I'm of the "no signal across the cap" persuasion and thus tend to use rather large ones, say 470-1000 uF. Rx, more properly called Rs, will be measured in milliohms for such caps. There's usually no problem with up to a volt of reverse voltage, but you can always use back to back caps or even run a high value resistor from the junction to a supply rail, to insure the caps are never reversed biased at all.
Linsley Hood's discussion was not over Rs. If you refer to http://www.low-esr.com/QT_LowESR.pdf his argument relates to Xs in figure 1.
While Linsley-Hood published audio designs in his day job was in the research laboratory of a maker that supplied electrical grade film to capacitor manufacturers. Part of his duties were to visit customers and test their product for various electrical behaviours.
Cyril Bateman published articles in Electronics and Wireless World in 2002 on the subject of capacitor sound. He offered boards for distortion measuring equipment that showed in general that plastic capacitors were better than electrolytic types.
Some electrolytics measured better than others and that might account for audiophile preferences for some brands. Black Gate electrolytics were tested-from memory they did not live up to their audiophile reputation.
I understand why the Silicon Chip example had appeal to you however your comment about "no signal across the cap" seems odd and perhaps not to be taken too literally. Can you share a bit more information on the supporting theory.
Michael Jonassen
While Linsley-Hood published audio designs in his day job was in the research laboratory of a maker that supplied electrical grade film to capacitor manufacturers. Part of his duties were to visit customers and test their product for various electrical behaviours.
Cyril Bateman published articles in Electronics and Wireless World in 2002 on the subject of capacitor sound. He offered boards for distortion measuring equipment that showed in general that plastic capacitors were better than electrolytic types.
Some electrolytics measured better than others and that might account for audiophile preferences for some brands. Black Gate electrolytics were tested-from memory they did not live up to their audiophile reputation.
I understand why the Silicon Chip example had appeal to you however your comment about "no signal across the cap" seems odd and perhaps not to be taken too literally. Can you share a bit more information on the supporting theory.
Michael Jonassen
Hello Andrew,
Due attention to that detail is given by most designers, although that may not be evident to a handyperson making simple modifications to his/her amplifier.
It is difficult to encapsulate the thinking in a neat and succinct way. I can see you have done that to keep things simple.
There is a subtle difference between your point and the one I was attempting to make.
There has to be an albeit attenuated a.c. signal from the amplifier output appearing across the NFB dc blocking cap for a series feedback amplifier with any gain.
In a conventional series feedback amplifier, the output signal arrives at the inverting input terminal - part of the differential stage that sums (mixes) all connected inputs.
That from the NFB feedback resistor from the output is one,as is the decoupling arm to ground.
Electrolytic capacitors have their shortcomings so deficiencies in the decoupling arm (made small by the network attenuation) will appear as an a.c. input at the summing terminal.
A sufficiently large value capacitor there will help to swamp out conflicts with input signals. However the physical characteristics of capacitors are also important according to Linsley-Hoods argument.
Linsley-Hood had friends who drew attention to subjective benefits of plastic capacitors and he used these in NFB decoupling arms in his later designs in preference to the electrolytic variety.
He would not have gone to the trouble without investigation as plastic caps are both large and expensive. The references I supplied give a supporting argument.
I am not trying to persuade anyone to use plastic caps to replace electrolytics in the NFB dc blocking position.
A decent size one is needed in this case and an electrolytic is the only practical choice.
A 1000 uF will give a lower cutoff, however the amplifier in question appears to use a non-polar part and I doubt if these are obtainable for values over 470uF.
If I am saying anything it is to try a selection of brands and types of electrolytic cap
to see if you have any preferences.
Michael Jonassen
Due attention to that detail is given by most designers, although that may not be evident to a handyperson making simple modifications to his/her amplifier.
It is difficult to encapsulate the thinking in a neat and succinct way. I can see you have done that to keep things simple.
There is a subtle difference between your point and the one I was attempting to make.
There has to be an albeit attenuated a.c. signal from the amplifier output appearing across the NFB dc blocking cap for a series feedback amplifier with any gain.
In a conventional series feedback amplifier, the output signal arrives at the inverting input terminal - part of the differential stage that sums (mixes) all connected inputs.
That from the NFB feedback resistor from the output is one,as is the decoupling arm to ground.
