Do You Have a Solution for Bad Behavior of V-I Limiters in Bridged Amplifiers?

Hi everyone!

Back in 1980, when I was designing and building my first “serious” stereo audio power amplifier, I found the safe operating area (SOA) protection circuit to be the most challenging part. After I was finally satisfied with the V-I limiter performance in stereo mode, I then discovered that it did not perform at all well when the two channels were bridged. When limiting was triggered, the voltage across the load would “snap” to ground producing nasty sounds from the speaker (please see the attached file for a simulation demonstrating the problem). I invested more time looking for a fix, but found none that were esthetically pleasing. As I never used this mode, I eventually lost interest in pursuing it.

Jump ahead 40+ years and this issue still nags me every now and then. There must be people on this forum who have come up with a good V-I limiter solution for bridged amplifiers. I know that most of you will say “easy, don’t use V-I limiting.” This response certainly has merit and may be the best way to go, but it does not really address the question. A well-designed V-I limiter extracts the most power from the output stage while providing reliable protection.

I know the this topic was discussed in the Amplifier based on 2N3055 thread but I didn’t see a solution if it was there. My searches may have missed other discussions. Does anyone have something to contribute?

Thanks in advance!

Bruce
 

Attachments

Bridging, two closed loop amplifiers ("servo's"), and two VI-limiters. That's four equations and four variables to get into control. Keeps itching for fourty+ years onwards. What circuits / amps are involved?
Do the limiters (non-inv and inv sides) 'know' their mutual existance? Are they 'coupled' in any way? It resembles to me like somesort of an unwanted astable multivibrator arrangement. Nice puzzle!
 
Everything you say is correct, but my question is not how to keep it from triggering, but to make it behave better when it does.
Oh, VI limiters are NASTY!

They just do their job, very well.

Being into the Guitar Amp business, I never use them because many users will eventually plug an extra cabinet "simply because it´s lying around" and it will sound UGLY.

I use plain current limiters which sound like regular clipping, of course heatsinks overheat a lot, and a bimetal switch takes care of that, turning amp off for many minutes until cool-down.

This is psychologically more acceptable to users.
 
  • Like
Reactions: Citizen124032
The current limiter should never trip in normal use. If it trips, the amplifier needs to handle more current.

If the load is anomalous, have the current limiter shut down the amplifier completely.
Ed
Ed,

I agree with your first sentence and this is how I design my amplifiers. This thread was intended to be more from an "academic exercise" standpoint to design, or find a design for, a V-I limiter for bridged amplifiers that limits cleanly.

Regarding the second sentence, I have not done this in my amps to date but I agree that it is a good idea.

Bruce
 
I use plain current limiters which sound like regular clipping, of course heatsinks overheat a lot, and a bimetal switch takes care of that, turning amp off for many minutes until cool-down.
It takes a while for the thermal shutdown. Do the output transistors stay within their published SOAs during this time with the plain current limiters activating? Perhaps the guitar waveforms are forgiving or do you have extremely robust output stages?
 
Again and awaiting, the involved issues.
Attached is the schematic of the SOA protection for one amplifier (both are the same) in the simulation referred to in the attachment to Post #1. Points C and D are the bases of the pre-driver transistors in the output triples. The +/- Sense points are the emitter-resistor/output-transistor-emitter nodes of one transistor pair. The emitter resistors are 0.33 ohms and there are 6 pairs of output transistors. Only one pair is used for current measurement. The Output Bus is the load connection point. This is a pretty common commercial configuration.

The exact schematic is not really important. I used this as it is simple and makes the simulation results easy to understand. What is important is that diodes D2 and D5, along with resistors R2, R4, R6, and R7, provide fold-back V-I current limiting as described in the attachment to Post #1.
 

Attachments

A bridged amplifier with a 4 ohm load.

essentially seeing 2 ohms on each side.

Your just basically designing individual 2 ohm amplifiers.
Were the VI limiter sensing is set to not trigger to early.

it is the usual problem I have seen.
Most VI limiters are set to start grabbing around 2 ohms for current levels in a 4 ohm amplifier.

In a bridged case, limiter set to 2 ohm current levels is gonna pop off too early.
Threshold needs to be lower at 1 ohm current levels.

Very easy to do in simulation.
Bring the amp to clipping with a 1 ohm load.
Set the trigger to barely grab at 1ohm not 2.
 
Last edited:
  • Like
Reactions: rayma
The circuit is best left in the past.
Ed
Ed, I'm beginning to think that you want this to nag me for another 40 years 🙂. True, it is probably not worth investing the time to pursue a solution at this point [unless someone is bored and looking for a design challenge], but I thought that with the enormous number of Members here, someone must have looked into this in the past, or noted how some commercial amplifier dealt with it. Many high powered commercial amps always operate in bridge mode, but I suspect that most designers may not have worried about this.

Bruce
 
A bridged amplifier with a 4 ohm load.

essentially seeing 2 ohms on each side.

Your just basically designing individual 2 ohm amplifiers.
Were the VI limiter sensing is set to not trigger to early.

it is the usual problem I have seen.
Most VI limiters are set to start grabbing around 2 ohms for current levels in a 4 ohm amplifier.

In a bridged case, limiter set to 2 ohm current levels is gonna pop off too early.
Threshold needs to be lower at 1 ohm current levels.

Very easy to do in simulation.
Bring the amp to clipping with a 1 ohm load.
Set the trigger to barely grab at 1ohm not 2.

All true, WhiteDragon, but I WANTED the simulation to trigger! I was trying to demonstrate the type of behavior that I was looking to eliminate.

It can be shown in the simulation (using a sine wave instead of a triangle) that the clipped waveform with "bat ears" has almost twice the harmonic content of a sine wave with a clean hard limit at the same level. The THD of the bat wing waveform is ~50%!

Bruce
 
No, I-V limiters do what they say on the tin, and if that interrupts a normal signal, it won't sound nice.
I think the solution you are looking for is to parallel a few more transistors and heatsinks on your output so that the I-V limter does not trip at the lower effective load seen by a bridged amp. And as others have mentioned, the load of a bridged amp is four times the power, so you need to be considering some seriously heavy output stages to solve it. Some heavy duty amps are designed to work into 1 ohm continuously, for that reason. Can be done, but it is not trivial.
 
  • Like
Reactions: JMFahey
Although I appreciate all of the good and well intentioned advice above, I am sill hoping that someone can directly address the original question.

The engineer in me hates to be defeated by a technical challenge. To word it differently: Given two amplifiers, each with optimally designed V-I limiters, is it possible to operate the amplifiers in bridge mode without incurring additional audio degradation?

Citizen124032 and both hinted at what is necessary. Any solutions or ideas for one?