This was actually a pretty fun thread. I classify gaincloning like other crazy tweaks, homemade magic bricks, interconnects made from magent wire, freezing CD's, etc. They're reasonably inexpensive to experiment with and they're fun. So, perhaps you could take your arguments elsewhare and we'll get back to gaincloning for fun. That's why we're in this hobby right? I'm sure there are other forums for arguing. You could even exchange emails and have a good one-on-one.
Now to the real topic. Would an input follower like (ZENV4) work as an input buffer for paralleling these devices?
Now to the real topic. Would an input follower like (ZENV4) work as an input buffer for paralleling these devices?
An input follower like (ZENV4) will work as an input buffer for paralleling these devices as long as you use an output cap ( for isolating DC from the input of the LM3875 ) and "dividing" resistors. A BS250 will also work and according to those that built the ZENV4 it sounds better than the ZVP device.
You are right about going back to the subject. Life is too short for arguing.
You are right about going back to the subject. Life is too short for arguing.
Goodmorning everybody,
The most strange thing happend yestday late at night: I build myself a gainclone. I couldn't resist anymore:
Thus:
I had a +/-30 supplylying around, simple 4 diode rectifier, 3000uF capacitance each rail.
I also had a 15cm/5cm/4cm heatsink lying around.
I allready had 4 LM3886T's for some time now, I think I bought these right after Rod Elliot published his project. I though that these were the same as the LM3875T, but selected for better specs and power (too bad).
Anyways: used a 10K muting resistor that the LM3875 doesn't need, used a 220K feedback resistor and 10K input (Vin-) resistor and a 10K to ground instead of a volume pot, Vin+ directly to groud, no caps, nowhere, so DC coupled. Some WBT RCA connectors, some OK interconnectors and my standard 6Ohm 2-way speakers. So that's 4 resistors (metalfilm), some wire and connectors and that's all
I got 17.8mV/28,3mV offset at output and it is dead quit without input, absolutely no humm (all grounds go to pin 7). Then the music started playing: Nice, but to my ears a little cold, hi-hats of the drumkit all sound alike, voices are accurate and intimite but lack warmth and are a bit distanced, there's plenty reverb but again it's cold reverb. Bass is well defined and strong but has less texture than I'm used to. All in all after some 3 hours of listening I still enjoy my JLH 10W Class A better.
I used cheap part that were lying around, soldered everthing neatly within two hours and I am very pleased with the results.
Maybe the difference in sound is pleasing some people, but it is not the Highest-Fi. Maybe much like I like bordeaux wine, but to avoid making that boring, now and then I also drink some cheap italien, chili etc. It makes me allways come back to bordeaux....
gr,
Thijs
The most strange thing happend yestday late at night: I build myself a gainclone. I couldn't resist anymore:
Thus:
I had a +/-30 supplylying around, simple 4 diode rectifier, 3000uF capacitance each rail.
I also had a 15cm/5cm/4cm heatsink lying around.
I allready had 4 LM3886T's for some time now, I think I bought these right after Rod Elliot published his project. I though that these were the same as the LM3875T, but selected for better specs and power (too bad).
Anyways: used a 10K muting resistor that the LM3875 doesn't need, used a 220K feedback resistor and 10K input (Vin-) resistor and a 10K to ground instead of a volume pot, Vin+ directly to groud, no caps, nowhere, so DC coupled. Some WBT RCA connectors, some OK interconnectors and my standard 6Ohm 2-way speakers. So that's 4 resistors (metalfilm), some wire and connectors and that's all
I got 17.8mV/28,3mV offset at output and it is dead quit without input, absolutely no humm (all grounds go to pin 7). Then the music started playing: Nice, but to my ears a little cold, hi-hats of the drumkit all sound alike, voices are accurate and intimite but lack warmth and are a bit distanced, there's plenty reverb but again it's cold reverb. Bass is well defined and strong but has less texture than I'm used to. All in all after some 3 hours of listening I still enjoy my JLH 10W Class A better.
I used cheap part that were lying around, soldered everthing neatly within two hours and I am very pleased with the results.
Maybe the difference in sound is pleasing some people, but it is not the Highest-Fi. Maybe much like I like bordeaux wine, but to avoid making that boring, now and then I also drink some cheap italien, chili etc. It makes me allways come back to bordeaux....
gr,
Thijs
Cold sound
Apparently you are using the inverted version, from the parts you describe.
If you do you can try three things that might improve the sound:
1) Use a 100K multiturn pot from non-inverting pin to ground to null the DC offset you describe. That DC seems to cause that cold impersonal sound you describe.
2) Bypass the output resistor, if you used it.
3) Force biasing into class A, using a 15K/2W resistor from output to V-, as was suggested on the Forum.
These changes seem to have brought audio improvements, according to the gainclone site. Even if the chip used is the 3875 instead of the 3886.
Carlos
tschrama said:I allready had 4 LM3886T's for some time now, I think I bought these right after Rod Elliot published his project. I though that these were the same as the LM3875T, but selected for better specs and power (too bad).
