Re: Re: Re: Re: Re: Re: The latter one
IRF460? that's a high voltage component with heavy gate capatance (450 is lighter).
I would rather use 530/9530 -- something often used for robot motor drivers, they are rugged enough to switch 24V. I have several ones of them, also have 540/9540 but they are twice "heavyer".
IRF460? that's a high voltage component with heavy gate capatance (450 is lighter).
I would rather use 530/9530 -- something often used for robot motor drivers, they are rugged enough to switch 24V. I have several ones of them, also have 540/9540 but they are twice "heavyer".
classd4sure said:
Re: Re: Re: Re: Re: Re: Re: The latter one
Yeap that's my point exactly it is a heavy one to switch and works just fine (with gate current in the mA range), they did three quarters of my troubleshooting for me, but I'm not saying design your amp to use them. It doesn't take anywhere near the current you think it does to drive them.
Now I have FDP3672 in their place, a very efficient device, and thanks to having used the IRF460's to get most tweaking done, I have yet to blow a set of the FDP3672.
Kenshin said:
Yeap that's my point exactly it is a heavy one to switch and works just fine (with gate current in the mA range), they did three quarters of my troubleshooting for me, but I'm not saying design your amp to use them. It doesn't take anywhere near the current you think it does to drive them.
Now I have FDP3672 in their place, a very efficient device, and thanks to having used the IRF460's to get most tweaking done, I have yet to blow a set of the FDP3672.
BTW Kenshine the IRF460's I had used for troubleshooting stages of my amp actually came with the heatsinks that I still have in there, cut from the old ACER monitor. Don't neglect old scrap for parts they can save you a bundle and get you back in operation in a pinch.
I used to live in a small retirement town where all I had was radio shack and the staff broke out into a sweat everytime I walked over to their parts shelf. Scavenging is a great way to get parts and there's so much old junk around it's not hard to come by. Here we have "clean up week" where in the spring everyone puts their junk on the curb and everyone else goes shopping through it. I know it sounds funny but alot of people get their furniture that way being this is a third world country I'm in.
Anyway, old computers (good for their SMPS's which just rock for class D amp parts, 105 deg caps, MOSFETs, small signal BJT's, varistors, all kinds of heat sinks, I've never found a coil I liked in one but you can use what's there just for kicks.)
The monitors also have some of that along with nicer mosfets, oddly the SMPS's I've found usually didn't have mosfets but power BJT's. Both are excellent sources of small heatsinks which are perfect for DIY homebrew experiments. You can even scavenge such things as the heatsink isolators and mounting hardware.
When it comes to higher end mosfets samples were an alternative for me. I don't think I"ve abused the system, and told them exactly what it was for both times. They even authorised shipping the same sample of one I'd already tried but ran out of, to better evaluate them with two other kinds I wanted to try as well.
A small handful of highly researched high end selections will last a long time if you use cheap and easy to find industrial type devices for the initial power up/tweaking stages, and precisely because of their greater parasitics slowing them down more, they're more immune to the process (take alot more of a beating before they cook, (this can also make redundant the need of snubbers around them in certain cases), even if they are running just at the point of meltdown for weeks because of bad cross conduction or whatever.
Once you have the circuit half tamed you can swap to more efficient mosfets and the difference is from heatsinks that are medium hot/very warm to "is this thing on?" cold.
No kidding I used from 10 to 100 ohms gate resistors, with a mix of (at the same time.. low stock) turn on and turn off driver BJTs.
It matters little how you how mix them provided they're wisely selected and similar in nature.
I've mixed the mosfets up too, that was worth a good extra 8 hours of play time last new years.
Finally left it with 100 ohm gate resistors, it provides the coolest mosfet, lowest EMI which is highly noticeable. The heat for a lower turn on resistor (cross conduction wise) can be easily tolerated with a certain limite of course, but the EMI from such a fast switch is unbearable with a homebrew /breadboard construction. Most of the extra heat of the industrial part is from the reduced slew rate allowing more time between states, but like I said before they seem love the abuse.
The 100 ohm turn on resistor is small enough to still switch it easily with such an efficient mosfet, and the IRF460's switch inefficiently but still switch.
