Hi All,
I am interested to know how one determines if a zobel network is required. I currently run my lm3875 gainclone without one and am very happy with the sound. There is no audible sign of any problems.
Firstly, in your experience is a zobel often required with a gainclone amplifier? And secondly, does not running a zobel risk damaging my amplifier and/or speakers? I prefer the idea of not adding any circuitry that is not required, but would obviously add the zobel if damaging the amplifier/speakers was a possibility.
Would I be able to hear any problems caused by not using one, or are the problems purely in the ultrasonic spectrum?
Your advice is appreciated.
Regards,
Greg.
I am interested to know how one determines if a zobel network is required. I currently run my lm3875 gainclone without one and am very happy with the sound. There is no audible sign of any problems.
Firstly, in your experience is a zobel often required with a gainclone amplifier? And secondly, does not running a zobel risk damaging my amplifier and/or speakers? I prefer the idea of not adding any circuitry that is not required, but would obviously add the zobel if damaging the amplifier/speakers was a possibility.
Would I be able to hear any problems caused by not using one, or are the problems purely in the ultrasonic spectrum?
Your advice is appreciated.
Regards,
Greg.
Last edited:
Not an easy one to answer. Some amps will run quite happily without, and needing one is definitely not a sign of a problem amp.
If you asked me if a chip amp needed one I would say yes yes yes.
It helps with stability and provides a known impedance at hf... so much depends on the phase margins and loop gain of the amp in question. And the load it works into.
As to values, what you see in any amp, 0.1uf in series with 10 ohms is fine.
You would need to test your amp and look at squarewave performance etc and stability into capacitive loads,,, which may be destructive. Your amp may be fine or running on the verge of instability... impossible to say. Very important to connect the Zobel return to the correct ground so it doesn't cause a problem rather than help stop one.
If you asked me if a chip amp needed one I would say yes yes yes.
It helps with stability and provides a known impedance at hf... so much depends on the phase margins and loop gain of the amp in question. And the load it works into.
As to values, what you see in any amp, 0.1uf in series with 10 ohms is fine.
You would need to test your amp and look at squarewave performance etc and stability into capacitive loads,,, which may be destructive. Your amp may be fine or running on the verge of instability... impossible to say. Very important to connect the Zobel return to the correct ground so it doesn't cause a problem rather than help stop one.
Agree with Mooly. Solid state amps, being mostly a constant voltage source, don't usually get too upset. ( If they do, build a better amp) Tube amps can. The Zobel is more often used as part of the crossover circuit closest to the driver to make the filter network more predictable. Amp load is a side effect.
Linkwitz has a spreadsheet that helps, but I have found only testing to be accurate. I have found needing caps more than twice the calculated value, and resistors plus or minus 50%. So, if you don't have something like a WT-2 to measure the impedance, you are working blind.
Linkwitz has a spreadsheet that helps, but I have found only testing to be accurate. I have found needing caps more than twice the calculated value, and resistors plus or minus 50%. So, if you don't have something like a WT-2 to measure the impedance, you are working blind.
Yes... somewhere where it has no influence... which if the PCB is correct will be the speaker return point.
The zobel that fits one may be grossly wrong for another. It pays to look carefully at this, or you may have unexplainable failures with IC amps.
0.1U and 10R as a standard cure is no cure.
& here is why:
IC Audio Power Amplifiers and Zobel Networks: One Size Does Not Fit All | Digital Home DesignLine
Another matter is that the standard loudspeaker cable and loudspeaker termination can have some pretty wild resonances in the low MHz range, killing any kind of amps like flies.. A HF load like the zobel may help.
But
0.1U and 10R as a standard cure is no cure.
& here is why:
IC Audio Power Amplifiers and Zobel Networks: One Size Does Not Fit All | Digital Home DesignLine
Another matter is that the standard loudspeaker cable and loudspeaker termination can have some pretty wild resonances in the low MHz range, killing any kind of amps like flies.. A HF load like the zobel may help.
But
Does it matter what type of capacitor is used for an LM3875s zobel network? TI make no mention of the type in their application notes...
A member here did write the following a while back:
http://www.diyaudio.com/forums/chip-amps/157692-gainclone-transformer.html#post2032919
"The capacitor for speaker output zobel (output RC) included with the chipamp.com kit is the wrong type. An inexpensive little polyester/mylar dip cap is typically used for zobel, not high-efficiency polypropylene, so if you get poor quality upper treble, then replace that little cap"
Is this really the case? I have a small 0.1uF polypropylene cap here which came with my Audiosector Gainclone kit, and wondered if I should use it in light of the above quote...
