SystemD_2kW, any interest for an open design? - Page 32 - diyAudio
 SystemD_2kW, any interest for an open design?
 User Name Stay logged in? Password
 Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read Search

 Class D Switching Power Amplifiers and Power D/A conversion

 Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
diyAudio Member

Join Date: Dec 2003
Location: Munich
Quote:
 Originally Posted by Workhorse ...you have to parallel them almost 30X in order to ensure trouble free operation when pumping multikilowatt into 2 ohms load.You haven't seen them blasting and evaporating in the air. Kanwar
Hm, this is sounding like a serious comment.
2x5parallel is for 1.2kW into 2R.
For multiple kW it would be two half bridges with 2.4kW into 4R and summing up the caps to a 10x parallel structure.
The true current load is depending on the layout and the equivalent circuits for the small caps and e-caps. My detailed simulation of the switching stage is delivering a current load of 0.59Arms per SMD cap when powering 35Ap sines (2.4kW into 4R) with a 1/8 duty cycle.
Together with the 0R7 ESR this turns into losses of 244mW per cap. That's how I came to the conclusion that it is close to the limits, but not catastrophic. I fact my thermal measurements while hottrodding the amp shows no self heating of the caps, which indicates in reality less than the calculated losses.
Their temperature is following very close the surrounding temperatures of the MosFets and the heat sink above, 3-5 deg less than the close surroundings.
Despite seeing an acceptable situation from all that, it would be silly to ignore your input, especially as it sounds like you already have burned fingers.
The PCB easily allows to double stack (back pack) each cap.
This would sum up to 20x parallel for multiple kW, which is coming close to your advice.
So I will go on hottrodding the amp with just 5 parallel caps and change the BOM to double stacked for future builds.

diyAudio Member

Join Date: Dec 2003
Location: Munich
Quote:
 Originally Posted by TroelsM Hi If I have a amp with post- and pre-filter-feedback then I need to break the loop after the two are joined right? - that would be at the comperator-input for a lot of designs. It might be very rudimentary , but especially for switching circuits I find it hard to fully grasp. For a selvf-osc D-amp, we have a delay(phase) that is dependent on the output volume (because freq drops as volume goes up) This level-dependent phase confuses me even more as a "small-signal-sim-model" would not be correct.
With pre and post filter feedback, you can only specify the prefilter loop gain and the post filter loop in seperate by breaking the related loop at the concerned different points.

Delay & phase:
The phase is usually dominated by the filter.
Furtheron the relevant delay which slightly impacts the phase of the forward path is the propagation delay of chain, not the period of the switching frequency.

diyAudio Member

Join Date: Dec 2003
Location: Munich
Quote:
 Originally Posted by Cresnet Then I think that you get around 620W @ 8R Load?
About that. Slightly more, because the rails will sag less than with 4R.

Edit:
There seems to be a factor 2 error in your calculations, which I did not notice at first glance.
Unbridged you may end up around 350W into 8R.
Bridged it will be around 1300W into 8R.

Quote:
 Originally Posted by Cresnet if I understand you properly, its much harder to make a stable D amplifier @ 8R load?
No, there is nothing harder. It is just about two topics:
-Which speed you want to achieve over which range of load impedances. You can always compensate it towards slower step response and avoid overshoot under all conditions.

It is a choice which the design engineer can make and there is a certain area of reasonable settings, depending on personal taste.

Last edited by ChocoHolic; 1st March 2013 at 07:27 PM. Reason: see Edit....

diyAudio Member

Join Date: Dec 2003
Location: Munich
Settling and verifying the over current protection between 50A...60A gave me many more insights in this amp. In any case it was necessary to settle stronger snubbers for Drain-Source and Gate-Source in order to keep the disturbance level low even at such high currents. Also having the additional shottky inside the current detection path of the lower MosFet is not intended by the IR chip, but can be handled.

My test load was series connection of 1uH and 47mOhm.
The first screen shot sows the voltage across the 47 mOhm right in the moment when the over current protection is acting. It acts at 2.5V which translates to 53A.
How did I catch this moment? Simple. The protection has a auto reset function, so I simply let the load connected and the input signal large enough to trigger the shut down.
While the amp is continuously running into shut down again and again – you can get every shut down event with the singe shot function of the scope.

For the second and third screen shot I used the amp with DC coupled input and gave some DC input signal resulting in a DC output which give stable signals for the measurements.
The level was adjusted to 50A load current, close to shut down.
I did not chose a resonant snubber because the disturbances showed dual ringing modes at 35MHz and 125MHz. The screen shots show that these disturbances are now catched by the snubbers almost completely.

