bypassing elcap of feedback network
Hi, Dr H,
due to not so good experiences with "elcap sound" in the past, I soldered a polypropylene (MKP, Wima, 3.3uF) parallel to the elcap.
One of the problems with elcaps in that position is, that there they are operated at near 0 V or even in slight reverse (tens of mV) conditions. As the elcap is an electrolytic device, it is kept alive (to maintain its capacitance value) by the presence of some DC bias which re-forms the dielectric layer. The gradative worsening of the elcap may not be immediately apparent, but with time passing it is sonically noticeably. In my GB300D kits I shorted this elcap and built a DC servo loop amp on a small board using a OP97 IC opamp. Now, under all circumstances, DC output level remains in the 200uV area (due to offset values of this IC and no offset compensation used).
Some audio enthusiasts dislike this approach, but I find it a way better solution that living with a doubtful elcap.
If you would like to get the schematic, I will be glad to supply it .
ciao.
Dr.H said:
Hi, Dr H,
due to not so good experiences with "elcap sound" in the past, I soldered a polypropylene (MKP, Wima, 3.3uF) parallel to the elcap.
One of the problems with elcaps in that position is, that there they are operated at near 0 V or even in slight reverse (tens of mV) conditions. As the elcap is an electrolytic device, it is kept alive (to maintain its capacitance value) by the presence of some DC bias which re-forms the dielectric layer. The gradative worsening of the elcap may not be immediately apparent, but with time passing it is sonically noticeably. In my GB300D kits I shorted this elcap and built a DC servo loop amp on a small board using a OP97 IC opamp. Now, under all circumstances, DC output level remains in the 200uV area (due to offset values of this IC and no offset compensation used).
Some audio enthusiasts dislike this approach, but I find it a way better solution that living with a doubtful elcap.
If you would like to get the schematic, I will be glad to supply it .
ciao.
elcaps
Hi KLe,
of course, we also could discuss the matter of the 2 bootstrap elcaps (10uF) being replaced by some good foil caps. But in the bootstrap position the elcaps are well biased, so that the electrolytic dielectric is kept in good shape. This was exactly the weak point regarding the feedback elcap which acts as an AC short to gnd for the 1k-Ohm feedback network resistor.
ciao.
KLe said:
Hi KLe,
of course, we also could discuss the matter of the 2 bootstrap elcaps (10uF) being replaced by some good foil caps. But in the bootstrap position the elcaps are well biased, so that the electrolytic dielectric is kept in good shape. This was exactly the weak point regarding the feedback elcap which acts as an AC short to gnd for the 1k-Ohm feedback network resistor.
ciao.
Hi stoeffle
That feedback to ground electro is 47uF (as designed).
Doug Self recommended increasing the size of that electro to 250/400uF, to overcome the issues you have written about (especially the resulting low frequency non-linear distortions). Interestingly, I have seen some schematics with sizes of 1000uF for this electro.
stoeffle, what are your thoughts of this approach, that is increasing the size to 250/400uF or higher.
ciao
That feedback to ground electro is 47uF (as designed).
Doug Self recommended increasing the size of that electro to 250/400uF, to overcome the issues you have written about (especially the resulting low frequency non-linear distortions). Interestingly, I have seen some schematics with sizes of 1000uF for this electro.
stoeffle, what are your thoughts of this approach, that is increasing the size to 250/400uF or higher.
ciao
larger caps on NFB branch
Hi KLe,
basically, IMO no problem in increasing that cap's value.
The designed 47uF yields a 3dB cut-off freq of some 3.4Hz,
250uF 0.64Hz and 470uF 0.34Hz. As you see, from the audio point of view nothing thrilling. But, on switch-on, the DC offset (plopp! noise) may take longer to settle, until this cap is charged to its final voltage value. This may do some harm to your speakers unless you shorten the output leads during power-up sequence.
Placing a good foil cap in // with this electrolytic cap will, for sure, be a sonic improvement as the non-linearities of the latter are shorted out (at least reduced by large amounts).
ciao.
KLe said:Hi stoeffle
That feedback to ground electro is 47uF (as designed).
Doug Self recommended increasing the size of that electro to 250/400uF, to overcome the issues you have written about (especially the resulting low frequency non-linear distortions). Interestingly, I have seen some schematics with sizes of 1000uF for this electro.
stoeffle, what are your thoughts of this approach, that is increasing the size to 250/400uF or higher.
ciao
Hi KLe,
basically, IMO no problem in increasing that cap's value.
