Is this really such a bad thing sound-wise?
From what I can gather, the drive to the output device is shunted away
as
the current through the output 'sense' resistor is increased. It does not
appear to be a switching action, so how does it generate all these nasty
spiky waveforms I have been reading about?
Also, why is it called V/I limiting when as far as I can see it's just
current limiting?
Is the 'bad sound quality' of these schemes really just because the limit
has been set too low?
From what I can gather, the drive to the output device is shunted away
as
the current through the output 'sense' resistor is increased. It does not
appear to be a switching action, so how does it generate all these nasty
spiky waveforms I have been reading about?
Also, why is it called V/I limiting when as far as I can see it's just
current limiting?
Is the 'bad sound quality' of these schemes really just because the limit
has been set too low?
You did not show a circuit, so I can only answer in general.
As long as the current limiter is not activated, it should not affect sound quality. But the moment the current limiter trips, a strong non-linear feedback loop is created around the output transistor and the limiter shunting away the input signal. This feedback loop is often not very stable, also because of interaction with a complex load. So the output signal is not just topped off nicely as with voltage clipping, but is more or less oscillating around the current limit. This is VERY audible.
To stay within the safe operating area (SOAR) of the output stage often V/I limiting is used. With V/I limiting the maximum output current is made dependent on the actual voltage across the output transistor. A large voltage across the transistor asks for a low current limit to keep the dissipation (VxI) within limits. This is normally accomplished by using level shift resistors from the current limiting transistors to the ground or the supply lines. Although this is good for the SOAR, it makes the feedback loop even more unstable. Also the addition of capacitors in this loop to slow down the initial limiting action can affect the stability of the current limiting action, although it helps to allow for high peak currents.
V/I limiters make sense for resistive loads, but complex loudspeaker loads often create such phase shifts between voltage and current that when the current is high the output voltage is low and consequently the voltage across the output transistor is high. In that case the V/I limiter may trip very early, unexpectedly early. But this may save your output stage; reactive loads increase the power dissipation in the output stage considerably.
Steven
As long as the current limiter is not activated, it should not affect sound quality. But the moment the current limiter trips, a strong non-linear feedback loop is created around the output transistor and the limiter shunting away the input signal. This feedback loop is often not very stable, also because of interaction with a complex load. So the output signal is not just topped off nicely as with voltage clipping, but is more or less oscillating around the current limit. This is VERY audible.
To stay within the safe operating area (SOAR) of the output stage often V/I limiting is used. With V/I limiting the maximum output current is made dependent on the actual voltage across the output transistor. A large voltage across the transistor asks for a low current limit to keep the dissipation (VxI) within limits. This is normally accomplished by using level shift resistors from the current limiting transistors to the ground or the supply lines. Although this is good for the SOAR, it makes the feedback loop even more unstable. Also the addition of capacitors in this loop to slow down the initial limiting action can affect the stability of the current limiting action, although it helps to allow for high peak currents.
V/I limiters make sense for resistive loads, but complex loudspeaker loads often create such phase shifts between voltage and current that when the current is high the output voltage is low and consequently the voltage across the output transistor is high. In that case the V/I limiter may trip very early, unexpectedly early. But this may save your output stage; reactive loads increase the power dissipation in the output stage considerably.
Steven
Steven,
Hello again! Sorry about the assumption and not posting a schematic, but I've only ever seen 2 schemes for this kind of thing. Here is an example of what I was talking about
This to me looks like current limiting. However, I know of a slight variation of this circuit with a couple of caps and an extra diode in there that the manufacturer calls a V/I limiter. I go with your description of what a V/I limiter is and I cannot see how this circuit is one of those.
In any case, your excellent description has made me think not to bother anyway, and rely on rugged MOSFETS, rail fuses and gate zeners.
Thanks
Hello again! Sorry about the assumption and not posting a schematic, but I've only ever seen 2 schemes for this kind of thing. Here is an example of what I was talking about
An externally hosted image should be here but it was not working when we last tested it.
This to me looks like current limiting. However, I know of a slight variation of this circuit with a couple of caps and an extra diode in there that the manufacturer calls a V/I limiter. I go with your description of what a V/I limiter is and I cannot see how this circuit is one of those.
In any case, your excellent description has made me think not to bother anyway, and rely on rugged MOSFETS, rail fuses and gate zeners.
Thanks
There is a write up of this in the Self amplifier book. If the values of the resistors are chosen correctly there shouldn't be any sonic effect. The difficulty is that a direct calculation of the values is subject to uncertainties. A SPICE simulation followed by constructing a circuit is the easiest way to choose the values. You could just build a circuit and substitute resistors but that is time consuming and carries the risk of destroyed output devices.
Btw, I just worked out a cross-coupled version of a VI limiter circuit that looks like it should be able to provide protection against overbias and thermal runaway. If it works as well as I hope it will, it might make the output stage almost bulletproof, assuming the requisite amount of heat sinking and no resistor or connection failures.
It's available here http://www.soundlabsgroup.com.au/lcaudio/lc_audio_end_millennium_technical.htm 
I have a few 'noob' questions on the attached schematic part:
1) How to allow for larger voltage swings before the limiting comes in : enlarge R31 & R35 a bit I guess, that's what I figured out by myself. (act on higher voltage)
2) What are the purpose of C20 & C21? What happens if I increase the value? (if they're put in to prevent oscillation locally then I'm clear
3) How can I make this limiting circuit allow for large transient peaks? (add a time delay, not the same as 1).
1) How to allow for larger voltage swings before the limiting comes in : enlarge R31 & R35 a bit I guess, that's what I figured out by myself. (act on higher voltage)
2) What are the purpose of C20 & C21? What happens if I increase the value? (if they're put in to prevent oscillation locally then I'm clear
3) How can I make this limiting circuit allow for large transient peaks? (add a time delay, not the same as 1).
