Hi all
Has anybody here listened to various DIY discrete opamps?
I have made a clone JE990 and it is very good but wondered if there is anything A LOT better out there - soundwise?
I looked at the various Sam Groner opamps as well as the line-up modified 918 as well as the plenty on the net.
I suppose I am asking what shouldn't I bother with as a DIY experiment
Rob
Has anybody here listened to various DIY discrete opamps?
I have made a clone JE990 and it is very good but wondered if there is anything A LOT better out there - soundwise?
I looked at the various Sam Groner opamps as well as the line-up modified 918 as well as the plenty on the net.
I suppose I am asking what shouldn't I bother with as a DIY experiment
Rob
Here is just one thread from diyaudio on the topic of discrete op-amps. There are others such threads here as well, just search this site. Listening experiences will vary with application, system context, and circuit implementation details.
http://www.diyaudio.com/forums/digital-source/185945-discrete-opamp-class.html
http://www.diyaudio.com/forums/digital-source/185945-discrete-opamp-class.html
Last edited:
From LOW up to MID frequencies, discrete implementations are unsurpassed. From MID up to HIGH frequencies, enough ICs are competitive to discrete.
The only objective test - as i have mentioned a lot of times - is the use of instruments: an oscilloscope at the output, and a square wave as stimulus.
The only objective test - as i have mentioned a lot of times - is the use of instruments: an oscilloscope at the output, and a square wave as stimulus.
i simply do not see a real advantage of discrete opamps vs chips.
possibly can improve figures, but i allways had doubts about how usefull is it anyways.
I could be very wrong, but for example 0.001% thd or 0.00001 % thd makes no difference for me.
Then i check on valve amps and find rather high thd, reason to love a valve amp is its unique distorsion.
sounds a bit.. like split personality.
sure better to have and not need than need but not have, i agree with that, but is it realy worth it?
i dunno..
as a DIY project, creating Your own discrete opamp on the otherhand sounds more than nice
possibly can improve figures, but i allways had doubts about how usefull is it anyways.
I could be very wrong, but for example 0.001% thd or 0.00001 % thd makes no difference for me.
Then i check on valve amps and find rather high thd, reason to love a valve amp is its unique distorsion.
sounds a bit.. like split personality.
sure better to have and not need than need but not have, i agree with that, but is it realy worth it?
i dunno..
as a DIY project, creating Your own discrete opamp on the otherhand sounds more than nice
I see two big differences between discrete opamps designed for audio duty and designed to sound good, vs IC opamps designed to be universal.
First is the number of components used in the respective topologies.
Second is the minuscule output stage bias.
I think the factors of 100 or so between the two make for a well designed discrete generally sounding better than IC.
First is the number of components used in the respective topologies.
Second is the minuscule output stage bias.
I think the factors of 100 or so between the two make for a well designed discrete generally sounding better than IC.
It's unlikely.
There might be something that is minutely better, but a LOT better, no. There won't be anything that's a LOT better.
I say this without the benefit of listening. These days, for an amplifier to be a LOT better than another amplifier, one has to be a LOT worse than an amplifier can be.
You have to remember that amplifiers can only GET so good. There isn't room for improvement ad infinitum. Its an asymptote to perfection, not a straight line rising into the future for all time.
w
Very nice phrasing Wakibaki. Mathematics is the corner-stone of Physics.
ad infinitum = without interruption.
asymptote = not going to meet; never.
agreed! for this reason, a "good" op amp and a "hot" i.e. high bias current like AD811 for example) buffer (low or unity gain) is hard to beat.
mlloyd1
mlloyd1
PHP:
I say this without the benefit of listening.I really have to take anything you say with a pinch of salt at this stage as I asked for 'has anyone had any listening experience'. Most folks here have their own ideas about the 'whys and nots' of IC vs discrete but this is about - have you.
@ robmil
Yes, you are right. I think that i was not so clear by what i said. And yes, i have big listening experience with ICs working for many years with any professional audio mixing and sound processing device.
Where i say "discrete are unsurpassed from LOW up to MID frequencies" i mean:
Sound of discrete in Mids and Highs is "sweet" without colorations and screaming compared to ICs.
In Lows is extremely tight and "warm". You feel like you stay on a bandstand, between the bass amplifier and the drum kit.
Yes, you are right. I think that i was not so clear by what i said. And yes, i have big listening experience with ICs working for many years with any professional audio mixing and sound processing device.
Where i say "discrete are unsurpassed from LOW up to MID frequencies" i mean:
Sound of discrete in Mids and Highs is "sweet" without colorations and screaming compared to ICs.
In Lows is extremely tight and "warm". You feel like you stay on a bandstand, between the bass amplifier and the drum kit.
i simply do not see a real advantage of discrete opamps vs chips.
possibly can improve figures, but i allways had doubts about how usefull is it anyways.
I could be very wrong, but for example 0.001% thd or 0.00001 % thd makes no difference for me.
Then i check on valve amps and find rather high thd, reason to love a valve amp is its unique distorsion.
sounds a bit.. like split personality.
sure better to have and not need than need but not have, i agree with that, but is it realy worth it?
i dunno..
as a DIY project, creating Your own discrete opamp on the otherhand sounds more than nice
I find one particularly significant advantage that discrete op-amps offer over monolithic op-amps. Discrete implementations enable one to control the open-loop gain and distortion, and thereby, control the amount of negative feedback which gets ultimately employed. I, like many others, find low feedback circuits to in general sound superior to high feedback circuits. As you suggest, this is not really a question of objective distortion performance, it's a question of subjective listening performance. On strictly objective terms, it's none too easy to best the closed-loop performance of today's monolithic devices.
