Now i put some 0.47µF caps close to the fets and that took out most of the ringing but the overshoot remains so i need fast recovery diodes, atlest between drain and source but maube between gate and source as well to clamp the overshoot.
Only short the wire or put it bigger between your fets and output coil, this should fix your probleme!
fredos
www.d-amp.com
fredos
www.d-amp.com
TOINO said:
The HEXFRED Recovery time is 30nS......which is slower than Schottky.......which switches in range of PicoSeconds.....but you miss another point the avalanche breakdown voltage of mosfet is 300V and also the reverse recovery time of Body diode of mosfet is around 300nS whereas the HEXFRED is 30nS which would eventually prevent the schottky from reverse breakdown....by conducting the reverse spikes/transients much earlier then the body diode of mosfet....and hence schottky is always safe........
chris, need info about your miniUcD
****sorry to post it here******
Hello chris,
finally made up my mind to make UcD.. 🙂
bellow is a picture of your miniUCD....
#0>> whats the amp frequency??? i mean the oscillation freq?? can it be changed by altering R5,C2,R6?? as THEY are he only feedback elements....
#1>> circled in blue, some sort of voltage regulation right?
I can still use 2 2N3906, right? 🙂 as it wont take up much power on the input side..
#2>> in blue box, WHY parallel 4 10ohm 2W resistors?? i can see that it powers up the LO side of the MOSFET drive.. does it take up THAT MUCH current?? compared to he HI side?? the HI side has 1K (R22) dropping resistor w/the zener... how come?
#3>> green box, whats the use of this diode?? it looks like powering up the HI side from th LO side
#4>> boxxed in RED, If (say) I want to have HIGHER output power, surely It would mean UPPING up the power rails (+/-15V) to something like (+/- 35V or so)...IS it possible? without adding components?? or do I have to change something?? some component value perhaps??
#5>> what other N-Ch mosfets can I use besides IRF630?? can i use IRFZ44? if not, why not??
sorry for many questions...
regards,
raff
****sorry to post it here******
Hello chris,
finally made up my mind to make UcD.. 🙂
bellow is a picture of your miniUCD....
An externally hosted image should be here but it was not working when we last tested it.
#0>> whats the amp frequency??? i mean the oscillation freq?? can it be changed by altering R5,C2,R6?? as THEY are he only feedback elements....
#1>> circled in blue, some sort of voltage regulation right?
I can still use 2 2N3906, right? 🙂 as it wont take up much power on the input side..
#2>> in blue box, WHY parallel 4 10ohm 2W resistors?? i can see that it powers up the LO side of the MOSFET drive.. does it take up THAT MUCH current?? compared to he HI side?? the HI side has 1K (R22) dropping resistor w/the zener... how come?
#3>> green box, whats the use of this diode?? it looks like powering up the HI side from th LO side
#4>> boxxed in RED, If (say) I want to have HIGHER output power, surely It would mean UPPING up the power rails (+/-15V) to something like (+/- 35V or so)...IS it possible? without adding components?? or do I have to change something?? some component value perhaps??
#5>> what other N-Ch mosfets can I use besides IRF630?? can i use IRFZ44? if not, why not??
sorry for many questions...
regards,
raff
Workhorse said:
I steel think that is dangerous for a 20V Schottky without other precautions, but apparently is only my ignorance. If it works reliable on your commercial amplifiers is ok.
Thanks for your effort and explanations.
Anybody with experience out there could advance some comments and explanations about the reliability of Workhorse circuit components choice, please?
I ask because 20V Schottky seems to be a relatively cheap solution, but my instinct (not knowledge) tells me that is dangerous…
Classd4sure, Fredos , EVA, Pafi...?
I know that “nuestro hermano” will say that Igbt has no reverse recovery problems 😉 , but only with Mosfet it is possible to reach high frequencies without interleaved or multilevel solutions.
Besides, if 20V Schottky works, I do not understand why they are not a commercial Mosfet with this diode inside… if they produce IGBT with or without fast diode, why not
a Mosfet with serial Schotky?
These spikes due to the frewheeling diode taking some time to reach a low impedance state are very short and will cause little dissipation.
