Filters , poles, interaction, resonance

What people are working on at the moment
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#1 Filters , poles, interaction, resonance

Post by IslandPink »

I will probably get out of my depth pretty quickly here, but I'd just like to tackle the issue we discussed on Steve C's amp thread , first .

I wrote :
"Stephen, do you believe that RC filters in different parts of a power amp interact with each other in a resonant way ? "

Stephen wrote :
"In a vinyl system, you can push the LF low enough to excite resonances in other parts of the system and those can eat power and cause bad sound, overload etc. Cascade issues with RC filters can cause ripples like they do with different filter types. Is a ripple that is seen in some filter types a resonance? Are beat frequencies in power supplies that lead to motorboating a resonant RC filter interacting? This can be an issue in a two stage amp never mind a 3 stage. Are these just different names/terms for resonance?

A regulator can hide these issues and some people don't experience them as their first choice is to regulate because regulation is always better ;)

As usual, most of my experience is with phono stages, some with transformers, TVCs and so on so it's hard to isolate and say RC filters interact but I think yes they can in conjunction with circuit topology and PSU configuration.

cheers"
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#2 Re: Filters , poles, interaction, resonance

Post by IslandPink »

Now there's a lot in that reply, and I don't think you answered exactly the question I asked. You brought the power supply into the equation again. That's a different game, and the potential for oscillation in power supplies, where you have chokes and caps, is well known.

On the subject of what happens with the RC filters in an amp, I believe their effects just add.
For a good example of this, take the circuit here :
http://www.izzy-wizzy.com/audio/images/ ... k22aud.gif
In the section between the D3a and the 5687 you have at least three, maybe four filters. I don't believe there's any instability or oscillation from packing these RC elements together, certainly I've never had any problem myself, having build several versions of the 'all-in-one' RIAA.
"Once you find out ... the Circumstances ; then you can go out"
Cressy Snr
Amstrad Tower of Power
Posts: 10552
Joined: Wed May 30, 2007 12:25 am
Location: South Yorks.

#3 Re: Filters , poles, interaction, resonance

Post by Cressy Snr »

"It goes without saying that network analysis requires mathematical tools--in particular: algebra (to expand and reduce basic operations, to find roots, and to solve differential equations); the theory of functions and graphs (to plot waveforms as functions of time); trigonometry (to mark time and to find phase angles); differential and integral calculus (to freeze and sum quantities which are functions of time); complex number theory and vector analysis (to evaluate the phase shift between voltage and current waveforms and to plot and interpret poles and zeros); and, ideally, operational calculus such as the Laplace transform (to simplify operations and to transform frequency domain functions into time domain functions)."


Should be a walk in the park then.
good article here.
http://www.triodeel.com/feedback

Bet there's an app for that now.
Sgt. Baker started talkin’ with a Bullhorn in his hand.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#4 Re: Filters , poles, interaction, resonance

Post by IslandPink »

IslandPink wrote: Mon Nov 20, 2017 10:52 am Actually Stephen wrote -
"In a vinyl system, you can push the LF low enough to excite resonances in other parts of the system and those can eat power and cause bad sound, overload etc. "
This is a valid point, good that Nick reminded me of it. I think the main effect is saturation in the output transformer though. I still don't see any reason to do anything other than make all the other time-constants in the main amp as low as possible, eg. 2 or 3Hz . After all, this is a phono source issue. It seems logical to handle the subsonic phono problem in the phono amp, by choice of rumble filter. I'm with SJS on this - set your rumble filter for the best system compromise ( I use about 10Hz, but then my speakers are very sensitive so the main power amp is not that near clipping ) then get the other time constants 'out of the way'.
If I do additional high-pass filtering in the main amp, it will degrade low bass performance from digital sources, which don't have the subsonic rumble/warp issues. I'd prefer to get the best bass tone and timing I can from the digital material. Hence my policy is to eliminate or reduce any other time constants.

IslandPink wrote: Mon Nov 20, 2017 10:52 am Actually Stephen wrote -
A regulator can hide these issues and some people don't experience them as their first choice is to regulate because regulation is always better ;)
A regulated supply generally allows better fidelity at low frequencies, as it keeps the power supply Zout low down to subsonic frequencies , unlike any passive supply. For instance the shunt-reg main supply for my power amps has a Zout of around 2R down to below 10Hz, if the demand ( mA ) is not too high.
So the reg. supply will probably not help the output transformer overload situation, but it will sound better in my experience. I used a number of passive supplies till about 2008, then moved to series-reg ( a la Thorsten/Verdier circuit ) then shunt-reg about 5 years ago. The regulated supplies were a distinct step up, in each case.

