as I said in my first post in this thread.... whenever and wherever you have L's and C's in series, parallel, or series\parallel combinations you have all the necessary ingredients to form a resonance....
For example, power supplies and crossovers have potential resonances.... designers routinely work with and around this point (pardon the pun)... it's not rocket science. And frequently the "cure" is to deliberately park the resonance where it is not likely to be excited... and\or design an effective damping circuit or use other anti-resonance techniques.
Just as transformer designers will attempt to park the peak of a high frequency signal transformer resonance well out past 20 khz... it's not that they can get rid of all vestiges of the resonance... but they can tailor the amplitude of the peak, the breadth of the peaking, as well as it's frequency of occurence so that it is for all practical purposes well outside of the audio band.
but... if your point is that an output transformer (say a series fed SE type) which has the last power supply C in series with it.... is also capable of producing a low frequency resonance... then your correct. In practice an output trans will have greater immunity from low frequency resonances since the secondary load provides a lot of effective damping in most cases. But this is moreso a concern say for an IT wherein the secondary is not loaded.
Again... we would maybe want to vary or modify the different circuit parameters so that we can shape the resonance and\or that the resonance can moved well into a subsonic region where the likelihood of it being excited is quite, quite slim. As well as looking at our damping mechanisms.
And that is what Paul Joppa was trying to demonstrate in his post that resonances don't just occur with or within parafeed or grid choke applications but even a series fed conventional airgapped transformer can also have the same resonances if the conditions are ripe.
I'm sure that if David put as much effort into finding resonances with conventional series feed iron circuits as he has with parafeed apps and grid choke apps.... I have no doubt he could find or create equally interesting resonances (as PJ had shown) via the simplified pspice routines that he employs.
Much of this discussion seems to be purely academic. In the real world, with many hundreds of grid choke applications that are up and running.... I've yet to hear from a single enduser who has actually encountered an uncontrolled, tsunami sized resonance which encroached upon their musical enjoyment. And we can provide you with some good predictive tools to use when looking at using a grid choke as well as suggesting specific circuit values that work, work well, and sound grreat.
msl
Last edited:
Hi,
Something is wrong here.
I cannot be sure what exactly is going on here, many variables.
What I have found that with around 120H around 2K Anode Impedance plus the various unavoidable capacitors in a "LCL Coupling" system I was able to actually partially equalise out the LF droop of the output transformer and that I had to significantly damp the resulting resonant system to attain a flat response.
Ciao T
Something is wrong here.
I cannot be sure what exactly is going on here, many variables.
What I have found that with around 120H around 2K Anode Impedance plus the various unavoidable capacitors in a "LCL Coupling" system I was able to actually partially equalise out the LF droop of the output transformer and that I had to significantly damp the resulting resonant system to attain a flat response.
Ciao T
Thorsten,
I was really hoping for your input here, as the initial question came to my mind looking at your amps design. It looks like you have preference for chokes (load or grid) over resistors or solid state. And it looks to me that you know what you are doing.
I am trying to understand how things work. Going to the basic I would say that with a resistor the load line is straight, with CCS is straight and horizontal. Using a choke as a load, the load line will have a more complex shape, as the impedance will vary with frequency. Am I correct ? This maybe can be useful, but kind not so easy to use.
D.
I was really hoping for your input here, as the initial question came to my mind looking at your amps design. It looks like you have preference for chokes (load or grid) over resistors or solid state. And it looks to me that you know what you are doing.
I am trying to understand how things work. Going to the basic I would say that with a resistor the load line is straight, with CCS is straight and horizontal. Using a choke as a load, the load line will have a more complex shape, as the impedance will vary with frequency. Am I correct ? This maybe can be useful, but kind not so easy to use.
D.
Hi,
And it seems you did not get a huge lot of extra help in the thread so far.
Despite it's tremendous value, Morgan's book sadly omits a lot of older design techniques and (like some others here) seems to gratuitously focus instead of throwing as much silicon into the mix as can be done while still calling the result a "Tube Circuit".
There is certainly much that can be done in this regard, yet the traditional methods (which include grid chokes and anode load chokes and much else) also have much to commend them.
First, most of the benefits over suitable solid state solutions or related tend to be strictly subjective, hard to quantify as "better measurements" in the traditional sense.
That does not mean that they cannot be measured, but they need regular cross referencing to psychoacoustics and taking into account complex interactions that extend beyond just the amplifier(stage) in isolation.
