Sure, at the moment it's a Harmon Kardon AVR 230, also a very old Marantz as a sub-woofer amp, currently out of action.
The speakers are under development, the whole point of my inquiries into amps was to create a family of modules to drive my speakers after an active crossover.
Currently I have -
2 JBL 2245 18" sub-woofers each in 10 cu foot bass reflex.
2 JBL 2226 15" woofers in 4 cu foot sealed with a Q of 0.5.
2 JBL 2447 4" Voicecoil 1.5" throat compression driver mid/hi
2 JBL 2405 slot tweeters, currently unconnected.
1 JBL TLX6 3 way as a center speaker
I have 3 JBL 2453 4" VC mids that will eventually become the L C R once I construct a Center speaker of my own.
Also 2 JBL 2432 3" VC mids for 2 way surrounds, also once I construct the boxes and find some JBL 2206 or similar.
Kind of over-kill for a home system, but, you know...
Me too, I will try to do my bit, ask Br Cherry and post my own sims soon.
Best wishes
David
Last edited:
There seems to always have been rumours of the demise of lateral MOSFETs - possibly now with some truth. But their inflated price is offset by their ability (and reliability, not to mention simplicity) in high power configurations where not so many devices are required.
As I expressed previously in this thread, I believe the source of confusion is the definition of the load in the amplifier.
Firstly, since stray capacitive charging currents are largely common to both circuit halves, they do not appear in the load defined as the amplifier output (the loudspeaker). If the stray capacitances are not equal then there will will some effect but in reality this can be engineered (that is, it is normally engineered) to be negligible.
Secondly, given the size of the impedances in an audio amplifier and its power supply, the extra loading effects of the amplifier power input (the total load) are also much smaller than your reference infers in this case.
I hope this does not require any references or further explanation. As I said before, please feel free to add some stray capacitances to my previous diagram and trace the current paths in the circuit for yourself. Better still build one and measure it.
I find simplified diagrams avoid misunderstanding the fundamentals before adding complexity and clouding the issues.
I have actually built this kind of amp.
I measured very roughly the stray capacitances and got a value of around 10nf. The positive and negative stray currents do not cancel out each other ( why should they?) but it seems that the impact on the signal is quite small. I think the impedance of 10nf at 20khz is 800 ohms or so. A speaker of 8 ohm will reduce the amplitude with a factor of 100. But the NFB probably will have to take care of a small portion of mains distortion.
I have a rather effective mains filter at my disposal. With any amp, the improvments are audible when the filter is inserted. My floating PS amp doesn't improve more than any other amp so there are probably no severe issues.
But this will probably depend on circuit topology. My amp is a low feedback quite simplistic one and that short NFB loop doesn't have any problem cancelling out disturbances. A complex one like the popular "blameless" amps with a hefty amount of NFB will probably react more nervously on those stray currents.
I measured very roughly the stray capacitances and got a value of around 10nf. The positive and negative stray currents do not cancel out each other ( why should they?) but it seems that the impact on the signal is quite small. I think the impedance of 10nf at 20khz is 800 ohms or so. A speaker of 8 ohm will reduce the amplitude with a factor of 100. But the NFB probably will have to take care of a small portion of mains distortion.
I have a rather effective mains filter at my disposal. With any amp, the improvments are audible when the filter is inserted. My floating PS amp doesn't improve more than any other amp so there are probably no severe issues.
But this will probably depend on circuit topology. My amp is a low feedback quite simplistic one and that short NFB loop doesn't have any problem cancelling out disturbances. A complex one like the popular "blameless" amps with a hefty amount of NFB will probably react more nervously on those stray currents.
Thank you for an actual measured data point, always better than speculation.
The problem is not the load at 20 kHz but rather the load at frequency near the unity feedback crossover.
This is typically in the order of 1 MHz so 10 nF is quite serious, as Ric Lee has already noted.
Your low feedback amp may have had no problem but very low distortion becomes vastly more difficult.
Especially when efficiency requirements demand Class B+ operation.
Best wishes
David
Last edited:
Perhaps one should avoid floating PS'es generally. But in my special case, that particular configuration made it possible to implement several interesting things, such as simplicity, single ended class A, and most important, that power saving technique I used.
I don't think any topology has some sort of ultimate virtue. The thing is to make an amp that very likely will sound good.
I don't think any topology has some sort of ultimate virtue. The thing is to make an amp that very likely will sound good.
My notes here are not remotely speculation.
I have built many amplifiers over nearly 25 years using this basic topology - and without a single failure to date. Admittedly most of this work uses current drive or velocity-sensing motional feedback which I would advocate for anyone interested in obtaining the highest system performance.
They do not cancel, they just do not flow in the load (the loudspeaker) by virtue of being balanced - assuming the stray capacitances are equal - which they should be to a very good approximation. They do not appear in the loudspeaker because they are caused by a common signal voltage on both positive and negative supplies (and therefore stray capacitances): NO current in the stray capacitances therefore flows in the loudspeaker. The impact on the signal is therefore indeed "quite small" being in the order of non-existent (for the matched stray capacitances at least).
