Dear friends:
I have the circuit shown on the schematic enclosed. I have already fixed serious problems concerning the output mosfets and driver, which have been replaced. On the first test the amp worked fine but after some minutes the driver was muted by the protection circuitry, I first thought of a thermal issue but this was soon discarded as I began to detect where the origin was. I have not been able to understand the design and I am not very familiar to class D protection designs. What I have found is that this activation happens randomly and it could be in just 45 seconds of use or 45 minutes. Also I have done some measuring on normal condition and in protected condition but this it's been only useful to create more concern and doubts. In brief I can tell that Vbe in Q16 (in normal condition) fluctuates from 10 to 200 mv depending on the output signal (swinging with it) but when the problem happens the Vbe=0. and I can't see how this can be according to the surrounding circuits. I mean, this is a design that is working but I need to understand how.
If any additional information is required to go further away please do not hesitate to post.
Thanks for your time.
I have the circuit shown on the schematic enclosed. I have already fixed serious problems concerning the output mosfets and driver, which have been replaced. On the first test the amp worked fine but after some minutes the driver was muted by the protection circuitry, I first thought of a thermal issue but this was soon discarded as I began to detect where the origin was. I have not been able to understand the design and I am not very familiar to class D protection designs. What I have found is that this activation happens randomly and it could be in just 45 seconds of use or 45 minutes. Also I have done some measuring on normal condition and in protected condition but this it's been only useful to create more concern and doubts. In brief I can tell that Vbe in Q16 (in normal condition) fluctuates from 10 to 200 mv depending on the output signal (swinging with it) but when the problem happens the Vbe=0. and I can't see how this can be according to the surrounding circuits. I mean, this is a design that is working but I need to understand how.
If any additional information is required to go further away please do not hesitate to post.
Thanks for your time.
Well IMOP the thing is rather badly designed as it takes no account of the MOSFET turn on current spike that will occur as a result of reverse recovery in the other mosfets drain source diode and capacitance in the output inductor, the magnitude of recovery current has many causes including dv/dt, the amount of current it was conducting and temperature, given a music signal it will be all over the place!! The normal way this is dealt with is to have a time gate, what we call a blanking period around mosfet turn on to disable the overcurrent sensing at this critical time, the other issue is shoot through caused by insufficient deadtime (both mosfets conducting simultaneously). In summery I am not surprised at your problem and my first port of call would be the vendor for my money back!! sorry to sound unhelpful but its hard to fix a poor design!
Thanks for your interest and (to me) deep review. The funny thing is that some hundred, who knows if thousand of these system have been sold and are in service under a famous british company's name (unfortunately the only thing that remains british is the name, as the company was sold to the chinese). Despite this I am not saying you are not right, instead, even if my knowledge of class D amps is limited, I realize this is a quite simple circuit because I am able to understand almost everything. The only thing I miss is a proper understanding of the Q16 function because I think it is an overcurrent detector but I can't figure how in this specific problem the voltage between Vbe gets to zero (forget about open resistors R64 or R61 tested under both conditions: work and fail).
Finally, to give you an idea of my experience I'll tell you I have been servicing professional audio for more than 25 years. Unfortunately all the formation after my career was finished it's been self-formation because where I live it is impossible to attend upgrade courses. On the last years internet has been my everything and the only thing I really miss is time, like almost everyone these days.
Thanks again for yours, I mean time.
Finally, to give you an idea of my experience I'll tell you I have been servicing professional audio for more than 25 years. Unfortunately all the formation after my career was finished it's been self-formation because where I live it is impossible to attend upgrade courses. On the last years internet has been my everything and the only thing I really miss is time, like almost everyone these days.
Thanks again for yours, I mean time.