Attachments
V/I limiting can affect sound quality, depending on the implementation...
As people pointed out, it will oscillate when active.
But, when inactive, it acts as a nonlinear shunt capacitance.
My suggestion is simple enough... do it in the PSU instead... use a regulator with a linear foldback circuit, followed by a capacitor bank. That way, you get a soft clipping. Also strap diodes across the output for good control over the flyback. Calculate the capacitor values so that the time constant leaves you inside the SOA of the output transistors for anything resembling an audio signal.
There is really no limit to the variations you can do here... the trick is to isolate potentially damaging influences from the signal path where possible.
As people pointed out, it will oscillate when active.
But, when inactive, it acts as a nonlinear shunt capacitance.
My suggestion is simple enough... do it in the PSU instead... use a regulator with a linear foldback circuit, followed by a capacitor bank. That way, you get a soft clipping. Also strap diodes across the output for good control over the flyback. Calculate the capacitor values so that the time constant leaves you inside the SOA of the output transistors for anything resembling an audio signal.
There is really no limit to the variations you can do here... the trick is to isolate potentially damaging influences from the signal path where possible.
What was the previous value of C20 and C21?
I think that the basic flaw found on these simple current-limit circuits is that when one side reaches current limit threshold and the current limit transistor is turned on, the base drive of this side is shunted to the output and this means the base drive of the other side is boosted since the Vbe reduction applied to one side gets copuled through the Vbe multiplier [or bias reference] to the other side
This may effectively make the output swing in the opposite direction for some time causing obvious oscillation, and may also cause cross conduction if the other side turns on before the current limiting side had enough time to turn off. Osciillation is made worst by adding capacitors and because the output current varies very slowly due to loudspeaker inductive behavior
In these circumstances, I recommend shunting both upper and lower base/gate drives to the output at the same time independently of the polarity of the limited current. This just bypasses the output stage and leaves the VAS driving the load directly so unstability is less likely to happen. VAS should include local current limiting and output stage must include freewheel diodes to prevent it swinging past the supply rails when limiting complex inductive-capacitive loads
Another flaw is that there is no easy way to prevent excessive dissipation during short circuit and to know if the current limiting is being activated or not, so you may end hearing unexpected distortion without knowing where it's generated. To prevent this it would be a great idea to use a timer to maintain the base drive shunted for 250ms periods [or so] trying to return to normal operation at the end of each period and lighting a LED while limiting is on
The current limiter may end having as much complexity as the rest of the amplifier
I think that the basic flaw found on these simple current-limit circuits is that when one side reaches current limit threshold and the current limit transistor is turned on, the base drive of this side is shunted to the output and this means the base drive of the other side is boosted since the Vbe reduction applied to one side gets copuled through the Vbe multiplier [or bias reference] to the other side
This may effectively make the output swing in the opposite direction for some time causing obvious oscillation, and may also cause cross conduction if the other side turns on before the current limiting side had enough time to turn off. Osciillation is made worst by adding capacitors and because the output current varies very slowly due to loudspeaker inductive behavior
In these circumstances, I recommend shunting both upper and lower base/gate drives to the output at the same time independently of the polarity of the limited current. This just bypasses the output stage and leaves the VAS driving the load directly so unstability is less likely to happen. VAS should include local current limiting and output stage must include freewheel diodes to prevent it swinging past the supply rails when limiting complex inductive-capacitive loads
Another flaw is that there is no easy way to prevent excessive dissipation during short circuit and to know if the current limiting is being activated or not, so you may end hearing unexpected distortion without knowing where it's generated. To prevent this it would be a great idea to use a timer to maintain the base drive shunted for 250ms periods [or so] trying to return to normal operation at the end of each period and lighting a LED while limiting is on
The current limiter may end having as much complexity as the rest of the amplifier
@mikeks: thanks, mail sent I put 'noob' questions alert for a purpose 
@Eva: C20 & C21 are 10nf. Basicly, I'm limited to the same three questions in my previous post, I don't have the abilities to alter this drasticly.
@Angel: To do it in the PSU seems to be a bit tricky (at least for me), I have 3x10.000uf per rail and 1KVA trafo in hooked up with large gauge wire. Flyback diodes are present on the output as far as I can figure.
@thanh: Don't take it personally, but your posts are sometimes too short - be more specific Did you simulate/are we talking about the same circuit, etc...
The limiting circuit is from Randy Slone's OPTI-MOS v. 04 amp (I have a mail from him allowing me to post the schematic so here it is). I did not include the part for 'overload' which connects to the part of the circuit in question.
I put 'noob' questions alert for a purpose @Eva: C20 & C21 are 10nf. Basicly, I'm limited to the same three questions in my previous post, I don't have the abilities to alter this drasticly.
@Angel: To do it in the PSU seems to be a bit tricky (at least for me), I have 3x10.000uf per rail and 1KVA trafo in hooked up with large gauge wire. Flyback diodes are present on the output as far as I can figure.
@thanh: Don't take it personally, but your posts are sometimes too short - be more specific
Did you simulate/are we talking about the same circuit, etc...The limiting circuit is from Randy Slone's OPTI-MOS v. 04 amp (I have a mail from him allowing me to post the schematic so here it is). I did not include the part for 'overload' which connects to the part of the circuit in question.
Ok!
I want to say about my heart! But my knowledge is not good!I have simulated this protection circuit . The sim say that it is all ok. And I built it ,my amp has "aw aw...aww" sound. Can you understand? .I removed these capacitor and the hell sound is hide
. I see that if you built these capacitor and you're not lucky , your amp will not failure instanly . When you increase vulume very much , your amp can die or nor die- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.