasheet for IR2010 driver. :smash:

I would like your critics on this part. (where am i lagging... :rolleyes: )



Anyways, thanks for the valuable feedback Chris... :)


Best Regards

http://www.zetex.com/audio/audio05-1-2a.asp


This part right here:

"The ZXCD1000 class D controller IC is inherently
capable of driving even higher power solutions, with
the appropriate external circuitry
. However as stated
above the maximum supply voltage to the ZXCD1000
class D controller IC is 18V and the higher supply
voltages must therefore be dropped. Also due
consideration must be given to the ZXCD1000 output
drive levels and the characteristics of the bridge
MOSFET’s. The latter must be sufficiently enhanced by
the OutA and OutB outputs to ensure the filter and load
network is driven properly.If the gate drive of the
ZXCD1000 is too low for the chosen MOSFET then the
OUTA and OUTB signal must be buffered using an
appropriate MOSFET driver circuit. Additionally,
suitable magnetics are essential to achieve good THD
performance."


Not to mention parasitics, trust me at 500W you're in for some real trouble, call it a feeling.

Regards,
Chris
 
trust me at 500W you're in for some real trouble, call it a feeling

:xeye:

I thought this issue would be a bit easier to address that getting the optimum magnetics designed for high power... :cannotbe: was i so wrong in my interpretation..:rolleyes:

Can anybody suggest be a good Mosfet driver (which can handle more that 3000 pf of input gate to source capacitance within 50 nSec)

I stongly believe in keeping the dead time to less than or equal to 15nS for good THD performance (please comment)...


Thanks Chris for redefing my immediate objectives. I still dont know how to go about magnetics capable of handling 20~25 Amps of pulsed current into the transducer.

NOTE: The current design works on 200Khz +/- 50 Khz operating frequency with THD below 0.01%.
 
Chris is right, you will want to go with an all N-channel setup at 500W. It would be easiest to use a gate driver like the National LM5104 that requires a single input and provides dead time control. That way you can keep the ZXCD1000 and not have to add any additional circuitry between the PWM controller and gate drivers. However, I am currently using the LM5104 and it looks like it handles dead time control poorly. Plus there are better gate drivers [IMHO, which I unfortunately am finding out the hard way] with separate high and low inputs. I would suggest using a discrete design by taking the complimetary outputs from the ZXCD1000 and using one to drive each side of the bridge with the other output connected to a RC dead time adjustment and then inverted [HC04]. Does that make sense? Then you can use a myriad of different gate drivers - I suggest the IR2010 or HIP2100.

As far as magnetics, my best suggestions from others that I am passing on is to use a distributed gap common toroid. That gies you a smaller footprint and keeps your flux fringing down. I'd stay away from split gap toroids or separate inductors. You can achieve good results with any method, but I am just pasing on what I have heard. Sorry I can't help you with core material or anything.

Best regards in getting some great performance out of your amp!
 
What kind of output MOSFETs? Q1...Q4 is not marked on the schematic.

the output mosfets are from zetex ZXM64N035L3 and ZXM64P035L3 50.

Are yaar tum Air core inductor kiyon nahi use karte ho,
I have used air core inductors upto 50A , as they are unsaturable types...


At 500 watts of output power the amount of radiation would be very very high. that is y I am keeping myself from air core inductance KANWAR sahab... (hindi mein peheli baar post kar raha houn, acha laga...:))

I have got very good results by using gapped core inductor from pulse.

Actually getting ferroxcube gapped core at a competitive pricing from local vendor is giving me a hard time...

yesterday, I tried using a N67 based EE core (air gap in the central limb). Its getting very hot at just 155 watts...!!!

No success absolutely in this direction...:bawling:


hey kenshin,

I am thinking of keeping the rail voltages at 70 volts(good mosfets with better THD performance at lower voltages). Please comment...
 
"A very important feature of the Zetex solution is that the residual distortion is almost totally free of any crossover artifacts. This lack of crossover distortion sets the ZXCD1000 solutions quite apart from most other presently available low cost solutions, which in general suffer from severe crossover distortion
problems."

How could it be? The output stage is a external AC coupled N+P bridge built of diodes and resistors, driven by a single output of the IC. (notes that IR's gate driver could control BOTH gates directly). So the dead time is determined by external components and it's "unknown" to the controller. Also, the chip contains only oscillators and comparators, no any compensating circuitary.

Without any "zero dead time"or dead time cancelling techniques, how could it eliminite cross-over distortion in a open-loop configuration?
 
Kenshin said:
"A very important feature of the Zetex solution is that
the residual distortion is almost totally free of any
crossover artifacts. This lack of crossover distortion
sets the ZXCD1000 solutions quite apart from most
other presently available low cost solutions, which in
general suffer from severe crossover distortion
problems."