Such a weak drive is also paramount in managing generated EMI to make the sound enjoyable. There's posts around here where someone had their own homebrew and was complaining of all the parts around his mosfets heating up inductively (ever heard of induction heaters?), .... you need to do what you can and if that means 100 ohm turn on resistors so what.
It also goes to further speed turn off as they're hardly on to begin with, good engineering slams them on and overdrives them by a good margin, I haven't found that part of the book applies.
Since you know about saturation and all that now you see one reason why MOSFETs are a preferred class d output device.
I wanted a circuit that used a dual N channel output stage because without question it is superior in function, and wanted to learn how it was done.
It's not easy with respect to only the learning curve, that takes work, and alot of it. However it's surprising to see how easy it is and how few components you need. Freedom from the requirement of the perfect and non existent driver IC is great.
Now I wouldn't waste my time with a P channel at all.
I used to live in a small retirement town where all I had was radio shack and the staff broke out into a sweat everytime I walked over to their parts shelf. Scavenging is a great way to get parts and there's so much old junk around it's not hard to come by. Here we have "clean up week" where in the spring everyone puts their junk on the curb and everyone else goes shopping through it. I know it sounds funny but alot of people get their furniture that way being this is a third world country I'm in.
Anyway, old computers (good for their SMPS's which just rock for class D amp parts, 105 deg caps, MOSFETs, small signal BJT's, varistors, all kinds of heat sinks, I've never found a coil I liked in one but you can use what's there just for kicks.)
The monitors also have some of that along with nicer mosfets, oddly the SMPS's I've found usually didn't have mosfets but power BJT's. Both are excellent sources of small heatsinks which are perfect for DIY homebrew experiments. You can even scavenge such things as the heatsink isolators and mounting hardware.
When it comes to higher end mosfets samples were an alternative for me. I don't think I"ve abused the system, and told them exactly what it was for both times. They even authorised shipping the same sample of one I'd already tried but ran out of, to better evaluate them with two other kinds I wanted to try as well.
A small handful of highly researched high end selections will last a long time if you use cheap and easy to find industrial type devices for the initial power up/tweaking stages, and precisely because of their greater parasitics slowing them down more, they're more immune to the process (take alot more of a beating before they cook, (this can also make redundant the need of snubbers around them in certain cases), even if they are running just at the point of meltdown for weeks because of bad cross conduction or whatever.
Once you have the circuit half tamed you can swap to more efficient mosfets and the difference is from heatsinks that are medium hot/very warm to "is this thing on?" cold.
No kidding I used from 10 to 100 ohms gate resistors, with a mix of (at the same time.. low stock) turn on and turn off driver BJTs.
It matters little how you how mix them provided they're wisely selected and similar in nature.
I've mixed the mosfets up too, that was worth a good extra 8 hours of play time last new years.
Finally left it with 100 ohm gate resistors, it provides the coolest mosfet, lowest EMI which is highly noticeable. The heat for a lower turn on resistor (cross conduction wise) can be easily tolerated with a certain limite of course, but the EMI from such a fast switch is unbearable with a homebrew /breadboard construction. Most of the extra heat of the industrial part is from the reduced slew rate allowing more time between states, but like I said before they seem love the abuse.
The 100 ohm turn on resistor is small enough to still switch it easily with such an efficient mosfet, and the IRF460's switch inefficiently but still switch.
Such a weak drive is also paramount in managing generated EMI to make the sound enjoyable. There's posts around here where someone had their own homebrew and was complaining of all the parts around his mosfets heating up inductively (ever heard of induction heaters?), .... you need to do what you can and if that means 100 ohm turn on resistors so what.
It also goes to further speed turn off as they're hardly on to begin with, good engineering slams them on and overdrives them by a good margin, I haven't found that part of the book applies.
Since you know about saturation and all that now you see one reason why MOSFETs are a preferred class d output device.
I wanted a circuit that used a dual N channel output stage because without question it is superior in function, and wanted to learn how it was done.
It's not easy with respect to only the learning curve, that takes work, and alot of it. However it's surprising to see how easy it is and how few components you need. Freedom from the requirement of the perfect and non existent driver IC is great.
Now I wouldn't waste my time with a P channel at all.
Too right, one room in my house is stacked full of old equipment that people wanted to throw, there must be a fortune in parts hiding amongst the junk.