Anyways: used a 10K muting resistor that the LM3875 doesn't need, used a 220K feedback resistor and 10K input (Vin-) resistor and a 10K to ground instead of a volume pot, Vin+ directly to groud, no caps, nowhere, so DC coupled. Some WBT RCA connectors, some OK interconnectors and my standard 6Ohm 2-way speakers. So that's 4 resistors (metalfilm), some wire and connectors and that's all
I got 17.8mV/28,3mV offset at output and it is dead quit without input, absolutely no humm (all grounds go to pin 7). Then the music started playing: Nice, but to my ears a little cold, hi-hats of the drumkit all sound alike, voices are accurate and intimite but lack warmth and are a bit distanced, there's plenty reverb but again it's cold reverb. Bass is well defined and strong but has less texture than I'm used to. All in all after some 3 hours of listening I still enjoy my JLH 10W Class A better.
I used cheap part that were lying around, soldered everthing neatly within two hours and I am very pleased with the results.
Maybe the difference in sound is pleasing some people, but it is not the Highest-Fi. Maybe much like I like bordeaux wine, but to avoid making that boring, now and then I also drink some cheap italien, chili etc. It makes me allways come back to bordeaux....
Apparently you are using the inverted version, from the parts you describe.
If you do you can try three things that might improve the sound:
1) Use a 100K multiturn pot from non-inverting pin to ground to null the DC offset you describe. That DC seems to cause that cold impersonal sound you describe.
2) Bypass the output resistor, if you used it.
3) Force biasing into class A, using a 15K/2W resistor from output to V-, as was suggested on the Forum.
These changes seem to have brought audio improvements, according to the gainclone site. Even if the chip used is the 3875 instead of the 3886.
Carlos
Also some remarks from my side:
1. Try a 1.5 K 2W resistor from the output to the negative supply. Makes the amp a little smoother. So not 15 K !
2. Be patient, the character of the amp will change after the burn-in period
3. I doubt if the removal of the output resistor results in a smoother sound, my experiences are the opposite. (the amp gets more transparant, so you hear more artefacts on the cd's).
4. Yesterday I talked with Jean-Paul on the phone. He suggested a RC filter on the input to get rid of some HF crap. For example 1 K with 1.1 nF (polysterene, styroflex etc.) to earth. This has to be placed between the input and the pot. I did not try it yet, maybe today I will.
5. If you use normal diodes you could try snubbers to get rid of some HF noise on the supply. The power supply rejection rate of the opamps is worse for higher frequencies.
BTW: my amp really sounds different during different times of the day. At night the amp is much smoother. Maybe I should try some filtering on the supply, but I'm afraid that that will result in slower bass. Thorsten suggested a DC-killer.
Good luck,
Fedde
1. Try a 1.5 K 2W resistor from the output to the negative supply. Makes the amp a little smoother. So not 15 K !
2. Be patient, the character of the amp will change after the burn-in period
3. I doubt if the removal of the output resistor results in a smoother sound, my experiences are the opposite. (the amp gets more transparant, so you hear more artefacts on the cd's).
4. Yesterday I talked with Jean-Paul on the phone. He suggested a RC filter on the input to get rid of some HF crap. For example 1 K with 1.1 nF (polysterene, styroflex etc.) to earth. This has to be placed between the input and the pot. I did not try it yet, maybe today I will.
5. If you use normal diodes you could try snubbers to get rid of some HF noise on the supply. The power supply rejection rate of the opamps is worse for higher frequencies.
BTW: my amp really sounds different during different times of the day. At night the amp is much smoother. Maybe I should try some filtering on the supply, but I'm afraid that that will result in slower bass. Thorsten suggested a DC-killer.
Good luck,
Fedde
Hi All,
Thanks for the suggestions. I also can imagine that the whole setup may benifit from some "burn in time". The nice thing is that this amp seem really stable and reliable, no real heat produce, no noices, quite toriod tranny, it is all very nice, so I left the thing (look like two dead bugs on a heatsink lying on my floor ) on the whole night. Offset is spot on the same as yesterday evening.
About the 15K or 1.5K class A bias resistor to V-. Hmmmm, I don't buy that. 15K at 30V would draw 2mA, that's not enough for a headphone amp to bias (my headphone amp is biased at 250mA) and would change allmost nothing to the 26mA bias the output stage is normally operating in. Using a 1K would draw 30mA from the output. However the effect is that the lower part of the outputstage is biased at around 0 (ZERO) mA, While the bias of the upper part is not changed at all.
That's right, this would change a class AB amp (27mA bias is about 3mW class A, after that class B) into a class B amp whitout bias!
30mA of bias is no enough by a long way for any apriciateble class A power output! There has been plenty of articles about the use and useless increase of bias current (JLH, Self, Slone etc).
I really think we should take a closer look at the schematic of the datasheet and post some educated remarks.
