Hi,
I've recently completed a relay based R-2R attenuator and am looking for a buffer for it to be independent of interconnects and the following stage input impedance. My source is Subbu DAC and the power amp is MyRef Fremen Edition at the moment (~80k input impedance). I also have a parallel LM4780 chipamp from BrianGT (~20k input impedance). My R-2R attenuator impedance is 10k.
Just to note, I'm not an electronics expert (although have a related background) and designing my own schematics is in most cases out of question (except for relatively simple ones) but own PCBs I design happily for the existing schematics because I can understand them (up to certain level of complexity).
Now to the problem! There is a full zoo of discrete buffers around: B1, Bonsai, Calvin, RJM, Kuartlotron and the list will go on… On the other hand there are IC based buffers. I've recently visited many threads on this forum dedicated to discrete buffers but I did not see many ICs based solution while searching.
Personally, I am not really into trying everything possible and tweaking/testing forever... So being practical I would see at least one major drawback of the discrete variants – nearly extinct transistors for them plus the need to closely match the transistors or pay someone to do it for you.
On the other hand, looking at specs of ICs I've selected as a possible candidate the LME49990 alone or in combination with the LME49600 (here “The Wire” project comes to mind for which I have PCBs). The combo offers a possibility to use it as the headphone amp or the line level buffer at the same time. Both ICs have vanishing distortion figures and are stable when driving capacitive loads within reasonable range. So I suppose I won't hear the buffer itself in this case but rather my source and/or the amp/speakers.
The discrete counterpart which I consider at the moment is the B1 since I have a matched transistor set for it and an already designed small PCB. Of course the driving capability of the B1 does not come close to LME49600 but is near to the LME49990. If I would go with ICs combo I would see it powered by the TPS7A4700 based PSU designed right on the same board saving a lot of space in the chassis (Owen's boards are small but for the buffer only, the PSU can be even smaller I guess). Also it seems that the sound of B1 is largely influenced by the PSU, more than I would assume for an ICs based buffer. Salas DCB1 discussions being a good example for this (changing the PSU parameters change the sound and thus you actually hear the buffer). It also shows the size of a buffer with the dedicated PSU. Even Jung/Didden superreg for B1 would take a lot of space especially thinking of the dual mono configuration.
Now is the question which puzzles me! What is so “magical” about the discrete solutions compared to ICs which justifies them?
To narrow the discussion please use the above mentioned configurations for comparison if possible. Better alternatives of both types of buffers are also welcomed but it has a danger to start an endless discussion which is not my intention.
Please do not recommend to build both and decide by listening. I would not trust my ears and I do not have an instrument to measure the relevant parameters. I just want to make an informed decision on a better buffer for my particular situation based on arguments I can understand and sleep well after I make that decision
Any opinions are welcome!
Regards,
Oleg
I've recently completed a relay based R-2R attenuator and am looking for a buffer for it to be independent of interconnects and the following stage input impedance. My source is Subbu DAC and the power amp is MyRef Fremen Edition at the moment (~80k input impedance). I also have a parallel LM4780 chipamp from BrianGT (~20k input impedance). My R-2R attenuator impedance is 10k.
Just to note, I'm not an electronics expert (although have a related background) and designing my own schematics is in most cases out of question (except for relatively simple ones) but own PCBs I design happily for the existing schematics because I can understand them (up to certain level of complexity).
Now to the problem! There is a full zoo of discrete buffers around: B1, Bonsai, Calvin, RJM, Kuartlotron and the list will go on… On the other hand there are IC based buffers. I've recently visited many threads on this forum dedicated to discrete buffers but I did not see many ICs based solution while searching.
Personally, I am not really into trying everything possible and tweaking/testing forever... So being practical I would see at least one major drawback of the discrete variants – nearly extinct transistors for them plus the need to closely match the transistors or pay someone to do it for you.
On the other hand, looking at specs of ICs I've selected as a possible candidate the LME49990 alone or in combination with the LME49600 (here “The Wire” project comes to mind for which I have PCBs). The combo offers a possibility to use it as the headphone amp or the line level buffer at the same time. Both ICs have vanishing distortion figures and are stable when driving capacitive loads within reasonable range. So I suppose I won't hear the buffer itself in this case but rather my source and/or the amp/speakers.
The discrete counterpart which I consider at the moment is the B1 since I have a matched transistor set for it and an already designed small PCB. Of course the driving capability of the B1 does not come close to LME49600 but is near to the LME49990. If I would go with ICs combo I would see it powered by the TPS7A4700 based PSU designed right on the same board saving a lot of space in the chassis (Owen's boards are small but for the buffer only, the PSU can be even smaller I guess). Also it seems that the sound of B1 is largely influenced by the PSU, more than I would assume for an ICs based buffer. Salas DCB1 discussions being a good example for this (changing the PSU parameters change the sound and thus you actually hear the buffer). It also shows the size of a buffer with the dedicated PSU. Even Jung/Didden superreg for B1 would take a lot of space especially thinking of the dual mono configuration.
Now is the question which puzzles me! What is so “magical” about the discrete solutions compared to ICs which justifies them?
To narrow the discussion please use the above mentioned configurations for comparison if possible. Better alternatives of both types of buffers are also welcomed but it has a danger to start an endless discussion which is not my intention.
Please do not recommend to build both and decide by listening. I would not trust my ears and I do not have an instrument to measure the relevant parameters. I just want to make an informed decision on a better buffer for my particular situation based on arguments I can understand and sleep well after I make that decision
Any opinions are welcome!
Regards,
Oleg
Hi Oleg,
Want the truth? Either one can sound excellent. The IC buffer does need an op amp to drive it as they often need a low impedance source, but they are hard to beat for distortion and noise under load.