Another reason why current drive is preferable and why this topology with a simple resistive load is eminently suited to provide it.
There is absolutely no reason to suggest such a conclusion - even for conventional voltage drive as ATCs amplifiers reliably demonstrate so well, for example.
Last edited:
I assume that we agree that the stray caps are towards the primary winding. I can't see how one may be sure that the currents are common mode? The AC is rectified and the actual direction of the currents will depend on many things. As the 50hz mains signal will vary, different parts of the windings will receive more and less currents.
The whole thing is then rectified making it even more difficult to figure out.
But you perhaps are talking about another phenomenon, Bloke?
The whole thing is then rectified making it even more difficult to figure out.
But you perhaps are talking about another phenomenon, Bloke?
Common mode is not a good term here admittedly, but it describes two currents flowing in the two output stage halves that are nominally the same. Since the current in the loudspeaker load is formed from the difference between these currents, the common part of these currents does not flow in the loudspeaker - and therefore the normal definition of the "load".
Take two secondaries wound together such that the distributed, stray capacitances are nominally equal. Then float the secondaries by some signal voltage. The currents through both stray capacitances will then be nominally equal and proportional to the signal voltage - and thus not flow in the loudspeaker. Rectification thereafter is not relevant.
This is heavy.
Example: Assume a rising edge on the output. Then the positive PS half will push current toward the primary winding. The negative half rises also and will also push current into the primary winding. The two currents adds up and current will flow from the load into the amp.
Disregarding this, we must consider impurities on the mains signal. On a normal amp, they will go to earth. In this case they will go through the speaker.
But it's easy to make an experiment. Connect a toroid to the mains and attach one of the secondary windings to your hi end stereo speaker terminals ( the hot one)and try to figure out if it affect the perceived sound.
Example: Assume a rising edge on the output. Then the positive PS half will push current toward the primary winding. The negative half rises also and will also push current into the primary winding. The two currents adds up and current will flow from the load into the amp.
Disregarding this, we must consider impurities on the mains signal. On a normal amp, they will go to earth. In this case they will go through the speaker.
But it's easy to make an experiment. Connect a toroid to the mains and attach one of the secondary windings to your hi end stereo speaker terminals ( the hot one)and try to figure out if it affect the perceived sound.
hello soundbloke:
are you willing to share a little more info about what you mention in you post below?
AD797 driving paralleled, complementary lateral MOSFETs with floating supplies sounds very interesting.
I've read the article in Linear Audio and curious why more people haven't played with the ideas.
what did you do to stabilize it?
Years ago, I listened to the Hafler Transnova power amp on some good electrostatics and liked what I heard.
mlloyd1
are you willing to share a little more info about what you mention in you post below?
AD797 driving paralleled, complementary lateral MOSFETs with floating supplies sounds very interesting.
I've read the article in Linear Audio and curious why more people haven't played with the ideas.
what did you do to stabilize it?
Years ago, I listened to the Hafler Transnova power amp on some good electrostatics and liked what I heard.
mlloyd1
Since you are talking about driving this CE output stage with an OP amp I take it you
are not referring to a CFP? Have you considered the Bryston output stage with gain,
with diodes to correct the reverse Vbe issue that I've pointed out, driven by an OP amp?
I don't really like the Bryston but it has been pointed out to me that it has some positive
qualities mentioned in the Kolinummi book. Their design is similar to the old Tiger .01 amp.
Mark Johnson wrote this in Cordell's book thread:
Bob, have a look at the Kolinummi power amp book {sold on the Linear Audio website} pp. 232-233 and especially Fig 7.36
The simplified schematic used by Canadian manufacturer Bryston is presented in Figure 7.35. This circuit has some gain which is set with resistors R1 and R2. The transistors in the circuit remain forward biased at all times and the circuit can be thought of as non-switching even if transistor currents may go close to zero. Linearity also remains good at high frequencies and the mentioned problems of the conventional CFP are avoided.
The simulated power device currents of the three output stages discussed [CFP, EF, Bryston] are plotted in Figure 7.36. [CFP is poopy, EF is slightly less poopy] and the Bryston topology has a very wide and smooth transition.
are not referring to a CFP? Have you considered the Bryston output stage with gain,
with diodes to correct the reverse Vbe issue that I've pointed out, driven by an OP amp?
I don't really like the Bryston but it has been pointed out to me that it has some positive
qualities mentioned in the Kolinummi book. Their design is similar to the old Tiger .01 amp.
Mark Johnson wrote this in Cordell's book thread:
Bob, have a look at the Kolinummi power amp book {sold on the Linear Audio website} pp. 232-233 and especially Fig 7.36
The simplified schematic used by Canadian manufacturer Bryston is presented in Figure 7.35. This circuit has some gain which is set with resistors R1 and R2. The transistors in the circuit remain forward biased at all times and the circuit can be thought of as non-switching even if transistor currents may go close to zero. Linearity also remains good at high frequencies and the mentioned problems of the conventional CFP are avoided.