Thanks for your compliments 🙂 you are correct both Q16 and Q20 are current detectors and a very similar arrangement is often found in linear class A/B amplifiers. Sticking to Q16, it will not conduct any collector current till its VBE reaches about 650mV (this is temperature dependent) it is measuring the voltage across R62/63 however a divide by two is formed by R61/R64 so Q16 will start to conduct when the voltage across R62/63 reaches ~1300mV and this corresponds to a Q17 drain current of 26amps. In a class A/B C55 would have little effect, however in a switching class D it serves to average the signal. In a class D the current waveform is triangular as the inductor current is either discharged or charged, so in an ideal world the current detector meaures the peak current at the end of the conduction period, if you imagine the current starting at zero each time the switch turns on and rising to its peak at the end of conduction this is the easiest way to understand it. Unfortunately the world is far from perfect and it is normal for a large current spike to accompany switch on as a result of cross conduction and diode reverse recovery, this will partly charge C55 such that the over current circuit can be falsely triggered by a lower than intended peak current. As I said before there can be numerous reasons for this however given this is a Chinese product it is probably because the original design has not been adhered to, most likely the mosfets are not the originals, the output inductor has a crap core and is saturating and the deadtime is wrong, they may be good at manufacturing but as for understanding what the design constraints are, clueless!!
OMG as to how to fix it ewwwwwww better to chuck it maybe, let me know if thats impossible 🙂
OMG as to how to fix it ewwwwwww better to chuck it maybe, let me know if thats impossible 🙂
Thank you Master.
It is always an unbeatable experience finding people like you with the will to help others. One final thing. Can you give me any advice on a book or link where I can go any further on this specific subject. I have many documents but some of them are too simple as to understand problems like the one I have or, on the other side, too much scientific as to get any practical application. One of my favourites expressions and sayings is: Not everything is black or white, the grey colour exists"
Again, thank you and I hope I don't bother you with my insistence.
Juan
It is always an unbeatable experience finding people like you with the will to help others. One final thing. Can you give me any advice on a book or link where I can go any further on this specific subject. I have many documents but some of them are too simple as to understand problems like the one I have or, on the other side, too much scientific as to get any practical application. One of my favourites expressions and sayings is: Not everything is black or white, the grey colour exists"
Again, thank you and I hope I don't bother you with my insistence.
Juan
Thats a big ask Juan I cant do a brain dump hahha I did find this class D app note from IR discusses some of the issues http://www.irf.com/technical-info/appnotes/an-1071.pdf
A class D amplifier isnt that different from a switchmode DC-DC converter so if you try looking for stuff on buck converters also that will also show you some of the issues and you may need to get into a magnetcs a bit too 🙂 I dont know if you have a scope but you could do with being able to see the inductor current waveform to find out if its saturating, good luck 🙂
John
A class D amplifier isnt that different from a switchmode DC-DC converter so if you try looking for stuff on buck converters also that will also show you some of the issues and you may need to get into a magnetcs a bit too 🙂 I dont know if you have a scope but you could do with being able to see the inductor current waveform to find out if its saturating, good luck 🙂
John
OK, good homework for the weekend. About instrumentation, yes I have , well perhaps more than I really need (I have four oscilloscopes plus many other instruments), I'll give some more time to the amp and do more theoretical readings, I mean, DC readings without signal and then other readings with a fixed frequency signal to measure other values. I am no going to be very "heavy" but as soon as I have a conclusion, or maybe a solution, I'll let you know.
Again, Thank you.
Again, Thank you.
Dear Johnsnell:
Thanks for your support and time. I don't like introducing mods on designs that are on the market, even if I know that most times are a necessity rather than a caprice, and it is like a religion for some people to do this alterations. My main reason is: if it works why should I change? and if it doesn't work, there must be some fault on components, tracks or maybe solderings ( I'm not going to get the Nobel for this discovering, I know).
But as a courtesy to you, I'll give a try and see what happens. If this is successful I'll let you know.
Again, thank you
Juan
Thanks for your support and time. I don't like introducing mods on designs that are on the market, even if I know that most times are a necessity rather than a caprice, and it is like a religion for some people to do this alterations. My main reason is: if it works why should I change? and if it doesn't work, there must be some fault on components, tracks or maybe solderings ( I'm not going to get the Nobel for this discovering, I know).
But as a courtesy to you, I'll give a try and see what happens. If this is successful I'll let you know.
Again, thank you
Juan
- Status
- Not open for further replies.