my 1st ever D-amp, WORKING!!!

[kartino]

current limiting resistor, where he put the resistor? What purpose?I'm sorry not to follow the discussion.

 

[RX5]

Originally posted by classd4sure
I've encouraged Raff to try removing the current limiting resistors and see how it works.

soon I will... 30/30 @8A is jsut waiting around the corner... still need some tweaking... Not totally content with it..

Originally posted by kartino
current limiting resistor, where he put the resistor? What purpose?I'm sorry not to follow the discussion.

its the resistor I want to insert between the rails.. when using the 30/30 8A power supply.... 15R 10W

 

[kartino]

but why?

 

[classd4sure]

To limite peak current in case deadtime is way off at those potentials and also at much higher current output.

The goal is to limite the current to allow him to test and make any necessary adjustments before trying it with them removed. Just a precautionary step.

 

[RX5]

GUYS!!

I have connected it with +/-30V .. its ok... some adjustment of deadtime... and it sounds GOOD and LOUDER!

only problem is MORE heating of some parts though..... namely the power supplying section of the Hi/Lo side mosfet drivers....

NOTE: MOSFETS still dont heat up.. maybe only a few degrees at high temp +/- 2deg C.....

to anyone who wishes to build =poor mans D-amp UcD way, give it a go!!

its worth your while

Cheers,

Raff

 

[RX5]

wow!! powered it using +/-36V and....

poooooooooooffff!!
after a few minutes of operation....
smoking diode.. (diode between TIP31C Emmiter and Hi side gate driver)



changed it to 1.5A 100V FR diode.. still smokes after a few mins... maybe 3A will do??

 

[RX5]

I dont know how much current the Hi side draws from the TIP31C voltage regulator... that the diode burns UP

whew... I think proposed AMP is good only for +/-18 w/o any failure of componenets..

dont worry, NO =other= dead parts.. just THAT diode...

I was planning maybe a higher power version of the regulator section? on an adjacent board ... i dunno.. maybe in the near future...

 

[RX5]

Anyone??

anyways, I changed that DIODE (with HER306 3A 600V 50nS ) still heats up.... anyways, Im not THAT alarmed... might as well =think= of another scheme....(and NOT play the AMP for a while,supply of +/-36V.. but +/-18V IS ok!! > )

As I recall from my past PMs, I read -someone- told me power disipation is HIGHER in the Hi side compared to the Lo side... hmmmm so I think? this heating up is pretty normal? at higher voltage rails..

anyway, back to the drawing board.. IF I could figure it out.. how to lessen power dissipation of High Side driver .....

bye for now.... this thread might be OFF the charts for weeks...

any additional info, is MOST welcome

sincerely,

Raff

 

[kartino]

I use BYV type with no problem. Powered at 35+/- DC.

Seems that your amp have problem. Check your 12V zener at high side driver. I thought that zener dead then short circuited. Remember your modification? You add more starting supply for high side that may cause the 12V zener dead.

Use 5.6K or higher for the starting supply or your zener dead again.

 

[RX5]

I use BYV type with no problem. Powered at 35+/- DC.

Seems that your amp have problem. Check your 12V zener at high side driver. I thought that zener dead then short circuited. Remember your modification? You add more starting supply for high side that may cause the 12V zener dead.

Use 5.6K or higher for the starting supply or your zener dead again.

Kartino,

shorted?? nop.. that was a different board...that board(latest) I posted, has not seen "shorts" of ANY kind only a smoking diode at my current +/-36V....

zener in high side gate driver IS OK....

regarding the 5.6K, im using 20K now...

you say BYV?? i got some in my junk.. i think I got this on a dead PC PS./... labeled BYV27-200..... going to look/download at its specs..



=======================
P.S. anyone care to explain WHY high side takes up MORE current than Low side??

 

[kartino]

Hi,

But the only cause problem for the diode is shorted. Check the other possibility. I see your diode fast enough and still under breakdown. The heat only possible by overloaded. But not make sense overloaded by normal consumtion if you use 3A even 1A fast diode.

 

[RX5]

I see what you mean.. NO there is NO short in Hi side driver... if IT had short, I should hear distortion.. or a blown Hi side MOSFET... hopefully, there is none..

I got a "hint" though.. to whats causing this.. maybe 470R/330R....

 

[kartino]

Look strange for me!

Then of your high side circuit OK, you have only one possibilty, unproper diode.

 

[RX5]

pound for pound!! BASS

HELLO ALL!!

YES!!! UcD AMP rocks!!!

=figured it all now!!!!=

but before that, a small story 1st

You all know I was having trouble with the High side mosfet driver,

diode supplying is/was heating UP!!! I really was ABOUT to give up..

go back to +/-18VDC as the power supply.... all sulky and sad after

numerous circuit part value changing/swapping, I cleaned the circuit

board using alcohol and lacquer thinner to make it REALLY

clean....getting it ready for the +/-18V... sigh...

