Wahab, some good points here from your side.
With the EF3, like many things in electronics, you just need to learn the techniques and get comfortable with them. I think stabilizing (i.e. preventing localized HF parasictic oscillation) in an EF3 output stage is straightforward. However, fail to address the elements that cause instability, and you will have problems. I've built two high power EF3 amplifers - a 280W per channel one with a 21193/21194 output stage (FT circa 4MHz)and the latest 180W per channel amp with 1302/1381 (Ft circa 30MHz) and had no problems. In the Ovation 250, a simple L pad in the driver base circuit and base stoppers in the output devices fixed the 2MHz output stage parasitic oscillation. In the e-Amp, I added stopper RC networks between the driver stage and output stage power lines and was unable to provoke oscillation despite extensive testing with all sorts of loads.
Hi , Bonsai
As you know , i simulated your amp as a whole but also specificaly
the output stage s behaviour and it seems after some tries that
its compensation can be quite improved , removing completely
the spikes that appear when testing with a square wave , as
shown in your PDF , moreover letting the high slew rate unchanged.
Still need to check further the numbers and if you dont mind
i ll upload the picture and mods to the OPS compensation
in the dedicated thread about this amp.
As you know , i simulated your amp as a whole but also specificaly
the output stage s behaviour and it seems after some tries that
its compensation can be quite improved , removing completely
the spikes that appear when testing with a square wave , as
shown in your PDF , moreover letting the high slew rate unchanged.
Still need to check further the numbers and if you dont mind
i ll upload the picture and mods to the OPS compensation
in the dedicated thread about this amp.
OT posts removed
LS impedance dips
Hi KSTR, Steven & Stuart,
Wow, a lot of response on this topic. Thank you!
You have made perfectly clear that and amplifier may be subjected (under certain conditions) to very large output currents. You also made clear that there is not only one reason, rather there are two reasons why this happens:
1. 'Difficult' signals.
2. 'Difficult' loads.
Ad 1, this is also illustrated -for example- by Bob Cordell, 'Designing Audio Power Amplifiers', p. 374 and also discussed Keith Howard in Stereophile. There is nothing we can do about it, but happily such signals are more or less rare, about 1% of the time, depending on the source material, of course. One might argue that under rare conditions and of short duration higher THD level are acceptable, that is, provided that the protection circuit will not prematurely be activated.
Ad 2, as SY replied: They indeed exist. The Wilson WATTs are notorious for that. As far as I understand, the impedance dips are caused interaction between the drivers as well as (ill designed) cross-over filters. I think that designers of powers amps shouldn't accepts this, as, opposed to ad 1, this is something one can do about it.
And finally, I wonder how much amperes you will need to drive 'difficult' speakers with 'difficult' music as well? Maybe 100 A?
Anyhow, I refuse to design 'Wilson WATT friendly' amps. Speaker designers have to take their own responsibilities and make the impedance curves as flat as possible. If they refuse to do that, we, as amp builders should refuse to buy their stuff.
Cheers,
E.
Hi KSTR, Steven & Stuart,
Wow, a lot of response on this topic. Thank you!
You have made perfectly clear that and amplifier may be subjected (under certain conditions) to very large output currents. You also made clear that there is not only one reason, rather there are two reasons why this happens:
1. 'Difficult' signals.
2. 'Difficult' loads.
Ad 1, this is also illustrated -for example- by Bob Cordell, 'Designing Audio Power Amplifiers', p. 374 and also discussed Keith Howard in Stereophile. There is nothing we can do about it, but happily such signals are more or less rare, about 1% of the time, depending on the source material, of course. One might argue that under rare conditions and of short duration higher THD level are acceptable, that is, provided that the protection circuit will not prematurely be activated.
Ad 2, as SY replied: They indeed exist. The Wilson WATTs are notorious for that. As far as I understand, the impedance dips are caused interaction between the drivers as well as (ill designed) cross-over filters. I think that designers of powers amps shouldn't accepts this, as, opposed to ad 1, this is something one can do about it.