Second, as a rule, most current commercially available proper "grid chokes" tend to have >>1KH inductance and are wound in ways that minimises shunt capacitance.
Previously I have tested the S&B grid-chokes (no longer available) and the ones from Silk. Either had ridiculous levels of inductance and very low interwinding capacitance (I posted the S&B values on line but cannot now find them). I never tested Magnequest, but have to assume that they too work well enough with low capacitance and high inductance.
As a result of this replacing the grid leak resistor with a grid choke is in most circuits no problem, I remember recommending keeping drive impedance below 10KOhm for the S&B grid choke, I'd keep that for Silk and on the published specs for the Magnequest ones. This does exclude some driver circuits, such as WECO 91 derivates.
There are other vendors that promote items as "grid chokes" on e-bay, generally with quite low inductance values, I would avoid these.
As to "what does the grid choke do", it is something that for a number of reasons seems poorly suited to numerical analysis. It is also less than trivial to measure what really goes on.
Taking the S&B grid choke, it's inductance would fall progressively at an alarming rate as DC offset was increased, I seem to remember only around 500H at 1mA.
When driving the grid positive on peaks it means the inductance drops rapidly, which for example speeds up the recovery from overload over the simulations cited by Dave Slagle for example. As all of this is a highly dynamic system it is VERY difficult to nail down.
What I found is that with a grid choke and driving a SE amplifier having a low anode impedance driver stage (WE437A, resistive load at 12K) with signal representing around 6dB clipping overload for 1/10th of the time and at 1/10th of the nominal full output for 9/10th of the time. This resulted in a (predictable) increase of current in the output stage, the use of RC coupling actually led to a pinch off of current, reducing it well below the quiescent current.
Subjectively the amplifier with grid choke was able to play considerably louder subjectively undistorted with music and seemed to generally posess extra "macro dynamics".
Anode load chokes are a bit easier, they simply raise the impedance (and potential swing) compared to resistive load, noteworthy is that their lack of inductance makes a choke loaded driver distort more at low frequencies where the impedance falls. This distortion is in opposite polarity to the distortion of the output stage which also distorts more at LF due to the output transformers limited primary inductance.
Using a choke loaded driver stage CL coupled (with grid choke) to an output triode CAN exploit the various resonances and limitations such as to produce an amplifier that as a system has a flatter low frequency response and lower LF distortion than a traditional resistor (or CCS) loaded driver RC coupled and can produce more output power for a given THD limit plus it tends to recover from repeated transient overload better.
This means the Amplifier plays louder cleaner and often has a less "Aphex'ed" big bottomed bass than common SE Amp's.
To really exploit this it is necessary to have serious measurement gear and to be able to get samples of adjusted chokes to for example "tune in" the distortion cancellation at low frequencies.
So, as so often in audio it is necessary to actually research things empirically instead of attempting a mere intellectual (and/or PSPICE - BTW, I am a serious fan of Spice Simulation, I just do not mistake it for reality) analysis and to then discard the subject of the analysis as disagreeable with ones intellectual prejudices.
BTW, there is of course a further step past LCL coupling that with hindsight is quite obvious, but it took me several years to take it anyway.
Ciao T
And it seems you did not get a huge lot of extra help in the thread so far.
Despite it's tremendous value, Morgan's book sadly omits a lot of older design techniques and (like some others here) seems to gratuitously focus instead of throwing as much silicon into the mix as can be done while still calling the result a "Tube Circuit".
There is certainly much that can be done in this regard, yet the traditional methods (which include grid chokes and anode load chokes and much else) also have much to commend them.
First, most of the benefits over suitable solid state solutions or related tend to be strictly subjective, hard to quantify as "better measurements" in the traditional sense.
That does not mean that they cannot be measured, but they need regular cross referencing to psychoacoustics and taking into account complex interactions that extend beyond just the amplifier(stage) in isolation.
Second, as a rule, most current commercially available proper "grid chokes" tend to have >>1KH inductance and are wound in ways that minimises shunt capacitance.
Previously I have tested the S&B grid-chokes (no longer available) and the ones from Silk. Either had ridiculous levels of inductance and very low interwinding capacitance (I posted the S&B values on line but cannot now find them). I never tested Magnequest, but have to assume that they too work well enough with low capacitance and high inductance.