Electrolytic capacitors have their shortcomings so deficiencies in the decoupling arm (made small by the network attenuation) will appear as an a.c. input at the summing terminal.
A sufficiently large value capacitor there will help to swamp out conflicts with input signals. However the physical characteristics of capacitors are also important according to Linsley-Hoods argument.
Linsley-Hood had friends who drew attention to subjective benefits of plastic capacitors and he used these in NFB decoupling arms in his later designs in preference to the electrolytic variety.
He would not have gone to the trouble without investigation as plastic caps are both large and expensive. The references I supplied give a supporting argument.
I am not trying to persuade anyone to use plastic caps to replace electrolytics in the NFB dc blocking position.
A decent size one is needed in this case and an electrolytic is the only practical choice.
A 1000 uF will give a lower cutoff, however the amplifier in question appears to use a non-polar part and I doubt if these are obtainable for values over 470uF.
If I am saying anything it is to try a selection of brands and types of electrolytic cap
to see if you have any preferences.
Michael Jonassen
Find the correct value first. Try it, both measurements and listening. Only after that is it worth building a second to make the exotics comparison.
My finding, although I have not spent the money on many exotics, is that with the correct value in place, I cannot measure, nor hear, a difference by swapping in different makes, or types of electrolytics.
My finding, although I have not spent the money on many exotics, is that with the correct value in place, I cannot measure, nor hear, a difference by swapping in different makes, or types of electrolytics.
How does this discussion apply to amps with direct coupled inputs? I'm currently working on a GFA-555, which has no input coupling cap, and a 47uF/100V NFB cap. The R is 1K, which gives a 3.4Hz cutoff.
If I understand the previous posts correctly, because of the lack of an input cap there will be signal across the NFB cap therefore it should be a bipolar type?
If I understand the previous posts correctly, because of the lack of an input cap there will be signal across the NFB cap therefore it should be a bipolar type?
Does it have a Source with a direct coupled output?
The DC blocking cap does not need to be in the Power amplifier. The DC blocker performs just as well when located in the Source equipment.
Bipolar is probably never appropriate.
There is a small DC bias across the NFB cap. The AC signal modulates that DC bias. The cap should never see more than a few hundred mV of either polarity when the input DC blocker is omitted AND there is no significant DC offset presented to the amplifier.
Most polar electrolytics are OK with upto ~1000mV of reverse bias.
Inverse parallel diodes can be added in parallel to the NFB cap to protect the cap from damage if the amplifier goes faulty or misbehaves.
The DC blocking cap does not need to be in the Power amplifier. The DC blocker performs just as well when located in the Source equipment.
Bipolar is probably never appropriate.
There is a small DC bias across the NFB cap. The AC signal modulates that DC bias. The cap should never see more than a few hundred mV of either polarity when the input DC blocker is omitted AND there is no significant DC offset presented to the amplifier.
Most polar electrolytics are OK with upto ~1000mV of reverse bias.
Inverse parallel diodes can be added in parallel to the NFB cap to protect the cap from damage if the amplifier goes faulty or misbehaves.
Last edited:
The big disadvantage with locating the DC blocker in the Source is that the Source designer does not know what equipment it will have to drive.
Good designer will specify the optimum load impedance.
Some will fit a low value cap to save money.
Some will fit a plastic cap to improve quality.
Some will fit an electrolytic to save money.
Some will fit a big plastic cap that will allow good bass signal delivery with many/most loads.
Few if any of these caps will have the correct value to match up with the requirement to keep the LOAD NFB cap without significant AC signal.
for example.
I have a SS (FET) Conrad Pre-Amp. It has no electrolytics in it. The output DC blocking caps are 10uF polypropylene. These will pass extended bass into a 100k load. but limited bass into a 10k load. But the frequency response variation has nothing to do with what appears on the NFB cap.
The only person that can control both the passband of the Power Amplifier and the NFB cap AC signal is the power amp designer.
Good designer will specify the optimum load impedance.
Some will fit a low value cap to save money.
Some will fit a plastic cap to improve quality.
Some will fit an electrolytic to save money.
Some will fit a big plastic cap that will allow good bass signal delivery with many/most loads.