Notice that the frewheeling diode is also subject to high dissipation during these spikes and 20V is quite a high value. 5V to 15V are more realistic figures according to my measurements. Most high side gate drive circuits would never survive 20V negative spikes anyway, so I would be more concerned about that if I was you.
Concerning IGBTs, isn't a 50ns rise time from 0V to 500V enough for you? That's 10ns per each 100V, which is the voltage usually switched in MOSFET designs. I'm even willing to repeat the measurements because such a quick turn-off is hard to believe for me.
Notice that the frewheeling diode is also subject to high dissipation during these spikes and 20V is quite a high value. 5V to 15V are more realistic figures according to my measurements. Most high side gate drive circuits would never survive 20V negative spikes anyway, so I would be more concerned about that if I was you.
Concerning IGBTs, isn't a 50ns rise time from 0V to 500V enough for you? That's 10ns per each 100V, which is the voltage usually switched in MOSFET designs. I'm even willing to repeat the measurements because such a quick turn-off is hard to believe for me.
Pafi said:
Pafi!!!
My calculations tell that this core can only sotre 12.6mJ for a 30% inductance degradation, see:
Permeability is 125u and according to datasheet it drops to 70% for H=35 oersted.
H= .4*pi*N*I/Le (general H calculation with Le in cm)
Effective core length is 10.743cm, so:
35=.4*pi*(N*I)/10.743
Solving that equation, the resulting N*I product is 300.
On the other hand, energy storage is calculated as follows:
E=.5*L*I^2
or:
E=.5*AL*(N*I)^2
AL for that core is 281nH/turn^2, so:
E=.5*281e-9*(300^2)=12.6mJ
According to my calculations, only the 26uH permeability version of that core (A-087059-2) allows for 100uH 30A and 45mJ energy storage for a 30% permeability degradation with 42 turns.
One cause of difference is incremental/normal permeability, I took account normal permeability.
Actually I didn't calculate this, instead of that I relied my old experimences when I tried A-324117 in an amplifier, and at 7 turns * 90 A I didn't see high residual on output. But maybe this was only because of high carrier freq and 3 level modulation.
Sorry about this!
OK, then try that! 🙂
But there must be some reason why do they making those expensive and lossy powder cores! 🙂
Actually I didn't calculate this, instead of that I relied my old experimences when I tried A-324117 in an amplifier, and at 7 turns * 90 A I didn't see high residual on output. But maybe this was only because of high carrier freq and 3 level modulation.
Sorry about this!
OK, then try that! 🙂
But there must be some reason why do they making those expensive and lossy powder cores! 🙂
Hi Toino,
The Schottky in my circuit is STPS20150C 150V 20A and for High power applications I use 40CPQ045 45V 40A Schottky and HFA50PA60C ULTRAFAST RECOVERY & ULRTASOFT RECOVERY HexFred Trr=23nS ,600V 50A....for Freewheeling
There is no 20V I think you were mistaken.....
regards,
K a n wa r
The Schottky in my circuit is STPS20150C 150V 20A and for High power applications I use 40CPQ045 45V 40A Schottky and HFA50PA60C ULTRAFAST RECOVERY & ULRTASOFT RECOVERY HexFred Trr=23nS ,600V 50A....for Freewheeling
There is no 20V I think you were mistaken.....
regards,
K a n wa r
fredos, theres around 1.4cm between fets and op inductor, thats bad enuf to cause over/undershoot as well as ringing ?😕
I´d rather blame my crappy layout and poor wiring. When looking @ the board underside i find it very amazing that it works at all.
Still i think diodes are the best shoot, workhorse had them Drain to Source diodes in his schemo. This i have is a modified version of his circuit, halfbridge instead of fullbridge and TL494 instead of SG2524 (because they went to hide in my printer while unpacking and i dident find them until a few days later)
Actually it it more a modded version of ledmania´s circuit.
I´d rather blame my crappy layout and poor wiring. When looking @ the board underside i find it very amazing that it works at all.
