I think he means that oscillations due to the lack of an output filter can produce large voltage swings which can destroy the amplifier. Many amps will go Chrnobyl without a zobel because there is no load and the amp oscillates.
The zobel network is a fixed impedance once the cap drops out in parallel with the load. This maintains virtually the zobel resistor load on the amp once the impedance of the speaker skyrockets. It doesn't change the 'percieved' impedance to the amplifier because the DCR is virtually constant and the impedance will change only slightly with frequency as the low value zobel resistor dominates the amp's load response.
The zobel network is a fixed impedance once the cap drops out in parallel with the load. This maintains virtually the zobel resistor load on the amp once the impedance of the speaker skyrockets. It doesn't change the 'percieved' impedance to the amplifier because the DCR is virtually constant and the impedance will change only slightly with frequency as the low value zobel resistor dominates the amp's load response.
TerryG said:
Yes, by killing, I mean able to kill a living being. The voltage can go very high, depending on when something breaks down, for example, the output C will breakdown at a certain voltage which will likely stop this dangerous event. I`m not an expert on these issues though, but don`t plan to become one. I read about this issue in some application note/tutorial on class-D amps, just don`t remember anymore where I saw it :-(
Best regards
Gertjan
Evan,
I was looking back over your calculator you provided in this thread, and it contains has provision for "C across load" (the cap of the Zobel?), but it has no input value for the resistor in series with it. Doesn't the resistor have an effect on the Q factor? Am I correct in thinking the "C across load" is the capacitor in the Zobel circuit?
I am sure glad I ask questions about things I don't understand in other members posts, killing was rather vague in its context.
Terry😎
I was looking back over your calculator you provided in this thread, and it contains has provision for "C across load" (the cap of the Zobel?), but it has no input value for the resistor in series with it. Doesn't the resistor have an effect on the Q factor? Am I correct in thinking the "C across load" is the capacitor in the Zobel circuit?
I am sure glad I ask questions about things I don't understand in other members posts, killing was rather vague in its context.
Terry😎
Terry,
The C across the load is not the zobel capacitor. Check the datasheets for the Tripath chips and you'll see there is a differential output capacitor in parallel with the speaker load. That's the value of the cap you'd use for "C across load".
The datasheets also show how to calculate the Q of the filter and they use Ctot in the calculation instead of just the output capacitor (Co) in the LC filter because the differential output capacitor (Cdo) is in parallel with Co. Since Cdo is across the load, the capacitance in parallel with each Co (for bridged outputs anyway) is effectively 2*Cdo because you can view Cdo as two capacitors in series with a virtual ground in between them. So that's how they get Ctot=Co || 2*Cdo.
Brian
The C across the load is not the zobel capacitor. Check the datasheets for the Tripath chips and you'll see there is a differential output capacitor in parallel with the speaker load. That's the value of the cap you'd use for "C across load".
The datasheets also show how to calculate the Q of the filter and they use Ctot in the calculation instead of just the output capacitor (Co) in the LC filter because the differential output capacitor (Cdo) is in parallel with Co. Since Cdo is across the load, the capacitance in parallel with each Co (for bridged outputs anyway) is effectively 2*Cdo because you can view Cdo as two capacitors in series with a virtual ground in between them. So that's how they get Ctot=Co || 2*Cdo.
Brian
ghemink said:
Lethal voltages on the Tripath? Not likely on the single rail 12-14V versions. You could kill someone with 12 volts, if you tried really hard, but it ain't likely.
Even with the transistor bounce in the output inductors, you won't see much over 16 volts peak. Not gonna hurt you.
panomaniac said:
Lethal voltages on the Tripath? Not likely on the single rail 12-14V versions. You could kill someone with 12 volts, if you tried really hard, but it ain't likely.
Even with the transistor bounce in the output inductors, you won't see much over 16 volts peak. Not gonna hurt you.
If not for the 12V Tripath, possibly with other higher voltage Tripath or other systems. Even with a 12V Tripath, if the Q-factor gets high enough, you can get large voltages.
Gertjan
panomaniac said:
Lethal voltages on the Tripath? Not likely on the single rail 12-14V versions. You could kill someone with 12 volts, if you tried really hard, but it ain't likely.
Even with the transistor bounce in the output inductors, you won't see much over 16 volts peak. Not gonna hurt you.
Hell, I can do it with 9.
http://www.darwinawards.com/darwin/darwin1999-50.html
🙂
Gertjan is talking about how the Q of the filter can sharpen considerably when the load is detached under power. Such occurances have sent a few amps up in smoke, it's a real thing, so I'd imagine under the right conditions anything is possible.
classd4sure said:
See
http://eportal.apexmicrotech.com/mainsite/pdf/an39u.pdf
for more theoretical background why an LC filter without load can be dangerous.
That apex site has a load of material on Class D, many application notes there.
best regards
Gertjan
ghemink said:
Nice link, Apex does have a load of info.
I'd imagine since it comes from a resonance it would shock you even easier. Being at high frequency you'll present a lower impedance to the voltage than you normally would, right? But not so high that it wouldn't penetrate. Sounds dangerous enough anyway.
Regards,
Chris
Hmmm.... That Darwin Navy report seems dodgey to me. A multimeter is supposed to have very low current. It is limited for the very purpose of not damaging the device under measurement. This poor sailor was more fragle than a signal transistor or a small cap .
How many times have I checked a 9V battery by placing it on my tongue. Never knew I was taking my life into my hands. FWIW, my sister used to do pacemaker research. I'll ask her about the currents used. I know they use a cap to zap.