The designed 47uF yields a 3dB cut-off freq of some 3.4Hz,
250uF 0.64Hz and 470uF 0.34Hz. As you see, from the audio point of view nothing thrilling. But, on switch-on, the DC offset (plopp! noise) may take longer to settle, until this cap is charged to its final voltage value. This may do some harm to your speakers unless you shorten the output leads during power-up sequence.
Placing a good foil cap in // with this electrolytic cap will, for sure, be a sonic improvement as the non-linearities of the latter are shorted out (at least reduced by large amounts).
ciao.
Actually stoeffle
This amp will run without this electro, ie DC to ground. 1 or 2 people have done this and reported sound improvements. Though, I think Greg said that the amp is not quite as thermally stable, in this configuration. Does that sound right ...
Would removing it all together remove the other issues that we have been discussing.
This amp will run without this electro, ie DC to ground. 1 or 2 people have done this and reported sound improvements. Though, I think Greg said that the amp is not quite as thermally stable, in this configuration. Does that sound right ...
Would removing it all together remove the other issues that we have been discussing.
NFB network
Hi KLe,
first:
to reduce the plopp! noise at power on, this capacitor mus be reduced to its nominal designed value, that is 47uF. Any larger value will generate that turn-on-greeting-sound. To reduce capacitor non-linearities, you may add a 3.3uF MKS or MKP cap in //, as already stated in some post earlier. This litte add-on value will not cause a plopp! noise bigger than with the el-cap only.
second:
grounding the cold end of the 1K-Ohm NFB-network resistor will of course eliminate completely possible "cap-sound" problems, BUT, as Greg already stated, output offset may get thermally unstable as the closed loop gain of the amp now goes way down to and including DC (with the 47u cap in place you had a high-pass amplifier with a 3db cut-off freq around 3.4Hz; below 3.4Hz the gain would go to 0dB (x1), meaning 100% NFB at DC). So, if you want to have both a) no cap on the cold end leg and b) no thermal DC output offset problem, you should build a DC-servo. This servo will measure the DC offset at the output and inject a correction current into the inverting input of the amp, thus driving it to null the existing DC offset.
Hope this explanation helps.
ciao.
KLe said:Actually stoeffle
This amp will run without this electro, ie DC to ground. 1 or 2 people have done this and reported sound improvements. Though, I think Greg said that the amp is not quite as thermally stable, in this configuration. Does that sound right ...
Would removing it all together remove the other issues that we have been discussing.
Hi KLe,
first:
to reduce the plopp! noise at power on, this capacitor mus be reduced to its nominal designed value, that is 47uF. Any larger value will generate that turn-on-greeting-sound. To reduce capacitor non-linearities, you may add a 3.3uF MKS or MKP cap in //, as already stated in some post earlier. This litte add-on value will not cause a plopp! noise bigger than with the el-cap only.
second:
grounding the cold end of the 1K-Ohm NFB-network resistor will of course eliminate completely possible "cap-sound" problems, BUT, as Greg already stated, output offset may get thermally unstable as the closed loop gain of the amp now goes way down to and including DC (with the 47u cap in place you had a high-pass amplifier with a 3db cut-off freq around 3.4Hz; below 3.4Hz the gain would go to 0dB (x1), meaning 100% NFB at DC). So, if you want to have both a) no cap on the cold end leg and b) no thermal DC output offset problem, you should build a DC-servo. This servo will measure the DC offset at the output and inject a correction current into the inverting input of the amp, thus driving it to null the existing DC offset.
Hope this explanation helps.
ciao.
Hello,
I run my GB150D without any coupling cap since the beginning.(input and feedback) I first tried with the standard mounting but the offset wandered too much for my taste. Then I desoldered/resoldered one input transistor of each pair, so as to couple them thermically, face to face. This greatly enhenced the offset behaviour, +/-20mV max when hot, and is no more an issue with my speakers.
I made some tests with an 4,7µF MKP cap in the input and an 220µ/10V OSCON cap in the feedback path. Honestly, I couldn't notice any real difference. But as STOEFFLE told us, the feedback cap has not enough dc accros it, and things could sonically worsen with time. My caps are still on board but definitely shortened.