Few if any of these caps will have the correct value to match up with the requirement to keep the LOAD NFB cap without significant AC signal.
for example.
I have a SS (FET) Conrad Pre-Amp. It has no electrolytics in it. The output DC blocking caps are 10uF polypropylene. These will pass extended bass into a 100k load. but limited bass into a 10k load. But the frequency response variation has nothing to do with what appears on the NFB cap.
The only person that can control both the passband of the Power Amplifier and the NFB cap AC signal is the power amp designer.
Right, I'm getting it.
The actual cutoff frequencies don't matter, as long as they are low enough to not affect the audible range. What does matter is that the NFB RC has a lower cutoff frequency than the input (or source output) RC. Since the low frequency cutoff point will then be defined by the input RC, there will be no AC signal blocked by the NFB cap.
On design lineage - the designer may have been Nelson Pass who did some work for Adcom - this from an article on modifying the GFP-555 in audioXpress issue of Sept 2006.
The GFP-555 is a preamplifier probably from the same product series as Stormriders GFA-555.
The Author of the article was Charles Hansen a member who is the owner/designer of a high end brand.
The information is a little bit hearsay based on an enquiy to Victor Campos the current designer at Adcom at the time. His recollection was that the GFP-555 was probably designed by Nelson Pass before he (Campos) arrived at Adcom.
A block diagram of the original preamp shows two outputs lab and normal. here is a blocking cap on normal output but none on lab. The value of the blocking cap is not given.
Michael Jonassen
The GFP-555 is a preamplifier probably from the same product series as Stormriders GFA-555.
The Author of the article was Charles Hansen a member who is the owner/designer of a high end brand.
The information is a little bit hearsay based on an enquiy to Victor Campos the current designer at Adcom at the time. His recollection was that the GFP-555 was probably designed by Nelson Pass before he (Campos) arrived at Adcom.
A block diagram of the original preamp shows two outputs lab and normal. here is a blocking cap on normal output but none on lab. The value of the blocking cap is not given.
Michael Jonassen
Needless to say I would expect to find a d.c. blocking capacitor in the line feed from the preamplifier to the power amplifier input in the GFA-555. Another contributor to audioXpress Gary Galo has written articles on modifying other Adcom gear. Whether this includes power amp sections I cannot say - I bought selected issues of the publication off the shelf at an electronics hobby shop and I can relate only from the issue mentioned in my previous post.
Michael Jonassen
Michael Jonassen
I found a schematic for the GTP500 at audio-circuit.dk, but it was illegible. They also had a schematic for the GFP565, which was a good scan. It's normal outputs had a 4.7uF cap, with a 47.5K to ground. This gives a cutoff of 0.7 Hz. Too low for the NFB RC in the GFA555.
The LAB outputs are direct coupled to the output of the "Adcom 1A" IC, which is a five pin TO220 package.
The LAB outputs are direct coupled to the output of the "Adcom 1A" IC, which is a five pin TO220 package.
Hello Stormrider,
I see elsewhere on DIYaudio someone emailed Adcom and they sent him a pdf for your amp.
There was one given in the thread and it had a d.c. servo amplifier - basically a low-pass amplifier which samples the d.c. output level and feeds this back into the amplifier.
If yours has one of these then you don't want to mess around with the power amp capacitor values.
Michael Jonassen
I see elsewhere on DIYaudio someone emailed Adcom and they sent him a pdf for your amp.
There was one given in the thread and it had a d.c. servo amplifier - basically a low-pass amplifier which samples the d.c. output level and feeds this back into the amplifier.
If yours has one of these then you don't want to mess around with the power amp capacitor values.
Michael Jonassen
The schematic for both of my Mk. 1 GFA-555's is easily downloaded here:
Categorized Schematics and Service Manuals for free download
Mine don't have the regulated power supply section on the second page of the .pdf.
As you can see there, there is nothing but a series 1K, shunted by a 22.1K and 300pF, to the base of Q1. You must be looking at the GFA-555II, which does have an input cap and a DC servo.
Categorized Schematics and Service Manuals for free download
Mine don't have the regulated power supply section on the second page of the .pdf.
As you can see there, there is nothing but a series 1K, shunted by a 22.1K and 300pF, to the base of Q1. You must be looking at the GFA-555II, which does have an input cap and a DC servo.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.