Last edited:
I also use a JE-990 type circuit and find it sounds excellent. I have not found any IC or combination of ICs (e.g Jung's nested f/b scheme) that sounds quite as good.
One thing I did find with the JE-990 topology is that I'm not too happy wih the sound of the LM394. The SSM equivalent or a pair or 2SK170s sounds better, even a pair of 2Nxxxx sounds fine. Of course the JFETs will affect the loop gain somewhat.
My preferred option is: JFETs on the input (no input cap), no dc servo and a decent output cap.
I would like to try the Kaneda topology as some point to see if this is better than the JE-990. Anyone compared the 2?
One thing I did find with the JE-990 topology is that I'm not too happy wih the sound of the LM394. The SSM equivalent or a pair or 2SK170s sounds better, even a pair of 2Nxxxx sounds fine. Of course the JFETs will affect the loop gain somewhat.
My preferred option is: JFETs on the input (no input cap), no dc servo and a decent output cap.
I would like to try the Kaneda topology as some point to see if this is better than the JE-990. Anyone compared the 2?
It is extremely easy to run an op-amp output stage in class A - just load the output to one of the rails with a resistor or a current source. Keep in mind that when driving a typical 10k output load to 10VRMS, most 'audio grade' op-amps will still be in class A, and at the more normal 1-2VRMS, they will be solidly in class A even without any output pre-load to force them into class A.
I have built many circuits, both discrete and with op-amps. While discrete designs have their place, I think op-amps offer some outstanding benefits. If one can see past the ant-feedback b.s. that seems regretably to permeate our hobby, I think on a double blind shoot out (yes, DB is the only scientific way to ***** equipement subjectively), I'd put my money on a good op-amp.
If you are looking for something that changes the sound of the source material, then of course that is a differentr matter - here indeed you may end up preffering a discrete or tube design.
Some specialst applications will demand a discrete approach - e.g. MC head amp or a dynamic mic pre-amp (but THAT corp have that covered with an IC as well now), and it could be argued (though its really open to debate) they may offer some performance benefits on MM inputs.
For transparency, accuracy, speed, linearity, its op-amps that win every time.
I have built many circuits, both discrete and with op-amps. While discrete designs have their place, I think op-amps offer some outstanding benefits. If one can see past the ant-feedback b.s. that seems regretably to permeate our hobby, I think on a double blind shoot out (yes, DB is the only scientific way to ***** equipement subjectively), I'd put my money on a good op-amp.
If you are looking for something that changes the sound of the source material, then of course that is a differentr matter - here indeed you may end up preffering a discrete or tube design.
Some specialst applications will demand a discrete approach - e.g. MC head amp or a dynamic mic pre-amp (but THAT corp have that covered with an IC as well now), and it could be argued (though its really open to debate) they may offer some performance benefits on MM inputs.
For transparency, accuracy, speed, linearity, its op-amps that win every time.
IC op-amps. Heard many. Tried most everything you can do to improve the sound. O/P stage biasing works well even on reasonably biased IC op-amps. Use to use about 10mA and sometimes up to 20mA.
There is something else you can do that I never see mentioned on this forum, or anywhere else for that matter, that makes a significant inprovement. Maybe more than the O/P biasing. It has to do with common mode distortion - and I am not going to say anything more than that.
Designed a couple of discrete op-amps. Stopped using op-amps after that for serious audio reproduction. It is pretty easy to beat the best op-amps, even with all their tweaks, with a discrete op-amp. But hey why stay with closed loop when you have the option of open loop with a discrete design. Can't do that with an op-amp.
Then there is the question of thermionic vacuum tubes. Those things have no equal in the discrete world let alone IC op-amp world when applied in the right manner and in the right places. But the problem is dwindling supply of those parts unfortunately.
IC op-amps have a nasty gritty distortion that once you have heard and identified you don't want to go back to. The buggers can even be heard in the audio path when used for DC servos. I found a way around that too but that information is for sale only.
One of the other guys I did research with would probably give it away for free if you asked the right way. But hey some people prefer their egos boosted for their hard work and research rather than their back pockets.
Anyway, that's my findings.
There is something else you can do that I never see mentioned on this forum, or anywhere else for that matter, that makes a significant inprovement. Maybe more than the O/P biasing. It has to do with common mode distortion - and I am not going to say anything more than that.
Designed a couple of discrete op-amps. Stopped using op-amps after that for serious audio reproduction. It is pretty easy to beat the best op-amps, even with all their tweaks, with a discrete op-amp. But hey why stay with closed loop when you have the option of open loop with a discrete design. Can't do that with an op-amp.
Then there is the question of thermionic vacuum tubes. Those things have no equal in the discrete world let alone IC op-amp world when applied in the right manner and in the right places. But the problem is dwindling supply of those parts unfortunately.
IC op-amps have a nasty gritty distortion that once you have heard and identified you don't want to go back to. The buggers can even be heard in the audio path when used for DC servos. I found a way around that too but that information is for sale only.
One of the other guys I did research with would probably give it away for free if you asked the right way. But hey some people prefer their egos boosted for their hard work and research rather than their back pockets.
Anyway, that's my findings.
Useful observations, thanks. I've been wondering for a while if sound quality is mainly to do with the input stage's immunity to RF. As I recall the JE-990 has a unique feature - inductors to degenerate the input pair. Have you any experience with trying different inductors (values, styles) in the Re circuit? Does replacing just those two inductors with resistors make a difference to the sound?
"IC op-amps have a nasty gritty distortion that once you have heard and identified you don't want to go back to. The buggers can even be heard in the audio path when used for DC servos. I found a way around that too but that information is for sale only."
LOL - do you really want us to believe that?
LOL - do you really want us to believe that?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.