What i am thinking is that as RDS increases current through the fet will decrease slightly but there will be a lot more voltage dropped accross the fet so even though current throughput goes down the fet sees more power accross it and so has to dissipate it as heat. I think that is why i am seeing a very slight case of thermal runaway on my amp.
Mad.P
Mad.P
Kenshin said:
I'm pretty sure it's only negative in the small signal region, I've already dismissed this with class d because it doesnt' work in that region, and if it is it's going to melt down anyway from other obvious reasons.
Nitrate, I see your point. But that wouldnt' lead to thermal run away. Thermal run away to me means a situation where one device starts to current hog, allowing it to heat more, allowing it to hog more ...
You still are dealing with a positive temp co. and so it will keep having less and less current, and shouldn't form a run away condition in the traditional sense.
What you seem to be alluding to is a voltage thermal run away caused by a positive temp co? That's interesting.
I dont' know
If it is the case though, it seems an easy one to cure, or not even a problem, truly. OHms law is a linear relation anyway right?
Regards,
Chris
Sorry folks,
I keep picking at this, i'm still convinced my amp suffers from a mild form of thermal runaway. I'll show you why i'm thinking this:-
Lets assume perfect load, perfect power supply regulation and perfect filter. Lets do the math at the top of a sinewave to make it easy. This is what i'm getting at:-
let us presume we are at the top of a 35V sine wave. The speaker load is a nice 8R and the filter is transparent. We will ingnor any freq responce and capacitance ect. Let us say the FET's RDS on is approx 0.4R at room temp.
35V ( supply to load ) / (8R+0.4) = 4.1666 Amps flowing through everything.
4.16A * 8R ( speaker load ) = 33.328V across speaker
4.16A * 0.4R ( FET's RDS on ) = 1.664V across FET
4.16A * 1.664V = 6.922 Watts dropped across fet
Now Let us say the FET's RDS on is approx 0.8R because the amp is now getting quite warm:-
35V ( supply to load ) / (8R+0.8) = 3.977 Amps flowing through everything.
3.97A * 8R ( speaker load ) = 31.81V across speaker
3.97A * 0.8R ( FET's RDS on ) = 3.18V across FET
3.97A * 3.18V = 12.64 Watts dropped across fet
As you can see doubling the RDS causes the power dropped across the fet to double even though current drops slightly. I am convinced that this explains why my amp once warm stays warm. What do you think?
Mad.P
I keep picking at this, i'm still convinced my amp suffers from a mild form of thermal runaway. I'll show you why i'm thinking this:-
Lets assume perfect load, perfect power supply regulation and perfect filter. Lets do the math at the top of a sinewave to make it easy. This is what i'm getting at:-
let us presume we are at the top of a 35V sine wave. The speaker load is a nice 8R and the filter is transparent. We will ingnor any freq responce and capacitance ect. Let us say the FET's RDS on is approx 0.4R at room temp.
35V ( supply to load ) / (8R+0.4) = 4.1666 Amps flowing through everything.
4.16A * 8R ( speaker load ) = 33.328V across speaker
4.16A * 0.4R ( FET's RDS on ) = 1.664V across FET
4.16A * 1.664V = 6.922 Watts dropped across fet
Now Let us say the FET's RDS on is approx 0.8R because the amp is now getting quite warm:-
35V ( supply to load ) / (8R+0.8) = 3.977 Amps flowing through everything.
3.97A * 8R ( speaker load ) = 31.81V across speaker
3.97A * 0.8R ( FET's RDS on ) = 3.18V across FET
3.97A * 3.18V = 12.64 Watts dropped across fet
As you can see doubling the RDS causes the power dropped across the fet to double even though current drops slightly. I am convinced that this explains why my amp once warm stays warm. What do you think?
Mad.P
Hi all,
I still need help with THD measurement on my amp. Do any of you know how to do this in the real world using non-fancy test equipment or PC soundcard? I would be very greatfull if you could guide me down this path as i have NO previous knowledge of THD or any other distortion measurement. If i need to seek professional help on this please let me know. P.S my amp is now over 1/2KW and still sounding sweet
Regards,
Mad.P
I still need help with THD measurement on my amp. Do any of you know how to do this in the real world using non-fancy test equipment or PC soundcard? I would be very greatfull if you could guide me down this path as i have NO previous knowledge of THD or any other distortion measurement. If i need to seek professional help on this please let me know. P.S my amp is now over 1/2KW and still sounding sweet
Regards,
Mad.P
Hi there fellow DIYers,
I've been quiet on here for a while as i've been concentrating on the preamp part of my amp and waiting for some huge torroids and uprated FETs to arrive. They are finally here, but befor i utilise them i have one last challange. I seem to have a problem with the finals blowing due to the high damping factor of the amp. Even a small signal under short conditions will pop my final FETs due to the massive current thrown out of the amp in an attempt to re-create the signal on the output terminals.