As a result of this replacing the grid leak resistor with a grid choke is in most circuits no problem, I remember recommending keeping drive impedance below 10KOhm for the S&B grid choke, I'd keep that for Silk and on the published specs for the Magnequest ones. This does exclude some driver circuits, such as WECO 91 derivates.
There are other vendors that promote items as "grid chokes" on e-bay, generally with quite low inductance values, I would avoid these.
As to "what does the grid choke do", it is something that for a number of reasons seems poorly suited to numerical analysis. It is also less than trivial to measure what really goes on.
Taking the S&B grid choke, it's inductance would fall progressively at an alarming rate as DC offset was increased, I seem to remember only around 500H at 1mA.
When driving the grid positive on peaks it means the inductance drops rapidly, which for example speeds up the recovery from overload over the simulations cited by Dave Slagle for example. As all of this is a highly dynamic system it is VERY difficult to nail down.
What I found is that with a grid choke and driving a SE amplifier having a low anode impedance driver stage (WE437A, resistive load at 12K) with signal representing around 6dB clipping overload for 1/10th of the time and at 1/10th of the nominal full output for 9/10th of the time. This resulted in a (predictable) increase of current in the output stage, the use of RC coupling actually led to a pinch off of current, reducing it well below the quiescent current.
Subjectively the amplifier with grid choke was able to play considerably louder subjectively undistorted with music and seemed to generally posess extra "macro dynamics".
Anode load chokes are a bit easier, they simply raise the impedance (and potential swing) compared to resistive load, noteworthy is that their lack of inductance makes a choke loaded driver distort more at low frequencies where the impedance falls. This distortion is in opposite polarity to the distortion of the output stage which also distorts more at LF due to the output transformers limited primary inductance.
Using a choke loaded driver stage CL coupled (with grid choke) to an output triode CAN exploit the various resonances and limitations such as to produce an amplifier that as a system has a flatter low frequency response and lower LF distortion than a traditional resistor (or CCS) loaded driver RC coupled and can produce more output power for a given THD limit plus it tends to recover from repeated transient overload better.
This means the Amplifier plays louder cleaner and often has a less "Aphex'ed" big bottomed bass than common SE Amp's.
To really exploit this it is necessary to have serious measurement gear and to be able to get samples of adjusted chokes to for example "tune in" the distortion cancellation at low frequencies.
So, as so often in audio it is necessary to actually research things empirically instead of attempting a mere intellectual (and/or PSPICE - BTW, I am a serious fan of Spice Simulation, I just do not mistake it for reality) analysis and to then discard the subject of the analysis as disagreeable with ones intellectual prejudices.
BTW, there is of course a further step past LCL coupling that with hindsight is quite obvious, but it took me several years to take it anyway.
Ciao T
Au contraire, he was guided toward a rational decision rather than one guided by unsupported assertions and ex cathedra pronouncements by those with something to sell or promote, as well as a rational reason why plate chokes were useful in limited situations, but not in most others.
I well understand that this disagrees with the fashion statements that many like to promote, but accurate and engineering-driven answers are more likely to be correct.
Hi Sy,
The problem is that your brand of engineering fails to engineer on a system wide level. Moreover, if we are looking at "engineering-driven answers" the answer should be:
"Grid chokes!? You mean you use tubes and transformers? Throw them out and use Transistors/IC's/whatever."
When it comes to classical performance as characterised by traditional measurements I can only reject tubes and especially single ended amplifiers as performing hopelessly bad.
My ears tell me otherwise though.
The interesting part is how to reconcile this dichotomy and how to find ways to reliably make something my (and many others) ears agree with.
Meanwhile I'll leave you to make engineering driven choices that often fail to satisfy the ear, even though the AP Two is happy... ;-)
Ciao T
The problem is that your brand of engineering fails to engineer on a system wide level. Moreover, if we are looking at "engineering-driven answers" the answer should be:
"Grid chokes!? You mean you use tubes and transformers? Throw them out and use Transistors/IC's/whatever."
When it comes to classical performance as characterised by traditional measurements I can only reject tubes and especially single ended amplifiers as performing hopelessly bad.
My ears tell me otherwise though.
The interesting part is how to reconcile this dichotomy and how to find ways to reliably make something my (and many others) ears agree with.
Meanwhile I'll leave you to make engineering driven choices that often fail to satisfy the ear, even though the AP Two is happy... ;-)
Ciao T
Yes! I assume that your above statement regarding SS devices is based on your own empirical research with your own circuits.