The second screen shot is showing the turn ON transition of the low side MosFet in very detail with a time scale of 20ns/grid.
Upper trace Uds: 50V/grid
Lower trace Ugs: 5V/grid

The third screen shot is showing the turn OFF transition of the low side MosFet in very detail with a time scale of 20ns/grid.
Upper trace Uds: 50V/grid
Lower trace Ugs: 5V/grid

All measurements with 1:10 probes.
Attached Images
 0R047_53A.jpg (675.6 KB, 685 views) MosON_50A.jpg (722.8 KB, 623 views) MosOFF_50A.jpg (659.6 KB, 551 views)

diyAudio Member

Join Date: Dec 2003
Location: Munich
As a consequence of the adjustments for the over current protection and about reproduction of rectangles I went for new overall adjustment, whereof I will show updated measurements in the coming posts. V1.3 is born.

Attached two screen shots, which show the output signal and the hysteresis signal.
Towards high output levels you can see the hysteresis being consequently reduced, which is keeping the output voltage ripple reasonably low.
First screen shot without load, second with 2R driven into clipping.

Upper trace: Output voltage, 50V/grid
Lower trace: Voltage a non inverting input of the comparator, 1V/grid

All measurements with 1:10 probes.
Attached Images
 DynHyst10kHz_noload.jpeg (273.9 KB, 509 views) DynHyst10kHz_2R.jpeg (268.5 KB, 384 views)

diyAudio Member

Join Date: Dec 2003
Location: Munich
Rectangles:

The new adjustment also has a new frequency compensation.
It is the fastest adjustment, which ensures not overshoots at any load and signal level.
It is just 20% slower at lower and medium levels compared to V1.2, but I am using the clipping of the OP amp for artificially slowing down at very high levels.
Fast enough that I can accept it.

Attached 9 screen shots.
White trace: Output voltage
Red trace: Voltage at the non inverting input of the comparator, just to show again the action of the dynamic hysteresis at the 120V level.

30Vpp into 2R
30Vpp into 8R
60Vpp into 2R
60Vpp into 8R
120Vpp into 2R
120Vpp into 8R

All measurements with 1:10 probes
Attached Images
 Rect30Vpp2R.jpeg (348.5 KB, 99 views) Rect30Vpp8R.jpeg (345.9 KB, 65 views) Rect30Vppnoload.jpeg (347.0 KB, 43 views) Rect60Vpp2R.jpeg (332.0 KB, 28 views) Rect60Vpp8R.jpeg (333.5 KB, 29 views) Rect60Vppnoload.jpeg (339.9 KB, 19 views) Rect120Vpp2R.jpeg (308.2 KB, 28 views) Rect120Vpp8R.jpeg (319.9 KB, 38 views) Rect120Vppnoload.jpeg (331.6 KB, 48 views)

diyAudio Member

Join Date: Dec 2003
Location: Munich
THD and IMD
Attached Images
 THD_V1Dot3_1kHz.JPG (41.0 KB, 136 views)

diyAudio Member

Join Date: Dec 2003
Location: Munich
After some fine tuning I got a more consistent character.

Also fs is now again in the desired range.
In simulation 343kHz. In reality it is 329kHz.
Attached Images
 THD_V1Dot3_1kHz_finetuned.JPG (42.2 KB, 125 views)

 5th March 2013, 06:45 PM #319 diyAudio Member   Join Date: Dec 2003 Location: Munich ...sudden silence? I guess you want to say: Stop optimizing, but release the builders package. Ok, understood. And since the over current protection is working now as well, I think the design is mature enough to go for the builders package. This week and coming week will be very busy in work, so I cannot promise a schedule for the builders package - but I am starting to prepare the data for V1.3.
 5th March 2013, 08:32 PM #320 diyAudio Member   Join Date: Nov 2005 Dear Chocoholic, if followed this thread every day from the very first beginning. This is by far the most mature and well thought DIY project (concerning class-d) i am aware of! Thank you very much for the heap of theory and all the big and small advices, techniques and skills in swithmode circuit design you've presented here. Futheron i'd like to thank you for the time, money and effort you pushed into this ambitious project! So far I have learned a lot and I'm looking forward towards your builders-package. kind regards!

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is OffTrackbacks are Off Pingbacks are Off Refbacks are Off Forum Rules

 Similar Threads Thread Thread Starter Forum Replies Last Post nattawa Power Supplies 5 8th October 2016 08:28 PM Llama16 Group Buys 1 12th September 2012 05:23 PM SimontY Group Buys 2 4th March 2009 01:53 PM lawbadman Multi-Way 1 17th January 2008 12:15 AM gfergy Everything Else 1 15th April 2007 07:33 AM

 New To Site? Need Help?

All times are GMT. The time now is 02:00 AM.