A shunt-reg supply is the best of all because :
1. It provides a clean current return loop for the signal
2. Almost all the demand variation is taken up at the output of the PS circuit, in the shunt - so the rectifiers & passive smoothing section leading up to the shunt see hardly any demand and operate in a very optimal manner with no risk of overload or oscillation.

Sorry if this is stuff you already know but you did wink :wink:
"Once you find out ... the Circumstances ; then you can go out"
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#5 Re: Filters , poles, interaction, resonance

Post by izzy wizzy »

IslandPink wrote: Mon Nov 20, 2017 11:02 am I don't believe there's any instability or oscillation from packing these RC elements together, certainly I've never had any problem myself, having build several versions of the 'all-in-one' RIAA.
Agreed. If we're talking only the coupling RC networks in a stable amplifier and stable front end i.e. vinyl

However the original question was "Stephen, do you believe that RC filters in different parts of a power amp interact with each other in a resonant way?"

And therein lies, lets say, ambiguity in the question as to where these RC filters are? Are they the RC filters in the power supply used for decoupling? If so, yes, they can lead to motor boating if not done right which is a resonance.

I have to take a guess here (I don't use cathode cap decoupling) but I can imagine that changing the decoupling caps values could push a marginal amp into motorboating. Is that a reaction of the RC cathode decoupling reacting with the powqer supply RC decoupling which are all RC filters leading to motorboating? And likewise, poushing the LF lower by dcreasing the coupling caps could also lead to motorboating.

And I think that the origin of all this was your suggestion, not a bad one, that the LF pole of Steve's amp could be pushed lower and while in itself, no bad thing, Steve suggested it could lead to an extended LF response that meant LF instability in the LP system could overload the amps magnetics etc. So while the RC networks themselves won't resonate, resonant issues could result in pushing the LF poles lower, too low for comfort, too low for that system etc.

Does that make any sense?

cheers,

Stephen
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#6 Re: Filters , poles, interaction, resonance

Post by izzy wizzy »

Oh dear, we cross posted :)
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#7 Re: Filters , poles, interaction, resonance

Post by izzy wizzy »

IslandPink wrote: Mon Nov 20, 2017 10:52 am Actually Stephen wrote -
A regulator can hide these issues and some people don't experience them as their first choice is to regulate because regulation is always better ;)
A regulated supply generally allows better fidelity at low frequencies, as it keeps the power supply Zout low down to subsonic frequencies , unlike any passive supply. For instance the shunt-reg main supply for my power amps has a Zout of around 2R down to below 10Hz, if the demand ( mA ) is not too high.
So the reg. supply will probably not help the output transformer overload situation, but it will sound better in my experience. I used a number of passive supplies till about 2008, then moved to series-reg ( a la Thorsten/Verdier circuit ) then shunt-reg about 5 years ago. The regulated supplies were a distinct step up, in each case.

A shunt-reg supply is the best of all because :
1. It provides a clean current return loop for the signal
2. Almost all the demand variation is taken up at the output of the PS circuit, in the shunt - so the rectifiers & passive smoothing section leading up to the shunt see hardly any demand and operate in a very optimal manner with no risk of overload or oscillation.

Sorry if this is stuff you already know but you did wink :wink:
[/quote]

I have yet to have a good go with shunt regs. In fact I had many good goes over years with series regs of many kinds and to me, they just get in the way, even in the heaters. It's never been so much as what they do down low but up higher.

My wink was more about where I see many people assume regulation of any kind is the way forward. I can see commmercially, it definitely has a place.

I suppose I have an issue where if you change the "controlling" part of the reg and the sound changes, then what exactly is going on? It seems just another variable and IME, a variable that doesn't bring me joy and again, IME, I hear it as taking something away in that nothing is stable and then the reg has to correct/react and so you have an amp being modulated by a reacting regulator. It just doesn't sit well with me. I could be missing the point.

I also hear many say about the isolation they give. Many series regs don't isolate much at all higher up where RC decoupling performs very well. In a sim, feed a signal into the output of some regs and it comes out the input amplified as many series regs that way round are common base amplifiers. I did this a long time ago so I might not be explaining it very well. However the argument could go, well there won't be any signal developed to amplify as the impedance at the output is so low. Again, at low frequencies, yes but what about further up?