My empirical research tells me that tradional methods are hard to beat, but not unbeatable.
Sometimes Spice simulation resullts tell one that use of a coil is indicated, sometimes not.
That would be the "step back" -- removing one L and one C, leaving "L-coupling" ;-)
The SS equivalent is "gyrator coupling" and one place in my experience that a simple SS circuit seems to be at least as good as a coil. So IMO it's still a "tube circuit" just not a "coil circuit".
Cheers,
Michael
Hi,
Example.
Take:
Amplifier (high distortion, SE Amp)
Speaker (high distortion)
Connect correctly.
You attain low system wide distortion.
Now re-engineer the Amplifier based on your "engineering driven choices".
The Amplifier appears to perform better. In classic engineering terms.
Yet the whole system performs more poorly.
That is my "vague" criticism somewhat more detailed.
If it is vague to you, perhaps because you do not consider engineering choices on a systematic (as in including the whole system, not just a small sub-section of it)?
Unless you have considered the consequences in full a choice between A and B entails one should avoid guiding others. And often some of the consequences are unexpected and not easily predicted by conventional logic, such as the objective increase in power for a given THD in an amplifier employing gridchokes.
As someone who routinely employs resistors, current sources and chokes (not to mention capacitors, transistors, fets, bjt's and ic's) in all sorts of applications in Audio, I find all have merits and applications, should one wish to seek them out.
And knowing these merits (and demerits) in detail and how they interact with human perception helps in making the right choices (as does knowing the allowable BOM cost of course).
Ciao T
PS, I sell no grid chokes or get any money resulting from their sale, same for anode load chokes in fact.
Example.
Take:
Amplifier (high distortion, SE Amp)
Speaker (high distortion)
Connect correctly.
You attain low system wide distortion.
Now re-engineer the Amplifier based on your "engineering driven choices".
The Amplifier appears to perform better. In classic engineering terms.
Yet the whole system performs more poorly.
That is my "vague" criticism somewhat more detailed.
If it is vague to you, perhaps because you do not consider engineering choices on a systematic (as in including the whole system, not just a small sub-section of it)?
Unless you have considered the consequences in full a choice between A and B entails one should avoid guiding others. And often some of the consequences are unexpected and not easily predicted by conventional logic, such as the objective increase in power for a given THD in an amplifier employing gridchokes.
As someone who routinely employs resistors, current sources and chokes (not to mention capacitors, transistors, fets, bjt's and ic's) in all sorts of applications in Audio, I find all have merits and applications, should one wish to seek them out.
And knowing these merits (and demerits) in detail and how they interact with human perception helps in making the right choices (as does knowing the allowable BOM cost of course).
Ciao T
PS, I sell no grid chokes or get any money resulting from their sale, same for anode load chokes in fact.
The use of tubes would be indefensible for a genuine scientist. For this reason SY uses them entirely for the coolness factor and not for the sound. And chokes are apparently uncool.
It's not that, it's the selective criticism based on who's selling what. I've seen little if any criticism of the design theories of the likes of Siegfried Linkwitz or Earl Geddes who indeed have a great deal to sell. Unfortunately, their own pubic showings indicated that these theories didn't translate into realistic sounding systems especially when compared to the efforts of people who have employed products sold by Mike LeFevre and Dave Slagle. At some point you should simply make it clear that you are entrenched in a different camp and that criticisms of certain "fashions" don't necessarily reflect personal experience.
John
Hi,
Nothing. It has all to do with your approach.
Note, I commented on what gridchokes observably do in an actual amplifier above. Do you have any comments on that? Or can you suggest any interesting solid state circuit that provides similar? Now that would be an interesting debate.
Ciao T
Nothing. It has all to do with your approach.
Note, I commented on what gridchokes observably do in an actual amplifier above. Do you have any comments on that? Or can you suggest any interesting solid state circuit that provides similar? Now that would be an interesting debate.
Ciao T
At certain levels and certain frequencies with test signals, yes.
Here's a better way to low system distortion (though that's not the question regarding grid chokes, it's just an OT distraction you've introduced):
Amplifier (low distortion)
Speaker (low distortion)
The system advantage is that the low distortion is maintained over a wide range of frequencies and levels, and holds up with complex signals and over time.
But of course, it's less fashionable.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.