John.
A member here did write the following a while back:
http://www.diyaudio.com/forums/chip-amps/157692-gainclone-transformer.html#post2032919
"The capacitor for speaker output zobel (output RC) included with the chipamp.com kit is the wrong type. An inexpensive little polyester/mylar dip cap is typically used for zobel, not high-efficiency polypropylene, so if you get poor quality upper treble, then replace that little cap"
Is this really the case? I have a small 0.1uF polypropylene cap here which came with my Audiosector Gainclone kit, and wondered if I should use it in light of the above quote...
John.
Zobel are funny - if they work , the amp never oscillates - then they see next to no current, power
but you need to size them as if they will see many cycles of full output swing oscillation at any frequency up to nearly the chip amp GBW
and they need to be low inductance - extended metal foil construction caps would be best - dielectric quality way secondary
the damping R shouldn't be inductive wirewound - carbon composition works fine though in this position
but you need to size them as if they will see many cycles of full output swing oscillation at any frequency up to nearly the chip amp GBW
and they need to be low inductance - extended metal foil construction caps would be best - dielectric quality way secondary
the damping R shouldn't be inductive wirewound - carbon composition works fine though in this position
This is the type the kit came with - assume this is OK to use:
BFC241641004 Vishay BC Components | BC2054-ND | DigiKey
I think I have have 2R7 carbon comp. in the parts box. Will go check.
Thanks
Values aside, I would add that it is supremely important that the zobel goes back to signal ground for all sorts of reasons - that it is not put across the speaker terminals. One simple reason is the inductance of the additional path to ground. Another is that any sort of shared path is a bad idea. In purist terms, if you send it directly back to ground you have simply added two components (perhaps 3 including the L). Across the terminals you have changed the topology of the circuit hugely. R & L to terminals + C to Gnd, same on the way back with an RL and C in between.
Chris,
I think you explain it badly.
The Zobel is an HF load. The series capacitor ensures it can only be an HF load.
HF signals do not pass easily through inductors as loads .
The first priority for an amplifier output Zobel is that it must be capable of loading the output stage at HF. This demands that the decoupling ground on the PCB that is wrapped around the output devices and the supply pins is the RETURN point for the output Zobel.
If you want to analyse the layout effectiveness, then look at your layout like this.
Remove the main PSU. Remove the MF decoupling.
This leaves the HF decoupling as the ONLY source of current to feed the output devices. Draw the route that the HF transients must follow from decoupling to output pin/terminal. Now add on the HF Zobel load.
Where is the lowest impedance route for the HF load?
In my view it can only be from the output pin to the Decoupling Ground. That whole group of power input and power ground and output devices and Zobel and HF Decoupling must be laid out in a very tight and compact layout that minimises trace lengths to make the HF route for least inductance.
Finally, I think too many layouts concentrate on the inputs at the expense of what needs to happen at the output.
Since the output is the biggest stage, it should become the major layout target. Only after the output devices are laid out to minimise inductance is it then time to add on the earlier stages to optimise them.
I think you explain it badly.
The Zobel is an HF load. The series capacitor ensures it can only be an HF load.
HF signals do not pass easily through inductors as loads .
The first priority for an amplifier output Zobel is that it must be capable of loading the output stage at HF. This demands that the decoupling ground on the PCB that is wrapped around the output devices and the supply pins is the RETURN point for the output Zobel.
If you want to analyse the layout effectiveness, then look at your layout like this.
Remove the main PSU. Remove the MF decoupling.
This leaves the HF decoupling as the ONLY source of current to feed the output devices. Draw the route that the HF transients must follow from decoupling to output pin/terminal. Now add on the HF Zobel load.
Where is the lowest impedance route for the HF load?
In my view it can only be from the output pin to the Decoupling Ground. That whole group of power input and power ground and output devices and Zobel and HF Decoupling must be laid out in a very tight and compact layout that minimises trace lengths to make the HF route for least inductance.
Finally, I think too many layouts concentrate on the inputs at the expense of what needs to happen at the output.
Since the output is the biggest stage, it should become the major layout target. Only after the output devices are laid out to minimise inductance is it then time to add on the earlier stages to optimise them.