I recently saw on greg's forum that he's preparing a new release of his killer amp, with a front end made of an opamp that will nicely and definitely solve the offset problem.
Francis
I run my GB150D without any coupling cap since the beginning.(input and feedback) I first tried with the standard mounting but the offset wandered too much for my taste. Then I desoldered/resoldered one input transistor of each pair, so as to couple them thermically, face to face. This greatly enhenced the offset behaviour, +/-20mV max when hot, and is no more an issue with my speakers.
I made some tests with an 4,7µF MKP cap in the input and an 220µ/10V OSCON cap in the feedback path. Honestly, I couldn't notice any real difference. But as STOEFFLE told us, the feedback cap has not enough dc accros it, and things could sonically worsen with time. My caps are still on board but definitely shortened.
I recently saw on greg's forum that he's preparing a new release of his killer amp, with a front end made of an opamp that will nicely and definitely solve the offset problem.
Francis
I am convinced as well about the negative effects of the electrolytic cap to ground found in many amps. The problem can be due to dielectric absorbency in the cap. I have tried to remove
this cap in some amps and in my case they are thermal stable.
That isn´t the same as to claim they are always unnecessesary.
But DC-stability can eventually be done without a servo circuit or the blocking cap. My Dynamic Precision A1 amp has neither and is in fact DC-coupled but I don´t know how it´s done.
Generally speaking; the more of the gear in the chain that is DC coupled, the better bass reproduction is to be expected as phase shifts will be added to each other. But DC-coupling has its risks and if you use a record player as source a high-pass filter is a good thing to be considered.
this cap in some amps and in my case they are thermal stable.
That isn´t the same as to claim they are always unnecessesary.
But DC-stability can eventually be done without a servo circuit or the blocking cap. My Dynamic Precision A1 amp has neither and is in fact DC-coupled but I don´t know how it´s done.
Generally speaking; the more of the gear in the chain that is DC coupled, the better bass reproduction is to be expected as phase shifts will be added to each other. But DC-coupling has its risks and if you use a record player as source a high-pass filter is a good thing to be considered.
tinitus said:
Hi Tinitus,
Was your crossover placed at the input or at the output of your amp ? The currents involved at the input nodes of an amp are so low that I can't imagine how a sound deprivation could occur with a short nicely done.
Any idea of what can give such a negative effect ?
Francis
Re: NFB network
DC servo sounds really good, but, removing the electro or shorting it and thermally coupling the input transistor pair, sounds even easier.
stoeffle, yes, please send me the DC servo circuit diagram.
No feedback electro, let's see ...
1. Closed loop gain to DC
2. No capacitor non-linearities
3. No degradation of sonics with time
4. No plopp! noise (which could degradate speaker driver performance over time)
stoeffle said:
Thankyou stoeffle/rellumrellum said:
DC servo sounds really good, but, removing the electro or shorting it and thermally coupling the input transistor pair, sounds even easier.
stoeffle, yes, please send me the DC servo circuit diagram.
No feedback electro, let's see ...
1. Closed loop gain to DC
2. No capacitor non-linearities
3. No degradation of sonics with time
4. No plopp! noise (which could degradate speaker driver performance over time)
rellum said:
Yep, has some nice specs listed there, hope it sounds as good as it look.
Forgive my ignorance here, what are the pros and cons of DC coupling; degradation or destruction of the driver vs sonic improvement and what is a safe limit of DC drift or is that a silly question (I suspect so!). Is the sonic improvement worth the increased risk if something goes awry or is this as with all things audio a suck it and see scenario ?
Anyone done an a/b on standard gb150 to dc-coupled 150?
Cheers
Iain
Hi iihay
Yes, rellum's 150D is DC to ground (no feedback) electro, and he says it sounds better. If you do this, then you need to thermally couple the input transistor pair (back to back).
But, it is probably better that rellum explains the details, if he doesn't mind, and the difference in sound.
Yes, the new Concept amp lolls like it should be another Simple Killer Amp The specs are excellent.
Yes, rellum's 150D is DC to ground (no feedback) electro, and he says it sounds better. If you do this, then you need to thermally couple the input transistor pair (back to back).
But, it is probably better that rellum explains the details, if he doesn't mind
, and the difference in sound.Yes, the new Concept amp lolls like it should be another Simple Killer Amp
The specs are excellent.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.