I have difficulty with plain statements like my reg has xx ohms output Z. Well it might at 50Hz but what about 1kHz, 10kHz? There's so much gunge and rubbish flying about in both directions in an amplifier power supply, the reg has to perform not just in one direction.

It just isn't for me but then see above re my lack of experience with shunt regs. One day when I have more time ... you know the story.

cheers,

Stephen
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#8 Re: Filters , poles, interaction, resonance

Post by IslandPink »

Cressy Snr wrote: Mon Nov 20, 2017 11:31 am "It goes without saying that network analysis requires mathematical tools--in particular: algebra (to expand and reduce basic operations, to find roots, and to solve differential equations); the theory of functions and graphs (to plot waveforms as functions of time); trigonometry (to mark time and to find phase angles); differential and integral calculus (to freeze and sum quantities which are functions of time); complex number theory and vector analysis (to evaluate the phase shift between voltage and current waveforms and to plot and interpret poles and zeros); and, ideally, operational calculus such as the Laplace transform (to simplify operations and to transform frequency domain functions into time domain functions)."

Should be a walk in the park then.
good article here.
http://www.triodeel.com/feedback

Bet there's an app for that now.
I think the 'App' is called LT Spice, Steve !
However having skimmed through the extensive page of tech approach there, I think it's really aimed at getting an amp to be stable when it's using global feedback. I checked with Nick on this and he agrees. You could do it on a zero-feedback SE amp but it would only be for academic interest.

I'll get back to you later, Stephen - prob after I've called Pete ( 'I should coco' ) for a natter !
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#9 Re: Filters , poles, interaction, resonance

Post by IslandPink »

izzy wizzy wrote: Mon Nov 20, 2017 12:08 pm
1. Agreed. If we're talking only the coupling RC networks in a stable amplifier and stable front end i.e. vinyl

2. However the original question was "Stephen, do you believe that RC filters in different parts of a power amp interact with each other in a resonant way?" And therein lies, lets say, ambiguity in the question as to where these RC filters are? Are they the RC filters in the power supply used for decoupling? If so, yes, they can lead to motor boating if not done right which is a resonance.

3. I have to take a guess here (I don't use cathode cap decoupling) but I can imagine that changing the decoupling caps values could push a marginal amp into motorboating. Is that a reaction of the RC cathode decoupling reacting with the powqer supply RC decoupling which are all RC filters leading to motorboating? And likewise, poushing the LF lower by dcreasing the coupling caps could also lead to motorboating.

4. And I think that the origin of all this was your suggestion, not a bad one, that the LF pole of Steve's amp could be pushed lower and while in itself, no bad thing, Steve suggested it could lead to an extended LF response that meant LF instability in the LP system could overload the amps magnetics etc. So while the RC networks themselves won't resonate, resonant issues could result in pushing the LF poles lower, too low for comfort, too low for that system etc.

5. Does that make any sense?
1. Cool !

2. Well you are right to include the power supply as being in 'the amp' or course !
I originally meant in a more narrow way ( Narrow Way, Gilmour - very nice track )

However I think what I am getting at is that you need inductance to get instability or oscillation going. Nick behind the scenes dug out a good statement for me :

' "Electrical resonance occurs in an electric circuit at a particular resonant frequency when the imaginary parts of impedances or admittances of circuit elements cancel each other. In some circuits this happens when the impedance between the input and output of the circuit is almost zero and the transfer function is close to one."

Oscillation on the other hand needs gain or at least negative resistance to provide positive feedback and a filter network to select a frequency of oscillation '

3&4 : I think the scenarios you suggest are all rather dependant on having a power supply that is unstable or marginal. I have always used PSud to design passive supplies, and the typical 'small c > choke > big C' arrangement or 'small c > choke > medium C > choke > big C ' combo are easily made stable to step-changes or startup with a good range of C values. The only design task required is to get the re-charge time shorter without losing the stability.
I think if making bass time constants really low all through the amp ( or phono ) with a PS of this sort was going to result in oscillation , I'd have discovered that a long time ago. These sort of supplies just get a bit sluggish and add their own detrimental phase shift to that of the amp, when you load them with lower and bigger signal frequencies.

5. Quite a lot :D
"Once you find out ... the Circumstances ; then you can go out"
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#10 Re: Filters , poles, interaction, resonance

Post by IslandPink »

izzy wizzy wrote: Mon Nov 20, 2017 12:26 pm I have yet to have a good go with shunt regs. In fact I had many good goes over years with series regs of many kinds and to me, they just get in the way, even in the heaters. It's never been so much as what they do down low but up higher.