Sure won't argue that switching 12V into a coil can generate high voltages, just look at a car ignition coil. And I mean the primary side. Used to have one of those giant Accel coils in my Dodge. The primary side would ring at over 200V peak! It killed a lot of condensors and points.
So maybe it would be worth a measure to see what kind of voltage an unloaded tripath chip generates at the coil. I've only seen about 16-17V loaded.
How many times have I checked a 9V battery by placing it on my tongue. Never knew I was taking my life into my hands. FWIW, my sister used to do pacemaker research. I'll ask her about the currents used. I know they use a cap to zap.
Sure won't argue that switching 12V into a coil can generate high voltages, just look at a car ignition coil. And I mean the primary side. Used to have one of those giant Accel coils in my Dodge. The primary side would ring at over 200V peak! It killed a lot of condensors and points.
So maybe it would be worth a measure to see what kind of voltage an unloaded tripath chip generates at the coil. I've only seen about 16-17V loaded.
panomaniac said:Hmmm.... That Darwin Navy report seems dodgey to me. A multimeter is supposed to have very low current. It is limited for the very purpose of not damaging the device under measurement. This poor sailor was more fragle than a signal transistor or a small cap .
How many times have I checked a 9V battery by placing it on my tongue. Never knew I was taking my life into my hands. FWIW, my sister used to do pacemaker research. I'll ask her about the currents used. I know they use a cap to zap.
Sure won't argue that switching 12V into a coil can generate high voltages, just look at a car ignition coil. And I mean the primary side. Used to have one of those giant Accel coils in my Dodge. The primary side would ring at over 200V peak! It killed a lot of condensors and points.
So maybe it would be worth a measure to see what kind of voltage an unloaded tripath chip generates at the coil. I've only seen about 16-17V loaded.
Depends on other things too right? Stun guns are powered by 9Volts 🙂
Did you neglect to read the part towards the bottom where they said he pushed the meter probes "into" his thumbs (through the skin and into blood) to try and measure his "internal" resistance? That's all that makes it semi believable.
panomaniac said:
I have once measured the resonance of a filter made of 20uH iron powder toroid and 1uF SMD capacitor. Powered by 24V single supply, the output peak voltage reached 150V.
It's done on a half damaged self oscillation buck regulator protype. The output filter electrolyc capacitor is removed, only remaining the parallel 1uF. Then I take it for the experiment, just inject 30KHz signal into the control loop. The feedback system rated to output 5V is not removed, so it's NOT produce 50% duty cycle and maximium power.
If it's with full power resonance, the output voltage may even rise higher. Also note: the tiny 1uF 0805 capacitor rated at about 25V handled 150V voltage without breakdown, so it's unsafe to depend on the capacitor for overvoltage protection. What would real happen on T amp may depends on the chip's current capacity, since get high voltage out of resonance needs a large current. In my case, the transistors are IRF540N rated at 33A, but the current capacity of a TA2024 is less.
classd4sure said:
Kill people with multimeter measurements on hand? (I can't see the original links, only could guess it from your posts) , sounds unbelieveable.
The computer we using now could have 110V voltage on the case if not grounded, and output 1mA on the grounding conductor if grounded. Many people have "bitten" by it, but none are killed. The current is drained from the mains via the EMI filting Y cap -- ironicaly, these cap are REQUIRED by CE regulations.
Both applying on hands, 1mA 50Hz AC is safe, but a multimeter could KILL? (Note: 50Hz is the most dangerous frequency to the heart.)
How much is a multimeter's maximium output current, in the lowest resistor gauge?
Or maybe the poor sailor happened to have some inherent heart illness?
Note: When a patient have anything (liquid pipe, etc) connected directly to the heart, small voltage or current is really dangerous, a slight disturbance coupled from power system via distributed capatance could be lethal. The case is called "electric sensitive patient" (the phrase is translated from Chinese, so it meybe literally incorrect in English).Oh my GOODNESS my thread has gotten hyjacked!!! It's the DIY terrorist, holding my thread hostage until we make an amp that won't kill anyone.
I do supose this is a good topic to consider lethal voltages and all, but I have been shocked more times than I can't count them all (none of which have come from a class D amp yet), and I am still here. Well maybe not all here, but here enough to give you guys what for.
This reminds me, we use to make these coils with magnet wire around pencils in school that would shock you with a 9v battery. They use to blow peoples minds that they got shocked from a 9v battery. We also use to take the lead from our mechanical pencils and crank our benchtop powersupplies up to 32v and hook them up, they would turn white hot and burn through 2" of wood in a few seconds.
But I digress, the LC section of the output is mostly for the speaker, is there any considerations at all concerning the MOSFETS? Also do they or zobel circuits contribute or minimise power supply pumping?
Terry🙂
I do supose this is a good topic to consider lethal voltages and all, but I have been shocked more times than I can't count them all (none of which have come from a class D amp yet), and I am still here. Well maybe not all here, but here enough to give you guys what for.
This reminds me, we use to make these coils with magnet wire around pencils in school that would shock you with a 9v battery. They use to blow peoples minds that they got shocked from a 9v battery. We also use to take the lead from our mechanical pencils and crank our benchtop powersupplies up to 32v and hook them up, they would turn white hot and burn through 2" of wood in a few seconds.
But I digress, the LC section of the output is mostly for the speaker, is there any considerations at all concerning the MOSFETS? Also do they or zobel circuits contribute or minimise power supply pumping?
Terry🙂
- Status
- Not open for further replies.