By the way don't worry about using this thread for the UcD stuff, it's making quite interesting reading this end. I've been slowed down a bit by a small flood in my house but shall get back up to speed soon, are there any schematics of the design you are both talking about? I'd like to see what you guys are talking about as i still have problems with the self oscillating controll/modulation at this end of my new design... Maybe it would help me understand were i'm going wrong. P.S do you know any good suppliers for amp cases?
Regards
Mad.P
Yeah, IBM 
For stuff like this you can't beat a mid tower.
Here's my work.
http://www.diyaudio.com/forums/showthread.php?threadid=55385
For stuff like this you can't beat a mid tower.
Here's my work.
http://www.diyaudio.com/forums/showthread.php?threadid=55385
Hi everyone,
After much experimentation and carefull thought i have decided the best course for me is to stay with my origanal design and to concentrate all my efforts on that. I built a mock-up scaled down version of an oscillating amplifier and found it to be more guess work and luck than design due to the unstable and often uncalcalculable nature of this type of oscillator. Speed is a good thing for class D amps in moderation but when you start hitting a few MHz it becomes your worst enemy. Also speed is very variable and i found if i slowed the amp to a reasonable rate it slowed down more in use, slow enough to be heard. When i got a satifactory working model of an amp i gave it a listen but it seemed to lack brilliance and depth. Also unavoidable noises in the music were being genarated, almost inaudiable but there. Since i come from a digital background i think its best i stay with my first design as it is perfectly logical to me and i can't see any flaws with the principal. On that note i am cleaning up a few issues before i go higher in power. One isssue has only just raised its ugly head and i would appreciate everyones opinion on whats causing it. Last weekend me an my mate came back to my place from the pub ( as usual ). We were fairly smashed and i put some music on quite low on my carrier amp but instead of using the small 6 ohm test speaker i usally have hooked up i put a huge 8 ohm speaker on the amp. This speaker is quite large and has a crossover inside with tweet/mid/sub cones. The more we got drunk the more the music volume crept up. At about 70% max there was a funney noise a bit like clipping but sounding more raspy than a clip. It came and went, seemingly triggered by musical notes. The noises were not there when the amp was maxed out though. I was in no state to investigate that night so as a bad attempt at a cure i wired both speakers to the amp, 6 and 8 ohm rammed into the output terminals ( i was drunk )
I slowly increased the volume and to my amazement i had achived absolutly nothing. The nois was still there at the same volume. So i tried hammering the amp for a while thinking it would maybe 'warm up' when maxed out and overloaded with the two speakers ( i was very drunk ). Obviously i achived nothing and the amp doesn't get warm anyway so i admitted defeat and left the amp maxed out ( to avoid strange noises
)
After thinking this through i think it must be either a lack of capacitance on the supply allowing bus pumping to affect the amp, or the post feedback is causing trouble due to the large inductance of my big speaker/crossover. The noise is raspy clippy a bit like a trumpet but not as sharp as normal clipping. Have any of you ever experienced anything like this before??
Take care all
Mad.P
After much experimentation and carefull thought i have decided the best course for me is to stay with my origanal design and to concentrate all my efforts on that. I built a mock-up scaled down version of an oscillating amplifier and found it to be more guess work and luck than design due to the unstable and often uncalcalculable nature of this type of oscillator. Speed is a good thing for class D amps in moderation but when you start hitting a few MHz it becomes your worst enemy. Also speed is very variable and i found if i slowed the amp to a reasonable rate it slowed down more in use, slow enough to be heard. When i got a satifactory working model of an amp i gave it a listen but it seemed to lack brilliance and depth. Also unavoidable noises in the music were being genarated, almost inaudiable but there. Since i come from a digital background i think its best i stay with my first design as it is perfectly logical to me and i can't see any flaws with the principal. On that note i am cleaning up a few issues before i go higher in power. One isssue has only just raised its ugly head and i would appreciate everyones opinion on whats causing it. Last weekend me an my mate came back to my place from the pub ( as usual ). We were fairly smashed and i put some music on quite low on my carrier amp but instead of using the small 6 ohm test speaker i usally have hooked up i put a huge 8 ohm speaker on the amp. This speaker is quite large and has a crossover inside with tweet/mid/sub cones. The more we got drunk the more the music volume crept up. At about 70% max there was a funney noise a bit like clipping but sounding more raspy than a clip. It came and went, seemingly triggered by musical notes. The noises were not there when the amp was maxed out though. I was in no state to investigate that night so as a bad attempt at a cure i wired both speakers to the amp, 6 and 8 ohm rammed into the output terminals ( i was drunk )
I slowly increased the volume and to my amazement i had achived absolutly nothing. The nois was still there at the same volume. So i tried hammering the amp for a while thinking it would maybe 'warm up' when maxed out and overloaded with the two speakers ( i was very drunk ). Obviously i achived nothing and the amp doesn't get warm anyway so i admitted defeat and left the amp maxed out ( to avoid strange noises ) After thinking this through i think it must be either a lack of capacitance on the supply allowing bus pumping to affect the amp, or the post feedback is causing trouble due to the large inductance of my big speaker/crossover. The noise is raspy clippy a bit like a trumpet but not as sharp as normal clipping. Have any of you ever experienced anything like this before??