My main consern however is not the outputs blowing but rarther the way that they blow. They always go short. I have a fuse on the bridge to stop this from damaging any other part of the amp or supply but unfortunatly the FETs gate also shorts to the drain and source. This destroys the driver circuit as large currents can then flow between the main supply rails and drivers. The only way i can think of protecting against this without compromising performance of the driver stage is to use a fuse between the driver and output gate. I fear this will be too slow to blow however. Have any of you implemented an electronic protection circuit? If so how did it perform under fault conditions?
Have fun building all,
Mad.P
P.S I hear talk of an attempt at a valve powered classD. I don't fancy the chances of that succeding but i'm very excited and pleased to see sombody contemplating it
( If it works i for one will build one )
I've been quiet on here for a while as i've been concentrating on the preamp part of my amp and waiting for some huge torroids and uprated FETs to arrive. They are finally here, but befor i utilise them i have one last challange. I seem to have a problem with the finals blowing due to the high damping factor of the amp. Even a small signal under short conditions will pop my final FETs due to the massive current thrown out of the amp in an attempt to re-create the signal on the output terminals.
My main consern however is not the outputs blowing but rarther the way that they blow. They always go short. I have a fuse on the bridge to stop this from damaging any other part of the amp or supply but unfortunatly the FETs gate also shorts to the drain and source. This destroys the driver circuit as large currents can then flow between the main supply rails and drivers. The only way i can think of protecting against this without compromising performance of the driver stage is to use a fuse between the driver and output gate. I fear this will be too slow to blow however. Have any of you implemented an electronic protection circuit? If so how did it perform under fault conditions?
Have fun building all,
Mad.P
P.S I hear talk of an attempt at a valve powered classD. I don't fancy the chances of that succeding but i'm very excited and pleased to see sombody contemplating it
( If it works i for one will build one )
Hmm,
The resistor idea may be workable using tiny S.M resisters of about 5R in place of the conventional ones but like you say i don't think they will blow quick enough. I don't think it will be possible to current limit the fets whilst they are turned on as they are only on for a very short time, too short to allow for a reasonable measurement to take place, and limiting current with shunt resisters in the supply to the fets will only cause trouble with ringing and so on. I think the only soloution to this problem is to avoid shorting the speaker terminals and use very robust over rated output FETs. The avoidence of speaker shorts however will not be a total soloution as a very large and fast input to the amp also blows it. I found this out to my horror when i unpluged the phono connectors whilst the amp was switched on and not turned fully down!!
What can i do??
Have a nice christmas all,
Mad.P
The resistor idea may be workable using tiny S.M resisters of about 5R in place of the conventional ones but like you say i don't think they will blow quick enough. I don't think it will be possible to current limit the fets whilst they are turned on as they are only on for a very short time, too short to allow for a reasonable measurement to take place, and limiting current with shunt resisters in the supply to the fets will only cause trouble with ringing and so on. I think the only soloution to this problem is to avoid shorting the speaker terminals and use very robust over rated output FETs. The avoidence of speaker shorts however will not be a total soloution as a very large and fast input to the amp also blows it. I found this out to my horror when i unpluged the phono connectors whilst the amp was switched on and not turned fully down!!
What can i do??
Have a nice christmas all,
Mad.P
It is possible to monitor the MOSFET channel voltage drop with some digital gates, but getting the signal from the floating MOSFET source to the control is tricky. Now, you may be able to do the cycle-by-cycle current limiting process locally, referenced to each MOSFET source. Then, each MOSFET gate would be controlled by a single turn-on pulse so that if overload trippage occurs, the MOSFET remains off for the remainder of the cycle. It is not very simple and clean sounding, though. Have a great Christmas break.
thespeakerguy said:A little guidance - glad to see people trying this on their own
"1. The whole of the amp runs COLD even after a good hammering, EXCEPT for one part, the output filter inductor, it always runs hot - WHY? "
If an inductor is warm, that means it is saturating. Your switching frequency is likely too low, or the voltage which you are integrsting is higher than originally intended for the inductor. Or, the switching frequency is too high for the core material. Try using Coilcraft DO3340 devices, they are very linear and are suitable for mid power amplifiers.
I have been running the LC Audio Zap pulse for years, always upgraded to latest build and its filder coils also get a temperature about 40 degrees so I think that is normal.
I am really impressed by your breadboard construction, I used such a board in the past when trying out moving coils amplifiers.
Maybe I'll start to build my own soon
I have tried two types of current limiting in similiar (but single supply) circuit, both worked OK.
One is use fuse in the positive supply rail of a H bridge motor driver. The mosfets are IRF540/IRF9540, fuse reted at 10A is available for protecting them.
Another attempt is using series resistor in the positive supply rail and detect the voltage drop with a PNP transistor. It worked fine in a buck regulator convert 24V to 5V and output 7A.
