OK. I implemented Fredos proposals for the supply. It is fine.
Also I reduced VCC to 12V again.
- Slightly more effective dead time.
- Less driver losses.
- Slightly less dv/dt stress for reverse recovery diode (seems to be around 3kV/us in hard switching condition)
And I added +0.7V (just one diode in series to the zener) for the supply of the high side driver, means VCC and (VB-VS) have now the same value.
Over current protection seems to work reasonable also.
I reworked the geometry and wiring and blocking caps of the power stage and now there is no ringing anymore. Nice.
Going on in design I came to the conclusion that I will need quite a good battery of fast/small e-caps in addition to the 2uF film caps right at the power stage. Furtheron these caps must be able to handle high AC ripple currents. Already in idle situation the triangular HF ripple in the output filter and power stage will ramp between +/- 3A if we assume +/-50V rails and 40uH at 100kHz.
I will have to set up a new PCB for my final full bridge amp.
Besides doubts regarding the reliability of my first prototype (Yes, A-Spice.. you are right... to many blobby solderings, to often changed wirings..), I would also like to optimize the PCB and have to plan some place for proper rail caps right at the power stages.
...well, after all this I think in some weeks I will open a new thread which is related to the amp. Starting from now things will be less directly related to the IR chip, but more to the amp design itself.
My statement so far:
I like this IC, even if there are lots of shut down mechanisms inside, which can cause fault tripping. With a proper design the IC seems to act reasonable without to much hypochondric shut downs.
I will go on to use it in my Class D amp project.
Thanks to you all and your valuable hints!
Also I reduced VCC to 12V again.
- Slightly more effective dead time.
- Less driver losses.
- Slightly less dv/dt stress for reverse recovery diode (seems to be around 3kV/us in hard switching condition)
And I added +0.7V (just one diode in series to the zener) for the supply of the high side driver, means VCC and (VB-VS) have now the same value.
Over current protection seems to work reasonable also.
I reworked the geometry and wiring and blocking caps of the power stage and now there is no ringing anymore. Nice.
Going on in design I came to the conclusion that I will need quite a good battery of fast/small e-caps in addition to the 2uF film caps right at the power stage. Furtheron these caps must be able to handle high AC ripple currents. Already in idle situation the triangular HF ripple in the output filter and power stage will ramp between +/- 3A if we assume +/-50V rails and 40uH at 100kHz.
I will have to set up a new PCB for my final full bridge amp.
Besides doubts regarding the reliability of my first prototype (Yes, A-Spice.. you are right... to many blobby solderings, to often changed wirings..), I would also like to optimize the PCB and have to plan some place for proper rail caps right at the power stages.
...well, after all this I think in some weeks I will open a new thread which is related to the amp. Starting from now things will be less directly related to the IR chip, but more to the amp design itself.
My statement so far:
I like this IC, even if there are lots of shut down mechanisms inside, which can cause fault tripping. With a proper design the IC seems to act reasonable without to much hypochondric shut downs.
I will go on to use it in my Class D amp project.
Thanks to you all and your valuable hints!
Hello!
This is my first post on diyaudio, however i read this from the beginning to the end few times
! I decided to bring up this topic because I am developing a non-discrete UcD clone amplifier with this cute driver!
The complete schematic is here:
http://ftp.torespont.hu/solarian/ucd9.PNG
The prototype board (home made, so no through hole plating):
http://ftp.torespont.hu/solarian/ucd10.PNG
I got the pcb yesterday so i built it, but not the whole amplifier, firstly just the power stage, to see how it works! After some high side shutdown problems, i decided to bypass the high side overcurrent protection, bcos it was not correctly set, and i was too lazy! But the shutdown problems are still there, sometime the high side is just wont turn on for some cycles! After burning all the 3 drivers I bought im at lost! As i wrote above, i' ve read the whole topic already a few times, nothing solved this issue!
Thanks for the helps and advices in advance!
This is my first post on diyaudio, however i read this from the beginning to the end few times
! I decided to bring up this topic because I am developing a non-discrete UcD clone amplifier with this cute driver! The complete schematic is here:
http://ftp.torespont.hu/solarian/ucd9.PNG
The prototype board (home made, so no through hole plating):
http://ftp.torespont.hu/solarian/ucd10.PNG
I got the pcb yesterday so i built it, but not the whole amplifier, firstly just the power stage, to see how it works! After some high side shutdown problems, i decided to bypass the high side overcurrent protection, bcos it was not correctly set, and i was too lazy!
But the shutdown problems are still there, sometime the high side is just wont turn on for some cycles! After burning all the 3 drivers I bought im at lost! As i wrote above, i' ve read the whole topic already a few times, nothing solved this issue!Thanks for the helps and advices in advance!
Hi Solarian!
...yes, it's frustrating to run out of drivers.
I also had to re-order... And hell... transport from US to China + import taxes are not funny.... But I am glad that I can get these devices within short time and without complications.
It is hard to determine which shut down became active.
If there is no signal of the upper switch this can have two reasons.
1. Some shut down became active, but then also the lower switch should remain OFF. You can see at the CSD pin, if it is really a shut down or not.
2. If CSD does not show a shut, you might suffer from some misbehaviour which noticed... If your input signals are not clean, some spikes or short pulses (which do not reach proper input level tresholds...) may disturb the chip in a way that it turns ON or OFF the driver for long times. Fortunately I never observed that the IC turns ON both at the same time. This locking seems to be reliable.
Furtheron:
-You are heavily loading this small driver,
I had the impression that additional freewheeling schottky diodes at the driver outputs helped to avoid IC destruction.
-Have a look to the wire between upper and lower MosFet. This wire has very high di/dt and already some nH can generate serious voltage peaks.