Take care all
Mad.P
nitrate said:Does anyone think there is any point in adding a snubber network across the power switch stages? I mean what advantage if any is it going to give me sonicly? What can a little ringing do to my music
Mad.P
Hmmmm, start with a reduction in EMI and go from there? Made a big difference on my amp anyway. It'll cost you a few caps and maybe a resistor or two to try it
Hi Nitrate,
I ever got same problem at firts trial with my UCD. I find two problem.
First gate mosfet turn on - off time was not good designed. Then I fix the gate driver. I see you use MAX, since I never try it, I have no idea.
Second I find that the L/C output filter was not good, so it cause the sinewave is not good formed. Then I find a paper, how to create air core. I created a about 0.5mm2 wire, wound about 70 turns with dia. about 3cm. It worked fine with a non polar 2uf 250V that people use it for tweeter crossover.
But actually now I use an ex. PC smps core with about 50 turns. Still passing a very small heared weird sound.
I have experienced also that the output before filtered is sensitive. When I held the area then the sound quality degraded. It may need ground plane around the area.
You can try to redesign your filter.
Good luck
kartino
I ever got same problem at firts trial with my UCD. I find two problem.
First gate mosfet turn on - off time was not good designed. Then I fix the gate driver. I see you use MAX, since I never try it, I have no idea.
Second I find that the L/C output filter was not good, so it cause the sinewave is not good formed. Then I find a paper, how to create air core. I created a about 0.5mm2 wire, wound about 70 turns with dia. about 3cm. It worked fine with a non polar 2uf 250V that people use it for tweeter crossover.
But actually now I use an ex. PC smps core with about 50 turns. Still passing a very small heared weird sound.
I have experienced also that the output before filtered is sensitive. When I held the area then the sound quality degraded. It may need ground plane around the area.
You can try to redesign your filter.
Good luck
kartino
Yeh man,
I've been proding away with the output filter latley as i've got my hands on a decent sig-gen from a freind and a dummy load. I've found that i had too much inductance causing reduction of signal above 15K and also not enough capacitance causing a seriously messed up wavform at low freq's above 35Watt and my high freq sines were becoming triangle waves
So i've been scrounging high voltage non polerised caps from old pcb's at work ( not easy to find i must say ) and i've reduced the inductance to a now known value of 24uH ( still concerned about the heat ). I'll try the caps when i get home. The large inductance allowed me to use plenty of post filter feedback but now i'm struggling to apply hardly any without going into osc with this smaller inductor. I hope the extra caps i'm taking home will help. Caps in the feedback line to filter the feedback signal help me to apply more feedback but i don't wanna use them due to these extra caps causing a non-linear freq response.
I think i need to find a better feedback technique. It seems that the more i make this amp near-perfect the more i destroy the beautiful sound i started with
Keep up the good work all
Mad.P
I've been proding away with the output filter latley as i've got my hands on a decent sig-gen from a freind and a dummy load. I've found that i had too much inductance causing reduction of signal above 15K and also not enough capacitance causing a seriously messed up wavform at low freq's above 35Watt and my high freq sines were becoming triangle waves
So i've been scrounging high voltage non polerised caps from old pcb's at work ( not easy to find i must say ) and i've reduced the inductance to a now known value of 24uH ( still concerned about the heat ). I'll try the caps when i get home. The large inductance allowed me to use plenty of post filter feedback but now i'm struggling to apply hardly any without going into osc with this smaller inductor. I hope the extra caps i'm taking home will help. Caps in the feedback line to filter the feedback signal help me to apply more feedback but i don't wanna use them due to these extra caps causing a non-linear freq response.
