Hi all,
Do you wether the output transistors can blow up even without a load connected.
This is how it happened:
I was testing my (DIY) amplifier (push/pull class AB) and the sound was pretty nice, so I decided to look at the signal with a scope - without any load connected. the signal was neat and I increased the amplitude of the input signal (1 khz sine) until the amp started to clip due to the supply voltage ( +/- 60V). At this moment the fuses (3A) of my supply flashed up and my output transistors were short cicuited between collector and emittor. How can this happen without any load?
the output devices are rated 200W 16A 160V (BDV66D / BDV67D). first I was thinking about a too high quiescent current, but I'm not sure (the transistors were still cold). Is it possible that one of the transistors blowed anyhow, so that the other blew too due to the short circuit? It's also amazing that the transistors blew really fast, the fuses didn't blew first.
If someone recognize this problem please tell me!
BTW: the circuit looks like a standard class AB with two current sources.
Best regards,
HB.
Do you wether the output transistors can blow up even without a load connected.
This is how it happened:
I was testing my (DIY) amplifier (push/pull class AB) and the sound was pretty nice, so I decided to look at the signal with a scope - without any load connected. the signal was neat and I increased the amplitude of the input signal (1 khz sine) until the amp started to clip due to the supply voltage ( +/- 60V). At this moment the fuses (3A) of my supply flashed up and my output transistors were short cicuited between collector and emittor. How can this happen without any load?
the output devices are rated 200W 16A 160V (BDV66D / BDV67D). first I was thinking about a too high quiescent current, but I'm not sure (the transistors were still cold). Is it possible that one of the transistors blowed anyhow, so that the other blew too due to the short circuit? It's also amazing that the transistors blew really fast, the fuses didn't blew first.
If someone recognize this problem please tell me!
BTW: the circuit looks like a standard class AB with two current sources.
Best regards,
HB.
oh yes, other question:
what the hell is a (integrated) damper diode?? is that a kind of backwards diode to protect the transistors against currents of inductive loads??
what the hell is a (integrated) damper diode?? is that a kind of backwards diode to protect the transistors against currents of inductive loads??
Doing what you did should not cause any problem. The power supply is 120 volts peak to peak, and the transistors are rated 160 volts. Could there be an anomoly in the design. Could you get a diagram uploaded?
Sounds like that is what they mean. In TV's they keep the flyback voltage from going too high. They can lso do as you mentioned. If that is actually what they do in your transistors, they may have gotten overloaded with too much voltage and melted down. But I don't see why.
take a look at the manufacturer's chart on Safe Operating Area (SOA) -- all is not (necessarily) as it appears. Your transistors could also have suffered from insufficient heat sinking.
>At this moment the fuses (3A) of my supply flashed up and my output transistors were short cicuited between collector and emittor. How can this happen without any load?
Hello HB,
From my experience, this can be happened when some oscillation/instability occure near (after) the clipping point,
and a huge current flows through both output transistors.
This is also usual when you test your
amp with subsonic signal.
Could you please enclose the circuit schematic?
Hello HB,
From my experience, this can be happened when some oscillation/instability occure near (after) the clipping point,
and a huge current flows through both output transistors.
This is also usual when you test your
amp with subsonic signal.
Could you please enclose the circuit schematic?
"Integrated damper diode" would generally mean an integral diode that is normally reverse-biased, but will protect the device against reverse voltages.
I agree with Dmitri's opinion that instability was the likely cause of failure.
Cheers,
I agree with Dmitri's opinion that instability was the likely cause of failure.
Cheers,
Bipolar transistors have a finite turn-off time, so if the test frequency you used was somewhat high, one transistor wont turn off before the other one turns on, resulting in a rail-to-rail short.
Another "reason" (more like pure speculation) is that the amp was clipping, causing large HF components, making the devices not turn off fast enough. In TTL gates, transistors are not allowed to saturate because they'd turn off too slowly. Clipping usually means saturation, so I'm guessing this is the reason.
<flame_guard>BTW, MOSFETS will not suffer from this problem.</flame_guard>
Another "reason" (more like pure speculation) is that the amp was clipping, causing large HF components, making the devices not turn off fast enough. In TTL gates, transistors are not allowed to saturate because they'd turn off too slowly. Clipping usually means saturation, so I'm guessing this is the reason.
<flame_guard>BTW, MOSFETS will not suffer from this problem.</flame_guard>
It can be oscillation
To me it sounds like an oscillation problem. Some amplifiers have a tendency to oscillate when they are unloaded. And since you had the gain turned to over clipping point, this is absolutely a possible reason
To me it sounds like an oscillation problem. Some amplifiers have a tendency to oscillate when they are unloaded. And since you had the gain turned to over clipping point, this is absolutely a possible reason
Does the amp have a zobel network at its output?
You may have one already incorporated, but without seeing the schematic I'm just guessing.
With no zobel network and being unloaded, depending on closed loop gain, miller compenstion etc. etc.. your amplifier could easily become a high voltage oscillator, exceed its safe SOA of the transistors, and the smoke will come out!! Once you let the smoke out of the transistors, it's all over. The smoke must stay inside..
make sure you have a zobel network at the output (will help stop ringing if any on the output signal), good supply decoupling capacitors near the output BJT's (or FET's for that matter) for stability. Make sure your miller compensation around the VAS is adequate (phase angle should not reach -180 deg. before unity gain.) If you have paralleled output transistor make sure you use ballast resistors at the output of each transistor for proper current sharing.
This is where I'd start,
Good Luck!!
------>BLM<-------
You may have one already incorporated, but without seeing the schematic I'm just guessing.