The simulated power device currents of the three output stages discussed [CFP, EF, Bryston] are plotted in Figure 7.36. [CFP is poopy, EF is slightly less poopy] and the Bryston topology has a very wide and smooth transition.
soundbloke, what is the 'best' PA using this method you have encountered? ... 'best' measured as THD20k at 50W 8R or greater power and blameless (sorry
) behaviour in all other respects ...Would it be one of the ATCs?
I'm just trying to get a feel for what can be achieved in pure Power Amp terms using floating PSs.
It is addressed to the OP and anyone else who wants to answer. I think I mentioned the
Bryston to you in your other thread, but I don't think you mentioned an OP amp front end
there.
Ok, that's me then.
The CFP makes no sense to me as an output, for basically the reason you mention, so I don't intend to look at it further for this application.
The Bryston certainly looks like it should perform well but I haven't analysed it to determine if this is merely superficial, visual niceness or if it actually performs better.
The whole point of my idea is to push as much gain into the overall feedback loop as possible, so a local loop like the Bryston is not my ideal.
But it may be that it is not practical to have all the gain in the outer loop and that I need at least some local feedback.
In that case the Bryston may be an attractive option and I shall study it more.
As an aside, I don't want to preempt Richard, but since he hasn't answered your question - I think his comment "in my book" was just an expression, not a literal book.
But you have a copy of the Kolinummi book it seems (hard to tell where the quote from Mark Johnston ends, no quote marks) like it?
Best wishes
David
Last edited:
I don't like the Bryston amp and have had some very negative things to say about it.
There seems to be compensation sprinkled in in many places, no degen on the diff pair.
If you're familiar with the infamous Tiger amps it is clearly based on them. They were
ahead of their time, and a real bargain PE kit in the late 1960s early 70s. I built them
as a young kid was able to get them working but they blew up at the drop of a hat. On
the other hand once put in a system they might run for decades if not pushed hard. I've
posted threads with simulations of them that were not very good at the start and got a lot
better when Bob C's SPICE models came out.
The Bryston takes the Tigers a step further and uses both and N and a P type power
transistor for each rail. I've joked and called this the we can't make up our mind (CMOM)
output stage. That is when Mark Johnson posted the reference to the book, I suppose to
correct me. I've not ordered the book, but I will, and often I find errors in such things so
we'll see what happens.
I simulated the Bryston just to show that it has a reverse Vbe issue that I discovered
is common to all of these output stage with gain types that are similar to the Tigers. The
obvious solution is a diode which did not work with the Tigers, but does work with the
Bryston so perhaps that is a fix for this type of output stage.
The Universal and Plastic Tigers both blew up and had probably more than one problem.
Many have reported no problems with the Tigersaurus, and I can confirm that.
These are the problems with the troubled Tigers:
Undersized resistors, they burn up at full output.
The early comp outputs rated for 200W only do 1A at 50V due to SOA, dropping the rails
by 10V would help or modern devices.
Probably not enough RF filtering at the input, if you plug in an input while powered up
they go up in smoke. Probably cross conduction due to the fact that the outputs go deep
into saturation.
Excessive reverse Vbe on the drivers especially in clipping.
I suspect a stability issue.
The one thing that impressed me about the Bryston is that it is one of the few amps that
does not have rising (by much if any) distortion with increased frequency. This, combined
with the diode fix is making the design grow on me - we'll see.
People like them and they do not seem to blow up.
There seems to be compensation sprinkled in in many places, no degen on the diff pair.
If you're familiar with the infamous Tiger amps it is clearly based on them. They were
ahead of their time, and a real bargain PE kit in the late 1960s early 70s. I built them
as a young kid was able to get them working but they blew up at the drop of a hat. On
the other hand once put in a system they might run for decades if not pushed hard. I've
posted threads with simulations of them that were not very good at the start and got a lot
better when Bob C's SPICE models came out.
The Bryston takes the Tigers a step further and uses both and N and a P type power
transistor for each rail. I've joked and called this the we can't make up our mind (CMOM)
output stage. That is when Mark Johnson posted the reference to the book, I suppose to
correct me. I've not ordered the book, but I will, and often I find errors in such things so
we'll see what happens.
I simulated the Bryston just to show that it has a reverse Vbe issue that I discovered
is common to all of these output stage with gain types that are similar to the Tigers. The
obvious solution is a diode which did not work with the Tigers, but does work with the
Bryston so perhaps that is a fix for this type of output stage.
The Universal and Plastic Tigers both blew up and had probably more than one problem.
Many have reported no problems with the Tigersaurus, and I can confirm that.
These are the problems with the troubled Tigers:
Undersized resistors, they burn up at full output.
The early comp outputs rated for 200W only do 1A at 50V due to SOA, dropping the rails
by 10V would help or modern devices.
Probably not enough RF filtering at the input, if you plug in an input while powered up
they go up in smoke. Probably cross conduction due to the fact that the outputs go deep
into saturation.
Excessive reverse Vbe on the drivers especially in clipping.
I suspect a stability issue.
The one thing that impressed me about the Bryston is that it is one of the few amps that
does not have rising (by much if any) distortion with increased frequency. This, combined
with the diode fix is making the design grow on me - we'll see.
People like them and they do not seem to blow up.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.