I think i need to find a better feedback technique. It seems that the more i make this amp near-perfect the more i destroy the beautiful sound i started with
Keep up the good work all
Mad.P
Switchmode addiction
Hi Nitrate.
I really liked your original breadboard arrangement. It reminds me of the first digital-modulated open loop switching amp I built on breadboard in 1995. Except mine was flimsier and didn't have the 4 FETs on a heatsink, so it ended up with an MTBF of about 60 minutes! It lasted just long enough to get through the Rolling Stones CD 'More Hot Rocks' and I was ever so impressed with the way it sounded.
It is far too late for you now. You've been bitten by the Switching Amp Addiction and probably the rest of your life will be ruined!
When it gets too bad you can design and build a few switchmode PSUs for a bit of light relief and fun.
Good work
John
Hi Nitrate.
I really liked your original breadboard arrangement. It reminds me of the first digital-modulated open loop switching amp I built on breadboard in 1995. Except mine was flimsier and didn't have the 4 FETs on a heatsink, so it ended up with an MTBF of about 60 minutes! It lasted just long enough to get through the Rolling Stones CD 'More Hot Rocks' and I was ever so impressed with the way it sounded.
It is far too late for you now. You've been bitten by the Switching Amp Addiction and probably the rest of your life will be ruined!
When it gets too bad you can design and build a few switchmode PSUs for a bit of light relief and fun.
Good work
John
LOL,
Cheers Jhon. Glad you approve of my dodgey design techniques
Mad.P
P.S all,
I'm not trying to create some kinda ghostly image of myself by signing 'Mad.P'. It's just my nick name i've had all my life. It is short for Mad Professor. I also used this nick for many years as a C.B handle. My freinds named me Mad Professor when i was about 10 because i was always messing with the old electronics.
Regards
Leigh Pilkington
Cheers Jhon. Glad you approve of my dodgey design techniques
Mad.P
P.S all,
I'm not trying to create some kinda ghostly image of myself by signing 'Mad.P'. It's just my nick name i've had all my life. It is short for Mad Professor. I also used this nick for many years as a C.B handle. My freinds named me Mad Professor when i was about 10 because i was always messing with the old electronics.
Regards
Leigh Pilkington
Hi Leigh,
Almost sounds like you're already missing the robustness of the self oscillating amp, but maybe dont' realize it yet?
The hash you spoke of could possibly be due to EMI interfering with your carrier at some frequency /power where the Q of some resonance is just right.
Better designs won't suffer from frequency modulation much, and may in fact benefit from it. Like, what's wrong with slowing down Fs at full output power, providing it doesn't become audible? Set Fs to 450k or 500kHz, if it should slow to 400k or even 350 I dont' think you'll notice a loss of quality. You gave up on that too fast I think.
Regards,
Chris
Almost sounds like you're already missing the robustness of the self oscillating amp, but maybe dont' realize it yet?
The hash you spoke of could possibly be due to EMI interfering with your carrier at some frequency /power where the Q of some resonance is just right.
Better designs won't suffer from frequency modulation much, and may in fact benefit from it. Like, what's wrong with slowing down Fs at full output power, providing it doesn't become audible? Set Fs to 450k or 500kHz, if it should slow to 400k or even 350 I dont' think you'll notice a loss of quality. You gave up on that too fast I think.