And finally, I wonder how much amperes you will need to drive 'difficult' speakers with 'difficult' music as well? Maybe 100 A?
Anyhow, I refuse to design 'Wilson WATT friendly' amps. Speaker designers have to take their own responsibilities and make the impedance curves as flat as possible. If they refuse to do that, we, as amp builders should refuse to buy their stuff.
Cheers,
E.
Indeed , i suggested it in the last line of my previous post...
Sure that s a big constraint , the same one as if speaker s designers
where challanged to make practical designs that would have to deal
only with high output Z amplifiers, SET like...
Your call to do so. What would then be the maximum load current target for designing a 4 ohm friendly amp, say of 200W nominal (into 4ohm resistive)?
The whole topic was raised by the claim about the capacity of three pairs of lateral mosfets driving 4 ohm speaker loads (and not necessary Wilson PUPPY), with +/-60V rails, while keeping the distortions near clipping anywhere close to acceptable.
Hi Edmond
The problem surely arise with 4R speakers , but i doubt that 8R siblings
could manifest such low impedances as 2R at some frequencies unless
the design is badly misguided , be it electricaly or acousticaly.
This is unfortunately a general trend that speakers manufacturers
dont bother about providing stable loads to the amplifiers.
Perhaps that impedance curves should be provided along
with frequency responses curves , as it would allow the consumer
to choose more accurately a given pair of speaker , but this , i doubt
that it s in the manufacturer s interest...
Cheers,
One of the audio magazines fairly recently carried out this load current test. The report is linked in this Forum.
They used real music signals fed to three different speakers, specially chosen to be of severe reactance reputation, moderate reactance and a fairly gentle to amplifers.
As far as my reading of the report I came to the conclusion that all three speakers can demand upto five times as much current as the nominal speaker impedance would draw if it were simply a resistance.
That report of 5times made me change my design method of determining temperature de-rated SOAR so that the output stage could cope with 3times for a 100ms period instead of my previous 2times.
They used real music signals fed to three different speakers, specially chosen to be of severe reactance reputation, moderate reactance and a fairly gentle to amplifers.
As far as my reading of the report I came to the conclusion that all three speakers can demand upto five times as much current as the nominal speaker impedance would draw if it were simply a resistance.
That report of 5times made me change my design method of determining temperature de-rated SOAR so that the output stage could cope with 3times for a 100ms period instead of my previous 2times.
Some of the "difficult" speakers may just plainly suffer from inductors going into saturation under rated but worst-case conditions (the most evil signal). We just had a case of that in the lab where a new speaker design with a slightly underspec'd prototype xover kicked a 1.2kW@2R (and 1R capable) amp into current limit (at about 40 amps), at not extemely large SPL (I didn't even need ear protection). Small signal impedance never dipped below 3.2Ohms, though. Best practice is to overspec max inductor current (if not aircore) by a factor of two which regularily leads to 100A figures. Not every product manager is willing to spend the extra expense for that...And finally, I wonder how much amperes you will need to drive 'difficult' speakers with 'difficult' music as well? Maybe 100 A?
Makes you wonder if there wasnt a short on a filter board ,
something like 0.1mm large lost flux solder between two tracks ,
with the amp bursting this local low power low resistance...
40A should do it quite well.....
We checked that and there was no sign of things like that as the speaker behaved exactly as simulated at lower power levels, also the 'scope shots (of speaker current) showed a reproducible sharp increase once the peak current was above 20 amps or so, with a sinewave signal.... now we're building the mass production xo with the best inductors we can find and we'll see... the main culprit was an inductor in a parallel path (to smooth out the 2nd woofer impedance peak to make the xo work at all).Makes you wonder if there wasnt a short on a filter board ,
something like 0.1mm large lost flux solder between two tracks ,
with the amp bursting this local low power low resistance...
40A should do it quite well.....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.