Why not give it another try with my +/-36V... lets see if BYV27-200

heats up as fast as HER306... well, to my astonishment, IT NEVER

HEATED UP AT ALL!!! measured DC output at speaker terminal...

around 200mV.... yup still the same.... held the MOSFET heatsink..

nop... no HEAT.. it was at idle.... waited for a few MORE minutes to

really "let BURN" the diode... well, once again. it DID NOT HEAT UP!!!

what the.............

Hmmmmm I know WHAT the problem WAS....

>>>>>>>>SOLDER FLUX<<<<<<<<<<<

heck... that was it??? daMn!!! after a few headscratching and a

sleepless night!!!!

well, anyways, valuable lesson, on high frequency circuits, clean board

IS A MUST!!! whew!!!

MAN!! thank GOD I cleaned than damn board!!

now, Im playing amp at 80% max... sounds sweet.. especially the

bass!!! it POUNDS!!

now, On with the show.. Im planning to supply it with +/-48VDC @ 2A

next week.... but this supply will be MY last..

with THAT said, I am HAPPY and I can sleep nice..

final mod is the bigger heatsink for TIP31C.... it DOES heat up the ONLY part that really heats up!!

Special thanks to Chris for figuring out this circuit.... Bruno your UcD

Amp Rocks!! To Kartino, those helpfull tips... and to all the others who

have helped... and the "Development of a reference class D starting

point" thread!!! learned a great deal info there....

Cheers,
Raff

 

[classd4sure]

Good stuff.

You were correct to inrease the value of the precharge resistor, you could even go a bit higher with it. The only reason it was so low in the first place was merely for simulation purposes.

Of course, once you increase that, you have to wait a bit longer before enabling the output to oscillate, otherwise you'll end up unequal voltage on the high Vs low side driver supplies.

You always have to ensure the value of driver resistors which you select are such that they will not load the drivers down too much or the voltage will drop, causing increased heating all around.

More efficient mosfets (what are you using again?) and a series gate resistor of say 22 to 33 ohms should help to reduce some dissipation of the tip31, there should also be a resistor in series with it to limite current, I'll let you and your simulator work that out though.

Almost giving up over such a small problem is no way to keep this thing working ! Trust me I've had my sleepless nights. Though I find it difficult to believe solder flux would have been the cause of this? Strange anyway.

That resistor in the feedback loop you use to "limite small distortions" is also not the way. These should stay closely matched in value for both feedback loops in order to maximise the CMRR of the amp. Though if it works I guess it's a decent makeshift cure, perhaps the front end op amp stage would be another.

I don't think I'd push the rails to such a high voltage with this design if I were you, it's kind of pushing your luck!

Regards,
Chris

 

[RX5]

hello chris,

welcome back.....

You were correct to inrease the value of the precharge resistor, you could even go a bit higher with it. The only reason it was so low in the first place was merely for simulation purposes.

Im using 20k now....

Of course, once you increase that, you have to wait a bit longer before enabling the output to oscillate, otherwise you'll end up unequal voltage on the high Vs low side driver supplies.

so perhaps a turn ON (3secs) delay relay for speaker IS fine ? another thing, what will happen if turned OFF then ON again<during playing>? would that mean an unequal voltage? or would it just "charge along"??

You always have to ensure the value of driver resistors which you select are such that they will not load the drivers down too much or the voltage will drop, causing increased heating all around.

i tried 1k for 470/330... TIP31C dont heat much BUT the MOSFETS did.... so back to these values..

More efficient mosfets (what are you using again?) and a series gate resistor of say 22 to 33 ohms should help to reduce some dissipation of the tip31, there should also be a resistor in series with it to limite current, I'll let you and your simulator work that out though.

Im using IRF540N... ok ill try 22R.. I have used 22R with +/-18 supply.. resulted in hissing.. could you tip me of on how much current consumption the Hi side takes? (470R/330R/10R) how about Lo side?? ok ill add a series limiting resistor to both of them....

Though I find it difficult to believe solder flux would have been the cause of this? Strange anyway.

I dont know too... I desoldered all components in the high side and tested them if they are STILL ok.. well, they were... and I never changed parts either! maybe something came out right after re soldering them again.... clueless if flux would be it?? hhmmmm dunno

That resistor in the feedback loop you use to "limite small distortions" is also not the way. These should stay closely matched in value for both feedback loops in order to maximise the CMRR of the amp. Though if it works I guess it's a decent makeshift cure, perhaps the front end op amp stage would be another.

well, If I make them equal, amp gain is minimal... increasing that resistor in feedback loop increases gain, BUT "input side" should be lowered... to prevent overmodulation? (i dunno of right term??) balance.... IT does work though....