Regards,
Chris
Hi all,
Chris, i havn't totally given up on the self oscillating design yet, it still sits on its own breadboard under my bench. By suggesting that i set the max osc freq to around 500K tells me that i definatly have been running the amp far too fast, this will explain the funny artifacts and unstabilites i found present in the trial amp. I was running at around 2 to 3MHz and the frequency modulated quite alot. A 25Hz test signal slowed it down to under 1MHz. Anyway i'm gonna carry on and finish all the main issues of my carrier amp for now. I like this amp better, don't know why... perhaps it's because its my first classD
I certaintly won't give up on the self osc design though.... if anything it will spurr me on as i won't be able to leave it alone until it works. I must admit analouge electronics is not my strong point and i always avoid getting analouggy were ever i can. As they say 'Never run before you can walk' well in this arena i'm still crawling so i'll stick with the carrier amp for now as i understand whats going on properly but be sure as soon as i have mastered that amp i'll be re-awakening the self oscillator! I'm never satisfied until i've achived my goal, just can't leave these things
Weekend is here, time for a beer! I hope my amp survives this weekend LOL
Mad.P
Chris, i havn't totally given up on the self oscillating design yet, it still sits on its own breadboard under my bench. By suggesting that i set the max osc freq to around 500K tells me that i definatly have been running the amp far too fast, this will explain the funny artifacts and unstabilites i found present in the trial amp. I was running at around 2 to 3MHz and the frequency modulated quite alot. A 25Hz test signal slowed it down to under 1MHz. Anyway i'm gonna carry on and finish all the main issues of my carrier amp for now. I like this amp better, don't know why... perhaps it's because its my first classD
I certaintly won't give up on the self osc design though.... if anything it will spurr me on as i won't be able to leave it alone until it works. I must admit analouge electronics is not my strong point and i always avoid getting analouggy were ever i can. As they say 'Never run before you can walk' well in this arena i'm still crawling so i'll stick with the carrier amp for now as i understand whats going on properly but be sure as soon as i have mastered that amp i'll be re-awakening the self oscillator! I'm never satisfied until i've achived my goal, just can't leave these things
Weekend is here, time for a beer! I hope my amp survives this weekend LOL
Mad.P
Am i going mad?
Hey all,
Am i going mad... Just a quickie...
As you can see from my early schematics i feed the amp with +-35V. One would expect to see a brige voltage of 70V Pk-Pk. On the scope i do. Very nice it is too, nice and square ( rectangular ) with nice sharp edges. So why is it i can only get 40V PK-PK sine wave output before serious clipping?
This is under any load ( 8 to 2 ohm ), any frequency ( DC to 30KHz ), and any output filter configureation i try ( iv'e modded the filer to cut at 30KHz ish and give a flat freq response by using 1uF cap along with 24uH inductance ). Is this normal, do i have to rms the rms value or somthing? Can't be rms of 70V cos that comes out at 50V ish... I'm confused.... Please help on this befor i go do somthing like boost the supply volage up to 80 volts in an attempt to regain my apparent lost power ( maybe it's a you can't read the power on the scope thing.. )
If the scope is to believed i'm only getting 40V PK-
pk/2=20, 20/1.42=14.08RMS, 14.08*14.08=198.37, 198.37/4Ohm load = 49.59 Watts RMS!!
Can't be true can it
I've included the calcs cos it's supprising how many people don't know how to work out true power ratings of amps.
Whats going wrong.. if anything.. it may be true... Oh no... maybe i've seriously underestimated the power of my amp.... Nah, can't have... I'm confused
Regards
Mad.P
P.S if my calcs are inaccurate i appologise, before you follow my example i urge you to check the next few post's as i imagine they will take great plesure in informing evryone of any errors ( I'm well known for errors like this )
Hey all,
Am i going mad... Just a quickie...
As you can see from my early schematics i feed the amp with +-35V. One would expect to see a brige voltage of 70V Pk-Pk. On the scope i do. Very nice it is too, nice and square ( rectangular ) with nice sharp edges. So why is it i can only get 40V PK-PK sine wave output before serious clipping?
This is under any load ( 8 to 2 ohm ), any frequency ( DC to 30KHz ), and any output filter configureation i try ( iv'e modded the filer to cut at 30KHz ish and give a flat freq response by using 1uF cap along with 24uH inductance ). Is this normal, do i have to rms the rms value or somthing? Can't be rms of 70V cos that comes out at 50V ish... I'm confused.... Please help on this befor i go do somthing like boost the supply volage up to 80 volts in an attempt to regain my apparent lost power ( maybe it's a you can't read the power on the scope thing.. )
If the scope is to believed i'm only getting 40V PK-
pk/2=20, 20/1.42=14.08RMS, 14.08*14.08=198.37, 198.37/4Ohm load = 49.59 Watts RMS!!
Can't be true can it
I've included the calcs cos it's supprising how many people don't know how to work out true power ratings of amps.