-COM and VS must sense the drain-source-voltage-drop directly at the transistor pins and track routing must be a HF low impedance connection with small loop area.
-In my case also optimized blocking of the DC-rails was helpful.
Also posting 44....49 of this thread
http://www.diyaudio.com/forums/showthread.php?s=&threadid=91637&perpage=10&pagenumber=5
may help you. I will go on to post my findings and results there, but don't expect to much newa during next week.....
Good luck so far
Markus
...yes, it's frustrating to run out of drivers.
I also had to re-order... And hell... transport from US to China + import taxes are not funny.... But I am glad that I can get these devices within short time and without complications.
It is hard to determine which shut down became active.
If there is no signal of the upper switch this can have two reasons.
1. Some shut down became active, but then also the lower switch should remain OFF. You can see at the CSD pin, if it is really a shut down or not.
2. If CSD does not show a shut, you might suffer from some misbehaviour which noticed... If your input signals are not clean, some spikes or short pulses (which do not reach proper input level tresholds...) may disturb the chip in a way that it turns ON or OFF the driver for long times. Fortunately I never observed that the IC turns ON both at the same time. This locking seems to be reliable.
Furtheron:
-You are heavily loading this small driver,
I had the impression that additional freewheeling schottky diodes at the driver outputs helped to avoid IC destruction.
-Have a look to the wire between upper and lower MosFet. This wire has very high di/dt and already some nH can generate serious voltage peaks.
-COM and VS must sense the drain-source-voltage-drop directly at the transistor pins and track routing must be a HF low impedance connection with small loop area.
-In my case also optimized blocking of the DC-rails was helpful.
Also posting 44....49 of this thread
http://www.diyaudio.com/forums/showthread.php?s=&threadid=91637&perpage=10&pagenumber=5
may help you. I will go on to post my findings and results there, but don't expect to much newa during next week.....
Good luck so far
Markus
Thanks for the reply and the tips!
I order the ICs from Farnell, its way more cheaper than ordering it from USA, and i receive it in one day !!
Noone noticed my sillyness that i forgot to ground the latch input of the comparator (also i forgot to mention that i used that comparator to generate some square waves ) MAYBE the problem was there!(and my poor scope) I've redesigned the whole proto board, it will be modular for easier developing!
OH, and the most important: I was surfing on IR homepage, and suddenly noticed that IRS20954 is NRND, and IRS20955 became active! Ive read the application note about the differences: More accurate OCP and slightly modified init sequence! THATS GREAT!
I order the ICs from Farnell, its way more cheaper than ordering it from USA, and i receive it in one day
!!Noone noticed my sillyness that i forgot to ground the latch input of the comparator
(also i forgot to mention that i used that comparator to generate some square waves ) MAYBE the problem was there!(and my poor scope) I've redesigned the whole proto board, it will be modular for easier developing!OH, and the most important: I was surfing on IR homepage, and suddenly noticed that IRS20954 is NRND, and IRS20955 became active! Ive read the application note about the differences: More accurate OCP and slightly modified init sequence! THATS GREAT!
$2.5 1K unit, hm, it's seems slightly overpriced, for instance tda8939 was available for $3, but it was all in one without PCB problem etc. Actually, i could buy l6385 + fine dead time adjustment+OCP+bc856 as level shifter for <$2, or fully discrete, that else cheaper one. I guess IR still haven't mass sales of the class D driver, and i hope it will be soon yet. Pretty strange, that in the China mass production, very popular Linfinity's class D chip with si4558 so-8 P-N mosfet pair, BTW it's very similar to the njm4558 so-8 dual opamp, populated on the same PCB.
OFFTOPIC:
Hi bbp!
in your profile you say: ' I'm only a student.'
Don't you you think this is a valuable status?
From my perception this is a very valuable status. Good education and fundamental know how is one of the important factors to bring a country forward. Many Western countries seem to underestimate this ...and run in danger to loose the dominant position.
...sorry, I'm just a foreigner in China. From Germany.
But in fact there are also Class_D designers from China.
How about you? Do you design Class_D or SMPS?
If you wish to extend this OFFTOPIC conversation, let's better switch to mail or start a Texas-thread....
Hi bbp!
in your profile you say: ' I'm only a student.'
Don't you you think this is a valuable status?
From my perception this is a very valuable status. Good education and fundamental know how is one of the important factors to bring a country forward. Many Western countries seem to underestimate this ...and run in danger to loose the dominant position.
...sorry, I'm just a foreigner in China. From Germany.
But in fact there are also Class_D designers from China.
How about you? Do you design Class_D or SMPS?
If you wish to extend this OFFTOPIC conversation, let's better switch to mail or start a Texas-thread....
To ChocoHolic
To ChocoHolic,
At the beginning,I was so suprised that your english is so well,actually,you're a German in China.I'm so happy to meet you here,too!
My major course in the university is Electrical Science and Technology,so,I'm intrest in desingning amplifiers.Recentlly,I have a project base on TAS5261 from TI.
But I'm not a full member in Diyaudio yet,so I can't see your e-mail address now.I wish you could sent me an e-mail first,so that I can contact you.
Best wishes to my Germany friend!
To ChocoHolic,
At the beginning,I was so suprised that your english is so well,actually,you're a German in China.I'm so happy to meet you here,too!
My major course in the university is Electrical Science and Technology,so,I'm intrest in desingning amplifiers.Recentlly,I have a project base on TAS5261 from TI.
But I'm not a full member in Diyaudio yet,so I can't see your e-mail address now.I wish you could sent me an e-mail first,so that I can contact you.
Best wishes to my Germany friend!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.