Still i think diodes are the best shoot, workhorse had them Drain to Source diodes in his schemo. This i have is a modified version of his circuit, halfbridge instead of fullbridge and TL494 instead of SG2524 (because they went to hide in my printer while unpacking and i dident find them until a few days later)
Actually it it more a modded version of ledmania´s circuit.
Workhorse said:
Sorry Workhorse... I am a real (work)donkey!
My sincere apologies for this stupid mistake.
However……
Re: chris, need info about your miniUcD
Hi,
Rather than upload the image to some temporary file shack, link to where I posted it on the forum so that others may be able to follow along as well, and use the component names in your description instead of circling them.
Much of this is your basic "analog building blocks" which you ought to be able to recognize. I strongly recommend that you read the patent, which will help explain some of the parts you're looking at.
This entire circuit was a redesign to suit the parts I had at hand. They do not therefore necessarily representative of the optimal solution. In fact I found this version of the circuit to have a bit of a problem that I was not able to solve without the use of a scope, therefore I revised the drivers, and I recommend you look for that revised circuit, as I'm sure I uploaded it.
Many of your questions are answered in the reference design thread from which this came, it is there for you to read. This will answer just about everything you asked, and alot more. Take the time to read it. In doing so, you'll no doubt find one of the revised versions of this, which do actually work as intended.
You need not worry about upping the rail voltage until you've gone at least this far!
Good luck.
Chris
RX5 said:
Hi,
Rather than upload the image to some temporary file shack, link to where I posted it on the forum so that others may be able to follow along as well, and use the component names in your description instead of circling them.
Much of this is your basic "analog building blocks" which you ought to be able to recognize. I strongly recommend that you read the patent, which will help explain some of the parts you're looking at.
This entire circuit was a redesign to suit the parts I had at hand. They do not therefore necessarily representative of the optimal solution. In fact I found this version of the circuit to have a bit of a problem that I was not able to solve without the use of a scope, therefore I revised the drivers, and I recommend you look for that revised circuit, as I'm sure I uploaded it.
Many of your questions are answered in the reference design thread from which this came, it is there for you to read. This will answer just about everything you asked, and alot more. Take the time to read it. In doing so, you'll no doubt find one of the revised versions of this, which do actually work as intended.
You need not worry about upping the rail voltage until you've gone at least this far!
Good luck.
Chris
Tekko said:fredos, theres around 1.4cm between fets and op inductor, thats bad enuf to cause over/undershoot as well as ringing ?😕
I´d rather blame my crappy layout and poor wiring. When looking @ the board underside i find it very amazing that it works at all.
Still i think diodes are the best shoot, workhorse had them Drain to Source diodes in his schemo. This i have is a modified version of his circuit, halfbridge instead of fullbridge and TL494 instead of SG2524 (because they went to hide in my printer while unpacking and i dident find them until a few days later)
Actually it it more a modded version of ledmania´s circuit.
Hi Tekko
Do you recognise this one?
Please note the group position of 3x100nF capacitors and the way they are connected to the Drain mosfets for decoupling…
Looking around, is easy to found some good Pcb routing examples at this power level.
Yah but i cant see how the caps are connected.
Maybe LC could give a prototype to me for examinaton.
Maybe LC could give a prototype to me for examinaton.
Re: Re: chris, need info about your miniUcD
Hi Chris,
it was your mini UcD.... 🙂
its nice that the input IS reference to ground... i dont know if its your latest design 🙂
classd4sure said:
Hi Chris,
it was your mini UcD.... 🙂
its nice that the input IS reference to ground... i dont know if its your latest design 🙂
Are you kidding?
One cap between the + Drain and ground of top leg mosfet.
One cap between the - Source and ground of bottom leg mosfet
The other cap between + Drain and – Source.
Shortest connections possible as you see by the position of caps and mosfets on picture.
It is very easy to do in you double side Pcb
One cap between the + Drain and ground of top leg mosfet.
One cap between the - Source and ground of bottom leg mosfet
The other cap between + Drain and – Source.
Shortest connections possible as you see by the position of caps and mosfets on picture.
It is very easy to do in you double side Pcb
- Status
- Not open for further replies.