Having said that, I have had my best luck with a diamond buffer based discrete design. No surprise since most IC buffers are a diamond buffer inside. You really have but one circuit to choose, and the packages they come in. One thing a discrete design brings to the table is higher supply voltages. I like mine running between bipolar 25 to 35 volt supplies.
Don't skimp on the power supplies either as that can undo everything you worked so hard for with the buffer. I tend to use a little gain, but you can design for unity if you want.
-Chris
Want the truth? Either one can sound excellent. The IC buffer does need an op amp to drive it as they often need a low impedance source, but they are hard to beat for distortion and noise under load.
Having said that, I have had my best luck with a diamond buffer based discrete design. No surprise since most IC buffers are a diamond buffer inside. You really have but one circuit to choose, and the packages they come in. One thing a discrete design brings to the table is higher supply voltages. I like mine running between bipolar 25 to 35 volt supplies.
Don't skimp on the power supplies either as that can undo everything you worked so hard for with the buffer. I tend to use a little gain, but you can design for unity if you want.
-Chris
bootstrapped supply composite op amp + buffer - reduces even the input stage nonlinear common mode impedance
http://www.epanorama.net/sff/Misc/O...g to reduce distortion in op-amp circuits.pdf
Supply Bootstrapping Reduces Distortion In Op-Amp Circuits | New operational amplifiers optimized for high-performance audio and ultrasound applications combine extremely low total harmonic distortion plus noise (THD+N), -130 dB, with large output vo
I recommend looking beyond "buffer" IC with fixed internal unity gain - more options are available in CFA op amps designed for DSL driver applications - I've used the TPA6120 which is a audio speced version of the THS6012 for composite amp output stages
CFA op amps can be used unity gain with a feedback R - you can even size the R value to "overcompensate", increase stability margins
http://www.epanorama.net/sff/Misc/O...g to reduce distortion in op-amp circuits.pdf
Supply Bootstrapping Reduces Distortion In Op-Amp Circuits | New operational amplifiers optimized for high-performance audio and ultrasound applications combine extremely low total harmonic distortion plus noise (THD+N), -130 dB, with large output vo
I recommend looking beyond "buffer" IC with fixed internal unity gain - more options are available in CFA op amps designed for DSL driver applications - I've used the TPA6120 which is a audio speced version of the THS6012 for composite amp output stages
CFA op amps can be used unity gain with a feedback R - you can even size the R value to "overcompensate", increase stability margins
Last edited:
Thanks jcx!
The schematic you posted is a bit complicated for me to read right away, but I've got a basic idea.
So far I've been thinking of using the schematic from LME49600 datasheet (see attached). And not too fancy PSU of proven design.
Hi Keruskerfuerst,
I know there is a number of dedicated buffer ICs. How is the LME49600 part compared to your list?
The schematic you posted is a bit complicated for me to read right away, but I've got a basic idea.
So far I've been thinking of using the schematic from LME49600 datasheet (see attached). And not too fancy PSU of proven design.
Hi Keruskerfuerst,
I know there is a number of dedicated buffer ICs. How is the LME49600 part compared to your list?
Attachments
its hard to tell since many buffer IC datasheets that do give audio distortion numbers actually give it for the composite - not the just the buffer alone
certainly if you don't have the experience and the equipment then stick to the datasheet circuit or established project amp schematics
for few kOhm or less attenuator Zout the input common mode nonlinearity distortion shouldn't be above the noise in many audio op amps
certainly if you don't have the experience and the equipment then stick to the datasheet circuit or established project amp schematics
for few kOhm or less attenuator Zout the input common mode nonlinearity distortion shouldn't be above the noise in many audio op amps
Last edited:
I will definitely look in the given directions. When I decide on what I actually want to try and start designing a PCB, I'll come back with questions. As of now I'm still searching around. I also seems to more like the buffer IC approach rather than discrete. But who knows what I decide next day... The problem of making a choice kills me
Oleg
Oleg
Hi Oleg,
You can't go wrong with either approach. They both require the same skills from you, but the discrete approach requires you match some transistors. That demands a higher skill level. Designing a good PCB is challenge enough I would say.
Maybe after you build the buffer IC version you could go back and try to build one from scratch. Done properly, I have no preference for either.
-Chris
You can't go wrong with either approach. They both require the same skills from you, but the discrete approach requires you match some transistors. That demands a higher skill level. Designing a good PCB is challenge enough I would say.
Maybe after you build the buffer IC version you could go back and try to build one from scratch. Done properly, I have no preference for either.
-Chris
One slightly off topic question... but related. Is there an argument against placing the voltage regulators as close as possible to the buffer IC? In my naive view it should eliminate some problems and allow to remove small local electrolytic caps (those typical 47~100uF) next to the buffer IC supply pins? I assume that the main filter capacitors for the regulator itself are no further than an inch from the buffer IC? The interest here is to minimize the buffer board size.
OK. Now I see I have to start a new thread
OK. Now I see I have to start a new thread
Hi Oleg,
Build a normal power supply. Maybe build pre-regulators in that area. Create smaller regulators for each channel near the circuits along with RF bypassing. That should give you the best performance.
You don't want to mount filter capacitors away from the rectifiers. There are high surge currents that flow between the rectifiers and capacitors that also will include the transformer winding's. The pre-regulator removes the ripple and makes the local regulators run at lower temperatures.
-Chris
Build a normal power supply. Maybe build pre-regulators in that area. Create smaller regulators for each channel near the circuits along with RF bypassing. That should give you the best performance.
You don't want to mount filter capacitors away from the rectifiers. There are high surge currents that flow between the rectifiers and capacitors that also will include the transformer winding's. The pre-regulator removes the ripple and makes the local regulators run at lower temperatures.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.