My wink was more about where I see many people assume regulation of any kind is the way forward. I can see commmercially, it definitely has a place.

I suppose I have an issue where if you change the "controlling" part of the reg and the sound changes, then what exactly is going on? It seems just another variable and IME, a variable that doesn't bring me joy and again, IME, I hear it as taking something away in that nothing is stable and then the reg has to correct/react and so you have an amp being modulated by a reacting regulator. It just doesn't sit well with me. I could be missing the point.

I also hear many say about the isolation they give. Many series regs don't isolate much at all higher up where RC decoupling performs very well. In a sim, feed a signal into the output of some regs and it comes out the input amplified as many series regs that way round are common base amplifiers. I did this a long time ago so I might not be explaining it very well. However the argument could go, well there won't be any signal developed to amplify as the impedance at the output is so low. Again, at low frequencies, yes but what about further up?

I have difficulty with plain statements like my reg has xx ohms output Z. Well it might at 50Hz but what about 1kHz, 10kHz? There's so much gunge and rubbish flying about in both directions in an amplifier power supply, the reg has to perform not just in one direction.

It just isn't for me but then see above re my lack of experience with shunt regs. One day when I have more time ... you know the story.

cheers,

Stephen
I understand your concerns - there are ways of using simple voltage regulation ( eg. in filament supplies ) that will take away a lot of the music.
My uses of regulation have been mostly valve or VR-tube based, where the risks of spoiling the sound are a lot lower.
The series-reg for the power amp was like the Thorsten/Verdier one, but with an additional choke in the pre-reg section :
http://community.fortunecity.ws/rivende ... egacy.html
More recently I built this ( more or less ) :
http://enjoythemusic.com/magazine/equip ... x/fig2.gif
However I only used a small film cap on the output, no point in spoiling the clean return path with an ordinary electrolytic. Most of the signal circulates through the KT88. The cathode by-pass on there is a black gate NH.
On the phono amp since 2005 I have used a conventional smoothed PS into a Pimm CCS feeding a VR-valve stack ( 2x OD3 ) shunt providing 300V .
This was so much better than the previous passive supplies I'd used, that I never went back. Bear in mind however, it's only suitable to feed the final stage direct, there's an additional RC decoupling to the first stage.
"Once you find out ... the Circumstances ; then you can go out"
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#11 Re: Filters , poles, interaction, resonance

Post by izzy wizzy »

The 2nd link didn't work for me unfortunately. I've tried variations of Thorsten's thingy but they're not to my liking.

There is also another possibility to consider through all this although I can't say I've fully explored it but I might be part of the same school as Thomas Mayer and you might be more Gary Pimm :) however I'm no single ended person (at high level) more a balanced/PP kinda person and pentodes are OK too however a shunt glow valve reg for a power amp is something I like just haven't done it for phono.

I think all these different approaches depends on what we're after as an end result and I've come from regulate everything to regulate nothing and in so doing have found my listen is not analytical as much as it was and more a feeling type thing and yet my designing uses LTSPICE more than ever. It really is personal and no two of us are alike. So much of htis is way more how our brain's process the signal from our ears; probably more than we would like to admit :)

cheers

Stephen
User avatar
jack
Thermionic Monk Status
Posts: 5493
Joined: Wed Dec 29, 2010 8:58 pm
Location: ɐılɐɹʇsnɐ oʇ ƃuıʌoɯ ƃuıɹǝpısuoɔ
Contact:

#12 Re: Filters , poles, interaction, resonance

Post by jack »

Interesting thread - it is all about the sound - at least we all agree on that.

Like some others here, whilst the sound is primary, I like to understand what differentiates a mediocre system for one that (to me) sounds great...

...and that's where the maths comes in. LTspice is great - I use it a lot - but it's only theory.

One thing I use is ExtractModel from Derk Reefman (used with uTracer) - it generates real-world spice models from actual tubes - there is a substantial library of 100s of standard tubes that have been measured and then the models generated... I've attached a recent version to this post - note that the pin numbering scheme is not the same as the default scheme used by LTspice, so the symbols need a bit of playing with, though I can supply those too should you wish.
Attachments
TubeLib.inc
(104.28 KiB) Downloaded 233 times
Vivitur ingenio, caetera mortis erunt
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#13 Re: Filters , poles, interaction, resonance

Post by IslandPink »

izzy wizzy wrote: Tue Nov 21, 2017 1:03 pm The 2nd link didn't work for me unfortunately. I've tried variations of Thorsten's thingy but they're not to my liking.