I'm sorry to say but that article is pretty much bo***cks. The zobel has nearly nothing to do with compensating the inductance of a driver. To do so would require a uF or two even for a tweeter. It is much more what you said about the impedance of cables and out of audio band Zs. In short, it gives a largely resistive load for the amplifier to look at above a certain frequency, usually starting at 70 or 80kHz so that it can be flat when getting into the tricky regions.
Cordell has some rather marvellous graphs of what the impedance can look like and even says that the impedance of a loudspeaker in the MHz range is anyone's guess. For those who don't have the book, those graphs look like a mountain range!
A Member posted that link to that SPEAKER correction paper.
It was clearly discussing SPEAKER correction and it's effect on the amplifier.
An amplifier OUTPUT Zobel is a completely different animal.
The amp Zobel is what the AMPLIFIER needs to see as an HF load to maintain stability if there is NO LOAD at the speaker terminals.
If one wants good load tolerance from the amplifier it is usually addressed by converting the Output Zobel to an Output Thiele Network. Part of that Output Thiele Network can be located on the amplifier's speaker terminals. Cherry discusses this at length.
It was clearly discussing SPEAKER correction and it's effect on the amplifier.
An amplifier OUTPUT Zobel is a completely different animal.
The amp Zobel is what the AMPLIFIER needs to see as an HF load to maintain stability if there is NO LOAD at the speaker terminals.
If one wants good load tolerance from the amplifier it is usually addressed by converting the Output Zobel to an Output Thiele Network. Part of that Output Thiele Network can be located on the amplifier's speaker terminals. Cherry discusses this at length.
Andrew
Part 1. I think we are on the same track here, though I'm at a loss to understand why your explanation is clearer than mine. I think what you are doing is introducing the power supply impedance as a necessary part of the circuit (and I do agree with you here in general) and saying that really the only low impedance (or dominant) route, at the frequencies we are concerned with, is from the local bypass caps. This is probably true, though the extent would vary with the type of power supply (in a normal linear supply the impedance will eventually be dominated by the ESL of the reservoir caps, and you are saying that it is important to keep this circuit tight. This is probably generally correct, notwithstanding that some of us join these two caps up before taking them back to the main ground.
On the more complicated networks I'm with you there too. There is a good argument for having something across the terminals and reducing the amount of RF that comes back in. I'm just saying that it's not a good place to put a standard zobel, even though it looks pretty harmless at first glance.
Part 1. I think we are on the same track here, though I'm at a loss to understand why your explanation is clearer than mine
. I think what you are doing is introducing the power supply impedance as a necessary part of the circuit (and I do agree with you here in general) and saying that really the only low impedance (or dominant) route, at the frequencies we are concerned with, is from the local bypass caps. This is probably true, though the extent would vary with the type of power supply (in a normal linear supply the impedance will eventually be dominated by the ESL of the reservoir caps, and you are saying that it is important to keep this circuit tight. This is probably generally correct, notwithstanding that some of us join these two caps up before taking them back to the main ground.On the more complicated networks I'm with you there too. There is a good argument for having something across the terminals and reducing the amount of RF that comes back in. I'm just saying that it's not a good place to put a standard zobel, even though it looks pretty harmless at first glance.
We are on the same wavelength.
Maybe Members will read both versions and come to understand what is important.
Thiele's original has two options.
R+C to ground before the series connected R//L, where some values are either open circuit or short circuit.
R//L in series followed by R+C to ground. Again some component values are zero/open.
Both these work and Cherry's paper is well worth reading to gain a fuller understanding of the infinite series of component values that can be used.
Again we are in agreement that an EXTRA Zobel across the amplifier's Speaker Terminals offers some extra RF attenuation. Cherry's alternative NFB tapping point improves this even further. I started a Thread many years ago showing this, that developed into a grounding topic.
Maybe Members will read both versions and come to understand what is important.
Thiele's original has two options.
R+C to ground before the series connected R//L, where some values are either open circuit or short circuit.
R//L in series followed by R+C to ground. Again some component values are zero/open.
Both these work and Cherry's paper is well worth reading to gain a fuller understanding of the infinite series of component values that can be used.
Again we are in agreement that an EXTRA Zobel across the amplifier's Speaker Terminals offers some extra RF attenuation. Cherry's alternative NFB tapping point improves this even further. I started a Thread many years ago showing this, that developed into a grounding topic.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.