But somehow nobody take this gaincloning thing serious and just enjoys the hype.
gr,
Thijs
Thanks for the suggestions. I also can imagine that the whole setup may benifit from some "burn in time". The nice thing is that this amp seem really stable and reliable, no real heat produce, no noices, quite toriod tranny, it is all very nice, so I left the thing (look like two dead bugs on a heatsink lying on my floor
) on the whole night. Offset is spot on the same as yesterday evening. About the 15K or 1.5K class A bias resistor to V-. Hmmmm, I don't buy that. 15K at 30V would draw 2mA, that's not enough for a headphone amp to bias (my headphone amp is biased at 250mA) and would change allmost nothing to the 26mA bias the output stage is normally operating in. Using a 1K would draw 30mA from the output. However the effect is that the lower part of the outputstage is biased at around 0 (ZERO) mA, While the bias of the upper part is not changed at all.
That's right, this would change a class AB amp (27mA bias is about 3mW class A, after that class B) into a class B amp whitout bias!
30mA of bias is no enough by a long way for any apriciateble class A power output! There has been plenty of articles about the use and useless increase of bias current (JLH, Self, Slone etc).
I really think we should take a closer look at the schematic of the datasheet and post some educated remarks.
But somehow nobody take this gaincloning thing serious and just enjoys the hype.
gr,
Thijs
Hi fedde
be very careful with that RC filter...the output impedance of your pot will add up to r part of your rc filter... so you have to be very careful when choosing the cutoff frequency of your filter and remember that when you change the volume, the output impedance of your pot will change and so the cutoff frequency of your filter. Here is the formula for the guys that want to play whit this mod and don’t know how to calculate an RC filter: c=1/f*2*3.14*r
Regards
be very careful with that RC filter...the output impedance of your pot will add up to r part of your rc filter... so you have to be very careful when choosing the cutoff frequency of your filter and remember that when you change the volume, the output impedance of your pot will change and so the cutoff frequency of your filter. Here is the formula for the guys that want to play whit this mod and don’t know how to calculate an RC filter: c=1/f*2*3.14*r
Regards
tschrama said:
30 ma gives class A operation to about 10 milliwatts. Enough
for headphones.
I too have read lots of discussion about increasing bias current,
but they tend to dwell on the measurements, noting either
failure to measure much better, or measuring even worse.
From listening tests, it is my experience that greater bias
sounds better, measurements and theory aside.
Dear Nelson Pass,
My headphones (32 ohm Grado's) draws about 5mA rms at (very) high volume levels, so 30mA bias is enough. But for some linearity or constant transductiance 30mA is barely enough, that's why I use a much hihger bias point. Even for a headphone amplifier I would advice (much) more than 30mA.
My point more specific was:
1] 30mA is little in to 8 Ohm (7.2mW peak about 5mW rms) that I think (!) it can never make any difference.
2] Accoridng to my simmulations of the topology, the 'bias resistor to V-' method decreases(!) bias current through the output stage. The opposite of the claimed effect!
3] From point 1] and 2] I conclude that nobody take the effort to make some educated remarks.
But I would like to add that I haven't tried the 'bias resistor to V-' mod and therefore I cann't comment on the claimed increase sound quality. However since the mod would make one side of the outpustage allmost straving for bias current instead of increasing it while the other site is unaffected, I think I can not expect a real improvement.
Sincerely,
Thijs Schrama
My headphones (32 ohm Grado's) draws about 5mA rms at (very) high volume levels, so 30mA bias is enough. But for some linearity or constant transductiance 30mA is barely enough, that's why I use a much hihger bias point. Even for a headphone amplifier I would advice (much) more than 30mA.
My point more specific was:
1] 30mA is little in to 8 Ohm (7.2mW peak about 5mW rms) that I think (!) it can never make any difference.
2] Accoridng to my simmulations of the topology, the 'bias resistor to V-' method decreases(!) bias current through the output stage. The opposite of the claimed effect!
3] From point 1] and 2] I conclude that nobody take the effort to make some educated remarks.
But I would like to add that I haven't tried the 'bias resistor to V-' mod and therefore I cann't comment on the claimed increase sound quality. However since the mod would make one side of the outpustage allmost straving for bias current instead of increasing it while the other site is unaffected, I think I can not expect a real improvement.
Sincerely,
Thijs Schrama
Please don't get so personal...
I was talking about connecting the RC filter to the input of the complete amplifier, not the opamp.
I agree that the "class A-tweak" is not completely sensible. It seems to work though. I do not know why. The difference is quite small, but there is one. I noticed more smoothness, but also a little bit more detail and tigher bass. I am not 100% sure, but it really seems an improvement.
Fedde
I was talking about connecting the RC filter to the input of the complete amplifier, not the opamp.
I agree that the "class A-tweak" is not completely sensible. It seems to work though. I do not know why. The difference is quite small, but there is one. I noticed more smoothness, but also a little bit more detail and tigher bass. I am not 100% sure, but it really seems an improvement.
Fedde
fedde[/i] [B]my amp really sounds different during different times of the day [/B][/QUOTE] [QUOTE][i]Originally posted by Nielsio said:
just the slightest bit of the dreaded thread jacking here but you just gotta read this ... well you don't gotta but you might wanna cause it's funny as H E double tooth picks and somewhat informative as well
The Magic Hour An examination of the mystery of the middle of the night bloom!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.