...however a shunt glow valve reg for a power amp is something I like just haven't done it for phono.
TRex_fig2.gif
How would you do a VR tube shunt for a power amp without using an arrays of a dozen or more & trying to balance then ? - on the other hand I think Paul once did something like that !
On a phono it's easy because you only need to shunt 20ma or less.
"Once you find out ... the Circumstances ; then you can go out"
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#14 Re: Filters , poles, interaction, resonance

Post by izzy wizzy »

IslandPink wrote: Tue Nov 21, 2017 8:35 pm How would you do a VR tube shunt for a power amp without using an arrays of a dozen or more & trying to balance then ? - on the other hand I think Paul once did something like that !
On a phono it's easy because you only need to shunt 20ma or less.
I should have said for a power amp driver, sorry.

And that reg is more complicated than anything I've built. That kind of thing just doesn't sit well with me which might mean I'm resistant to it which could mean I'm unreasonable. On the other hand, I just can't see wny I need something like that.

I have this other theory on these things. I can't help but think a thing like that must have a sonic signature of it's own that it imprints on the circuit it's feeding. By it's very nature, it must react to what's going on with the power supply rails which let's face it, are moving in sympathy with the music being played. It has to react to maintain equilibrium and I can't see it doing that evenly over the frequency spectrum.

And so the designer has to be happy to not only deal with the circuit and it's non-linear goings on but also the non-linear actions of the regulator. From what I've heard, some are better than others but ultimately although the better ones can be analysed to make a better sound; tighter bass, possibly smoother highs or whatever, I find the message that people are playing music, the raw expression of this gets lost. It's so hard to explain. It's a delicate balance.

If your mains is all over the place, then regualtion may even be a necessity especially for heaters but I still don't care.

But then that's just me. I mean, I do stupid things like using MV rectifiers. There's no logic in doing that but I like the result and for safety, they're in a box and I can't even see them.

I've tried Garry Pimm CCS instead of plate choke in phono. Went back to plate choke. Tried CCS in tail of diff pair on output of phono vs resistor. Went back to resistor.

One day I might try a glow tube. At present I can't as I'm up against the current limit of the chokes in my power supply and I also can't raise the HT to allow the votage drop due to the cap ratings. Maybe I did that on purpose so I can't try them ;)

cheers,

Stephen
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#15 Re: Filters , poles, interaction, resonance

Post by IslandPink »

Try something on a driver valve first.

I can tell you that I'm definitely in your camp ( mostly ) with sonics. I've had a chequered history with solid-state bits in amps. I had exactly the same experience with a Gary Pimm CCS as an anode load in a phono. Wasn't natural-sounding after a few days - went back to a quality resistor. Tried a Mosfet source-follower to add 'drive' to a DHT driver, even with correct gate-stopper it came out after a couple of days. Batteries or LED's in cathodes ? - forget it.

Any change I've made of the years with a power amp has always HAD to preserve the midrange and treble tone, low level detail and overall realism. With the Verdier series-reg circuit, it was overall an improvement because the bass got a lot better and it helped to de-muddle the lower mids. Higher up it was a bit of a trade-off, I ended up with some series resistance on the output ( 1 or 2R ) and a quality cap after this ( to ground ) , this helped ( the return current at HF then goes back through the film cap ) .
The Shunt reg is the real daddy. Everything from low bass out the middle of the midrange improved again.
Do not be fooled into thinking it's really complicated . The bit on the right is just a 2-stage passive supply to get the ripple down, In mine I made this better than the Dick Olsher one and got about 2mV ripple at that point.
The network around the 6922 is just to set up and load the error-amplifier - hardly any current goes through that .
The relevant bit is the KT88 in triode-mode and its cathode by-pass cap. As long as the circuit demand stays within about +/-40mA, the only 'action' is really the current being modulated through the KT88. The passive bit of the supply on the right just sits there smoothing the rectifier ripple - it has no change of demand on it, hence is not really in the signal return loop.
I can guarantee you this supply has a better midrange than a passive supply, but with a rock-solid bass.

Currently I'm thinking about how I can upgrade it to a 700V supply for GM70's !
"Once you find out ... the Circumstances ; then you can go out"
Post Reply