----OR-----

I could probably increase BOTH<always same value> of them slowly -until- distortion comes in... BTW, laymans term for CMRR???

I don't think I'd push the rails to such a high voltage with this design if I were you, it's kind of pushing your luck!

well, just wish me luck.... I will do those last minute mods you suggest.. then will I opt for a higher voltage supply...

Regards,
Raff

 

[BWRX]

Re: pound for pound!! BASS

Hmmmmm I know WHAT the problem WAS....
>>>>>>>>SOLDER FLUX<<<<<<<<<<<

Though I find it difficult to believe solder flux would have been the cause of this? Strange anyway.

RX5, what kind of solder and flux did you use and how old was it?

I've heard that the residue left over from certain fluxes and solders can cause microprocessor circuits to do strange things...

 

[classd4sure]

Thanks for the welcome back. Been a rough weekend. Is three seconds fine? RC time constant * 5 is what's required to charge the cap to 99.3% of the source.

The high side driver can indeed require a little bit more juice, this is due to the miller effect. Google that.. should keep you busy for a week. While you're there google CMRR too.

"i tried 1k for 470/330... TIP31C dont heat much BUT the MOSFETS did.... so back to these values.."

It is the ratio of the 470/330 that most largely dictates how much the driver is loaded, along with choice of active components, and even the series 10ohm. Use your simulator to give an idea how this all works.

Also use it to experiment with the tip31C circuit, I believe your current limiting resistor is simply most easily placed in series with the collector lead to ground, but only if the heat really bothers you. Of course this will make it alot easier for you to load your drivers down too much. You really should experiment with this on the simulator for awhile.

Now when you say 46V @ 2amps, I know for certain you dont' mean with a lab supply that's actually limited to 2 amps. Much greater levels of current will flow than you're aware, it will draw what it needs given the rail voltage.

I really don't think this will hold together very well at 46V rails, but good luck, I look forward to those results.

I looked it up and it seems flux can indeed cause some weirdness/shorts. It should always be cleaned anyway, this is just another reason to do it.

 

[RX5]

Re: Re: pound for pound!! BASS

RX5, what kind of solder and flux did you use and how old was it?

I've heard that the residue left over from certain fluxes and solders can cause microprocessor circuits to do strange things...

I dont use flux alone....combined with a soldering lead

I use 60/40 or was that 40/60?? lead/flux ratio.... hhhhmmmmmm Been missing the basic info here... hahahah

 

[RX5]

Thanks for the welcome back. Been a rough weekend. Is three seconds fine? RC time constant * 5 is what's required to charge the cap to 99.3% of the source.

what I meant was adding a RELAY switch on the speaker output, say 3 secs "off", before connecting the speaker TO the AMP

The high side driver can indeed require a little bit more juice, this is due to the miller effect. Google that.. should keep you busy for a week. While you're there google CMRR too.

wikipedia.. miller effect.. gotta read that one later... CMRR ( Common Mode Rejection Ratio) ok!!

"i tried 1k for 470/330... TIP31C dont heat much BUT the MOSFETS did.... so back to these values.."

It is the ratio of the 470/330 that most largely dictates how much the driver is loaded, along with choice of active components, and even the series 10ohm. Use your simulator to give an idea how this all works.

been doing THAT .. changed it to 1k(470/330) its works on SIM BUT heats the MOSFETs in real life.so i got to "ratio" it out at higher values.. is that correct? whats the ratio?.. used 22R, no oscillation (dead) output is positive(where output to speaker is taken)....

Also use it to experiment with the tip31C circuit, I believe your current limiting resistor is simply most easily placed in series with the collector lead to ground, but only if the heat really bothers you. Of course this will make it alot easier for you to load your drivers down too much. You really should experiment with this on the simulator for awhile.

so a "common" value for the High an Low?? hmmm ok... been using different values for the Hi side and Lo side.... but changing supply to a lower voltage meant lowering AGAIN value of said resistor/s.... ok ill try the resisitor in collector leg.. in SIM...

Now when you say 46V @ 2amps, I know for certain you dont' mean with a lab supply that's actually limited to 2 amps. Much greater levels of current will flow than you're aware, it will draw what it needs given the rail voltage.

nop.. it aint a lab power supply.... its a 24-0-24 @ 2A transformer... once again, half wave 60Hz... I cant find 48-0-48 here..

I really don't think this will hold together very well at 46V rails, but good luck, I look forward to those results.

ok.. but I still will.... might as well put on protective "eyegear" in case a transistor blows up... hehehe

I looked it up and it seems flux can indeed cause some weirdness/shorts. It should always be cleaned anyway, this is just another reason to do it.

Ive read that in a textbook.... its ok with low frequency, BUT higher frequency, is where the prob of flux show up...