Whats going wrong.. if anything.. it may be true... Oh no... maybe i've seriously underestimated the power of my amp.... Nah, can't have... I'm confused
Regards
Mad.P
P.S if my calcs are inaccurate i appologise, before you follow my example i urge you to check the next few post's as i imagine they will take great plesure in informing evryone of any errors
( I'm well known for errors like this )
Hmmm,
The switching waveform is fine right upto about 98% then the p channel starts to jib a little in terms of being a little slow at saturating ( A slight bend appears at the top of the switch on were normally it is a lot sharper ). This is not ideal but i don't think it accounts for a 40% loss of power. The Switching wave is switching hard onto the rails. The driver outputs perfect pwm all the time, and after the cap coupling the waveform on the fet gates is still sharp. Do you think it may be my dead time? I have it set rarther high at the mo to avoid me blowing anything. It is currently 100n/s. Maybe the inductor is causing a lot of loss, afterall it does get hot and all that power loss in heat has to come from somwhere.
On a different note, i tried my amp with a sig gen to sweep accross the freq form DC to 20K. At first i found the amp started to tail off around 12.5K and was only half the gain at 20K. This was considered unacceptable but at the time the amp sounded graet as i had a lot of post filter feedback that totally silenced the amp, with reduced gain. I experimented with the filter and reduced the size of the inductor. This increased the HF ripple so i slapped a bigger cap on the output as well to keep this under control. Now the freq response it totally flat from DC to 35K not that im bothered about anything abouve 20K. Gain is up too. This would all normally be a good achivement however it comes at an unacceptable price. The price is i can no longer apply anything like affective amounts of feedback. Every time i try to close the loop either pre or post filter the amp oscillates unless it is a uslessly small amount. If i close the loop with caps in the loop i can get more feedback but nothing like the amount i had before, also the freq response is badly affected by having caps in the loop. I can't seem to get the balance between freq response and feedback right. I know the trouble must be with the inductor altering the phase of the signal at different frequencys ( High frequencys must be really alterd i think ). How do i work out the phase/freqency relationship for the filter or inductor? There must be a collection of formulas around for doing just this??
Hope you can help,
Mad.P
P.S Anyone think it's a sin to use optos on the output stage? Just seems an easy way of achiving level shift.
The switching waveform is fine right upto about 98% then the p channel starts to jib a little in terms of being a little slow at saturating ( A slight bend appears at the top of the switch on were normally it is a lot sharper ). This is not ideal but i don't think it accounts for a 40% loss of power. The Switching wave is switching hard onto the rails. The driver outputs perfect pwm all the time, and after the cap coupling the waveform on the fet gates is still sharp. Do you think it may be my dead time? I have it set rarther high at the mo to avoid me blowing anything. It is currently 100n/s. Maybe the inductor is causing a lot of loss, afterall it does get hot and all that power loss in heat has to come from somwhere.
On a different note, i tried my amp with a sig gen to sweep accross the freq form DC to 20K. At first i found the amp started to tail off around 12.5K and was only half the gain at 20K. This was considered unacceptable but at the time the amp sounded graet as i had a lot of post filter feedback that totally silenced the amp, with reduced gain. I experimented with the filter and reduced the size of the inductor. This increased the HF ripple so i slapped a bigger cap on the output as well to keep this under control. Now the freq response it totally flat from DC to 35K not that im bothered about anything abouve 20K. Gain is up too. This would all normally be a good achivement however it comes at an unacceptable price. The price is i can no longer apply anything like affective amounts of feedback. Every time i try to close the loop either pre or post filter the amp oscillates unless it is a uslessly small amount. If i close the loop with caps in the loop i can get more feedback but nothing like the amount i had before, also the freq response is badly affected by having caps in the loop. I can't seem to get the balance between freq response and feedback right. I know the trouble must be with the inductor altering the phase of the signal at different frequencys ( High frequencys must be really alterd i think ). How do i work out the phase/freqency relationship for the filter or inductor? There must be a collection of formulas around for doing just this??
Hope you can help,
Mad.P
P.S Anyone think it's a sin to use optos on the output stage? Just seems an easy way of achiving level shift.
It is not a sin to use opto-couplers for that purpose but it can be regarded as some sort of waste. Opto-couplers that are fast enough are still more expensive than many technically superior solutions.
Like:
http://www.nve.com/il700.htm
Regards
Charles
Like:
http://www.nve.com/il700.htm
Regards
Charles
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.