How could it be? The output stage is a external AC coupled N+P bridge built of diodes and resistors, so the dead time is determined by external components and it's "unknown" to the controller. Also, the chip contains only oscillators and comparators, no any compensating circuitary.

Without any "zero dead time"or dead time cancelling techniques, how could it eliminite cross-over distortion in a open-loop configuration?


I'm guessing it can't. Perhaps the chip is fast enough that with the right values and proper selection of FET and inherent delay it just works out that way. They still get away with saying that by using the phrase "with the appropriate external circuitry", and leaving it up to you to come up with it.
 
There are a few issues with your schematic I see:

1. The ZXM64x035L3 are rated at 35V. This isn't enough voltage to support 500W into 4ohm. And their current rating isn't great enough to support 500W into a lower impedance. I am surprised the FETs handled the 150W power from your original amp. I definitely wouldn't go any higher than that.

2. For D2-5 you used 4148s. These need to be 10-13V [or so] zeners. Their purpose is the clamp the gate voltage if it ever exceeds the Vgs of the FETs. You really don't need them, but in an emergency they might help.

No kidding about the dead time. Have you seen the 100W eval board layout? It's got to be terrible in audio performance. FETs scattered all over the place with long traces. It leaves plenty of room for measuring, heatsinking, and developing, which I imagine might have been the ultimate goal. They could be referring to the inherent crossover distortion advantage that BD modulation has over AD??
 
Razor_Edge said:

At 500 watts of output power the amount of radiation would be very very high. that is y I am keeping myself from air core inductance KANWAR sahab... (hindi mein peheli baar post kar raha houn, acha laga...:))


Chaturvedi , yaar why dont you shield the inductors in a steel chamber, i have done it with success and no EMI problems encountered.....

K a n w a r
 
1. The ZXM64x035L3 are rated at 35V. This isn't enough voltage to support 500W into 4ohm. And their current rating isn't great enough to support 500W into a lower impedance. I am surprised the FETs handled the 150W power from your original amp. I definitely wouldn't go any higher than that.

2. For D2-5 you used 4148s. These need to be 10-13V [or so] zeners. Their purpose is the clamp the gate voltage if it ever exceeds the Vgs of the FETs. You really don't need them, but in an emergency they might help.


Sorry I forgot to mention that intially I started with that configuration (ZXM64x035L3 and 1N4148 for power 24 Volts op voltage)

Right now the Mosfets used are IRF530 and IRF9530 with 10 volts zener. Also, planning to move on to IRF740 for higher powers >500 watts @ 70~90 volts.


Chaturvedi , yaar why dont you shield the inductors in a steel chamber, i have done it with success and no EMI problems encountered.....

Hmmm... I never thought about that. I guess its my ignorance about the subject.

A known devil is better than an unknown angel
But Kanwar saheb, what would you say about huge bundle of wire OR shearing DCR (1~1.2 ohms) eating precious resources like space on the PCB??

Can you suggest something about getting gap done in the toroids??


Does someone has a hands on experience or knows how to go about the gaps...? :D (Or may be a alternative like RM core based or even a air core with very low DCR)

I tried slitting the toroid with a high speed silicon carb rotary blade (1.15 mm groove). the core is chipping off like small feathers....
:xeye:
 
Razor_Edge said:


Hmmm... I never thought about that. I guess its my ignorance about the subject.

A known devil is better than an unknown angel
But Kanwar saheb, what would you say about huge bundle of wire OR shearing DCR (1~1.2 ohms) eating precious resources like space on the PCB??


Chaturvedi Ji,

yes it is huge bundle of wire, but you can lower its DCR by using MultiWire windings to reduce the skin effect...
 
Razor_Edge said:
Thanks Chris,

Could you please elaborate (foil)...

Did you meant warpping up the coils in a metiallic foil? :confused:


Bruno had said he gapped toroidal ferrite cores with a dremel mounted in a drill stand. I'd imagine with the right abrasivness, pressure and enough speed it wont' chip so much. Should take a few minutes to cut the gap that way, you can even experiment with lubrication for a cleaner cut? Depends what you get for a cutting disc I suppose.

Foil coils have been discussed in the class d forum but I can't think what thread it was now. They have a few advantages and you can build your own, most likely as an air core. I mean replacing the wire with insulated foil, not as some type of shield. Just another option to consider.
 
Checkout the Coopper Bussman (Coiltronics) DR and HC series. You might have to parallel a couple coils to get to your inudctance.

http://www.cooperet.com/products/products.cfm?page=high_current_inductors

I am using a DR127-8R2 8.2 uH which handles 12 amps.

What FETs will you use for the 500 watt amp? My rough calulation shows you must be using a 4 ohm speaker to get 150 watts from 35 Volts.

To get to 500 watts you will need a 65 Volt supply. Do you have a FET 16 Amps at 65 Volts?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.