One is use fuse in the positive supply rail of a H bridge motor driver. The mosfets are IRF540/IRF9540, fuse reted at 10A is available for protecting them.
Another attempt is using series resistor in the positive supply rail and detect the voltage drop with a PNP transistor. It worked fine in a buck regulator convert 24V to 5V and output 7A.
Hi all,
The magnetic couplers i use to send the signal accross to the drivers also come in a flavour that allows bi-directional communitation so i can very easily send a signal back to my control logic. i could in theory use this signal to shut down the op amps at the input to my amp, i don't think this will be fast enough though, and my use of a double feedback scheme ( havn't posted updated schematics yet ) is going to make the problem worse as the op amp will be shutting down at a moment when the bias voltage is not were it should be. The only way to safley shut down the amp is to turn off both fets and keep them off until the power to the amp is cycled. Stoping the amp oscillating however is not a good idea as what the high side driver and coupler are going to make of their failing bootstrap supply is anybodys guess!
I guess the question is How does one safley shut down one of these amps rather than when do i shut down. I'll be looking into this more over Xmas.
Hope you all have a brill holiday!! stay safe and don't electricute yourselfs!! LOL
Regards
Mad.P
The magnetic couplers i use to send the signal accross to the drivers also come in a flavour that allows bi-directional communitation so i can very easily send a signal back to my control logic. i could in theory use this signal to shut down the op amps at the input to my amp, i don't think this will be fast enough though, and my use of a double feedback scheme ( havn't posted updated schematics yet ) is going to make the problem worse as the op amp will be shutting down at a moment when the bias voltage is not were it should be. The only way to safley shut down the amp is to turn off both fets and keep them off until the power to the amp is cycled. Stoping the amp oscillating however is not a good idea as what the high side driver and coupler are going to make of their failing bootstrap supply is anybodys guess!
I guess the question is How does one safley shut down one of these amps rather than when do i shut down. I'll be looking into this more over Xmas.
Hope you all have a brill holiday!! stay safe and don't electricute yourselfs!! LOL
Regards
Mad.P
Hi,
At power on or during a power cycle long enough to allow the bootstrap cap to discharge at all, enough time delay needs to be implemented to allow the cap enough time to be recharged from a precharge circuit before it can switched. A set delay at power up before enabling the modulator should do.
Looking at the typical transformer based drivers you'll see they're made to hold the gate low by default, as are most drivers, then all you have to do to stop it is stop sending the signal to the drivers and the gates are automatically pulled low putting the output stage into a state of high impedance. With the right housekeeping circuitry for the bootstrap circuit you'll have no problems with it.
Regards,
Chris
At power on or during a power cycle long enough to allow the bootstrap cap to discharge at all, enough time delay needs to be implemented to allow the cap enough time to be recharged from a precharge circuit before it can switched. A set delay at power up before enabling the modulator should do.
Looking at the typical transformer based drivers you'll see they're made to hold the gate low by default, as are most drivers, then all you have to do to stop it is stop sending the signal to the drivers and the gates are automatically pulled low putting the output stage into a state of high impedance. With the right housekeeping circuitry for the bootstrap circuit you'll have no problems with it.
Regards,
Chris
Hooray!
Found a new job. I can relax again now
I didn't feel much like messing around with amplifiers lately as you can imagine, but I'M BACK!
Anyway i have a dilamer. I have a Musical Fidelity A100 class A amp. Half of the amp is charcoaled. It will take quite a lot to repair this amp. I was thinking about stripping the guts out of the amp, replacing them with one of my classD designs. Then i could go around local Hi-end audio outlets and demo the amp to see what they thought. It would be even more interesting if i did not tell them i had replaced the amp inside LOL. Do you think i would be wasting a good classA amp or do you think this would be a respectable end for the old amp?
Have fun all,
Mad.P
Found a new job. I can relax again now
I didn't feel much like messing around with amplifiers lately as you can imagine, but I'M BACK!
Anyway i have a dilamer. I have a Musical Fidelity A100 class A amp. Half of the amp is charcoaled. It will take quite a lot to repair this amp. I was thinking about stripping the guts out of the amp, replacing them with one of my classD designs. Then i could go around local Hi-end audio outlets and demo the amp to see what they thought. It would be even more interesting if i did not tell them i had replaced the amp inside LOL. Do you think i would be wasting a good classA amp or do you think this would be a respectable end for the old amp?
Have fun all,
Mad.P
Time to blow some life into this thread.
I am eventually gonna get some 6N137 optocouplers so i was gonna try this concept, i switched out nitrate´s discrete gate drivers for some gate drive IC´s. Heres a schem:
I am eventually gonna get some 6N137 optocouplers so i was gonna try this concept, i switched out nitrate´s discrete gate drivers for some gate drive IC´s. Heres a schem:
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.