With no zobel network and being unloaded, depending on closed loop gain, miller compenstion etc. etc.. your amplifier could easily become a high voltage oscillator, exceed its safe SOA of the transistors, and the smoke will come out!! Once you let the smoke out of the transistors, it's all over. The smoke must stay inside..
make sure you have a zobel network at the output (will help stop ringing if any on the output signal), good supply decoupling capacitors near the output BJT's (or FET's for that matter) for stability. Make sure your miller compensation around the VAS is adequate (phase angle should not reach -180 deg. before unity gain.) If you have paralleled output transistor make sure you use ballast resistors at the output of each transistor for proper current sharing.
This is where I'd start,
Good Luck!!
------>BLM<-------
OK, I'm back
thanks for all the replies!!
Yes, I've forgotten the zobel network - I wanted to place it directly on the speaker connectors but didn't used it when testing with the scope.
I will use two parallel pairs of output transistors, so do I have to use one or two zobel networks - I've seen amp designs with a zobel network between each pair.
I will try to check the phase shift of the amp, but I don't think that was the problem.
I'll post the new schematic soon!
HB.
thanks for all the replies!!
Yes, I've forgotten the zobel network - I wanted to place it directly on the speaker connectors but didn't used it when testing with the scope.
I will use two parallel pairs of output transistors, so do I have to use one or two zobel networks - I've seen amp designs with a zobel network between each pair.
I will try to check the phase shift of the amp, but I don't think that was the problem.
I'll post the new schematic soon!
HB.
Standard Test AC Power Connection.....
Sounds like you have a cross conduction problem, or oscillation problem.
Either way put a 60W lamp in series with the mains transformer primary, and if you get any fault conditions the lamp will glow and the output transistors will be saved while you find the source of the problem.
Eric.
Sounds like you have a cross conduction problem, or oscillation problem.
Either way put a 60W lamp in series with the mains transformer primary, and if you get any fault conditions the lamp will glow and the output transistors will be saved while you find the source of the problem.
Eric.
Hey thanks Eric!! Good trick!!
I've never had a variac so this is a good way to work safely! I was already using a lamp for decharging the caps...
I've probably already found the problem, although I didn't test it yet... I think the problem was between the driver transistors and the output transistors, I've made a kind of darlington with two separate transistors but I've never used a resistor between the two emittors of both drivers or a resistor between the base and the collector of each output transistor. I didn't do that because I din't know the exact reason why I should (although some guys here tried before....).
When no load connected, the drivers will see a very small load (large resistance) and their output capacitance won't get the change to decharge. So when the sine changes of polarity both transistors will conduct, and will be damaged. When small signals applied there might be no problem, but when pushing the amp to it's limits (to clipping) the capacitance has no change to decharge!!! The resistor(s) will help to lower the resistance that will decharge the output capacitance of the drivers. I've had the same problem with my selfmade lightcontroller: There I've used transistors to apply a signal to the input of an opamp... first the lamps keep on shining due to the capacitance. One simple resistor solved the problem.
So this is the kind what happened to my amp I guess, not due to the high frequencies but to the capacitance....
Can anyone confirm this reason?
Thanks again to all of you - you couldn't have known the exact reason without schematic- my mistake
best regards,
HB.
I've never had a variac so this is a good way to work safely! I was already using a lamp for decharging the caps...
I've probably already found the problem, although I didn't test it yet... I think the problem was between the driver transistors and the output transistors, I've made a kind of darlington with two separate transistors but I've never used a resistor between the two emittors of both drivers or a resistor between the base and the collector of each output transistor. I didn't do that because I din't know the exact reason why I should (although some guys here tried before....).
When no load connected, the drivers will see a very small load (large resistance) and their output capacitance won't get the change to decharge. So when the sine changes of polarity both transistors will conduct, and will be damaged. When small signals applied there might be no problem, but when pushing the amp to it's limits (to clipping) the capacitance has no change to decharge!!! The resistor(s) will help to lower the resistance that will decharge the output capacitance of the drivers. I've had the same problem with my selfmade lightcontroller: There I've used transistors to apply a signal to the input of an opamp... first the lamps keep on shining due to the capacitance. One simple resistor solved the problem.
So this is the kind what happened to my amp I guess, not due to the high frequencies but to the capacitance....
Can anyone confirm this reason?
Thanks again to all of you - you couldn't have known the exact reason without schematic- my mistake
best regards,
HB.
I fix a lot of big PA amplifiers and SMPS's, so I have a box fitted with a variac, lamp and isolation transformer final output.
I have a switch to short out the lamp, and a power switch between the variac and isolation transformer primary.
With this box I can run up just about any device safely, and this has saved expensive grief many times, and is required workshop equipment in my view.
Until you get a variac a range of lamp wattages is a good idea - ie 40W/60W/100W.
Eric.
I have a switch to short out the lamp, and a power switch between the variac and isolation transformer primary.
With this box I can run up just about any device safely, and this has saved expensive grief many times, and is required workshop equipment in my view.
Until you get a variac a range of lamp wattages is a good idea - ie 40W/60W/100W.
Eric.
Hi all,
I've just modified my amp, and until now it has been running fine.
Like I said before I've put a resistor of 150 ohm between the two emitters of both drivers. Although I didn't pushed the amp to its limits, the problem seems to be solved.
Here's a pic of my amp (under test conditions):
as you can see I used the lightbulb trick to limit the current
best regards,
HB.
to be continued....
I've just modified my amp, and until now it has been running fine.
Like I said before I've put a resistor of 150 ohm between the two emitters of both drivers. Although I didn't pushed the amp to its limits, the problem seems to be solved.
Here's a pic of my amp (under test conditions):
as you can see I used the lightbulb trick to limit the current
best regards,
HB.
to be continued....
Attachments
- Status
- Not open for further replies.