Tangent's ground channel w/o buffer

[morfic]

http://www.tangentsoft.net/elec/bitmaps/vgrounds/gnd-channel.png

What would the schematic look like with regard to the opamp's feedback loop if i left out the buffer if i don't have current needs to need the buffer?

Do i pretty much do this: http://www.tangentsoft.net/elec/bitmaps/vgrounds/vfb-opa.png

with this: http://www.tangentsoft.net/elec/bitmaps/vgrounds/tle2426cp.png

And do i keep R3 at 1K instead of the 100Ohm if followed by the buffer?
Omitting Cc?

Thanks,

Daniel

 

[morfic]

R3 being 1kOhm and R1 being 100Ohm, either on the opamp's output.
I'm tempted to guess I'd be using 1kOhm, Cc i don't see hurting anything either way though?
Only because it was the value he used in all other circuits where the opamp was used to split the rails w/o being followed by the buffer.

 

[jcx]

check if the tle2426 can provide the current and low impedance you need in your circuit - I believe it can handle a few 10s of mA wthout additional op amp or buffer

op amp "gnd buffer" can be intersting to analyse for stability with capacitive load


there are also other supply spitting options - usally just using a dual supply or the midpoint of an (even numbered) battery stack is simpler

 

[morfic]

The TLE2426 pdf talks about a MAX condition of +-80mA, i think Tangent mentions 20-40mA.
I think that that would be enough and play nice with the LME49720's max current (I'll be testing the circuit with a OPA2134 first.
And from all i read the LME49720 will love some .1uF bypass caps (metal film?).
I like the idea of tangent's active ground/ground channel, but since he mentions to use the same opamp and buffer in the ground channel as is used for actual amp circuit, i figured to take out the buffer since i am not using a buffer for left/right either.

Rather than just creating a nicely split power source i was interested in the active output ground of this.
Just would want to skip the buffer in the schematic, unless it makes sense to run a LME49720 for L/R and the ground channel with tle+opamp+buffer, even if there isn't a buffer after the LME49720.

If my company had paid me Thursday i would likely have at least the parts and listen to the basic cmoy with the dip8 tle rail splitter off Tangent's side, but that's pretty woulda/coulda/shoulda

So my question remains, can i strip his ground channel schematic of the buffer, or does it actually make sense to leave it even if the opamp for left and right channel does not have a buffer?

 

[nelsonvandal]

Is it a CMOY with active ground you're building, like a PINT or Mini3? When you say you don't have "current needs", what load are you thinking of?

LME49720 is peaking when run at unity gain. OPA2134 doesn't look too good either at unity gain. I'm not sure it will affect the sound.

I think it's a good idea to use different opamps in ground and L/R so you can blend the sound the way you want. Eg I find the combination of OPA134 and LME49720 to sound better than either of them used in both positions.

Sure you can use an opamp as active ground, but omit the 1k resistor in the feedback loop, or use a much lower value.

 

[jcx]

the TLE2426 data sheet is missing the critical "min output current" - the worst case (minimum) current it could sink/source so you have to derate the typical spec by "guesstimating" from the graphs - I wouldn't feel comfortable relying on more than 20mA from what I can see

the smallest short circuit spec is 31 mA - meaning the the output is only sinking that much while pulled all the way to the supply rail


echoing Nelson, you need to determine the active gnd load current requirement from the load impedance(s) and required output Vswing

 

[morfic]

Is it a CMOY with active ground you're building, like a PINT or Mini3? When you say you don't have "current needs", what load are you thinking of?

LME49720 is peaking when run at unity gain. OPA2134 doesn't look too good either at unity gain. I'm not sure it will affect the sound.

I think it's a good idea to use different opamps in ground and L/R so you can blend the sound the way you want. Eg I find the combination of OPA134 and LME49720 to sound better than either of them used in both positions.

Sure you can use an opamp as active ground, but omit the 1k resistor in the feedback loop, or use a much lower value.

Right, it's a cmoy, with Tangent's active ground, basically just making sure i am not way off course omitting the buffer.
A friend built his and everything draws 10mA according to his MM, interestingly it pulls the same idle. So i hope i get similar results once done. The phones driven are both 32ohm.
I Will then either use the 100ohm used in the active ground schematic between opamp and buffer or none.
It;s going to be a growing project, a buffer might end up in there later, this question relates to me always feeling the need to ask question if i am at least close, as i get going i hope to pick up enough knowledge along the way to only ask when really stuck
Your note about mixing opamps actually helps, this way i can look which of those i "know" (aka have listened to) are capable of more current for ground channel.

the TLE2426 data sheet is missing the critical "min output current" - the worst case (minimum) current it could sink/source so you have to derate the typical spec by "guesstimating" from the graphs - I wouldn't feel comfortable relying on more than 20mA from what I can see

the smallest short circuit spec is 31 mA - meaning the the output is only sinking that much while pulled all the way to the supply rail


echoing Nelson, you need to determine the active gnd load current requirement from the load impedance(s) and required output Vswing

From what i understand the TLE does not have to deal with the driver's current needs, it deals with the small currents to vground only.
Omitting the buffer, instead of say 250mA i would be limited to whatever the current the opamp is capable of handling, LME49710 may not be the one i end up with in ground channel. It's not capable of much more than the TLE for example.
LME49720 just happens to be the most enjoyable opamp to listen to.

EDIT: I'm almost tempted to use LM6172/LM6171, it sounded real nice in a portable amp i had, but i got very irritable with it after a while, not until later that i read it's a temperamental little gem, I guess the circuit was not made to work with that fast an opamp.
Since Tangent does show how to work with them on another page, i may have to try those, once i get a slower one working nice

Thanks to both of you,

Daniel

 

[nelsonvandal]

It's not always mentioned that the TLE2426 when used this way, sinking/sourcing the return currents, has it's own sound signature. I find it brighter than neutral and not as good as most of the modern opamps.

A cheap, fairly easy and good way (IMO the best way) to get the current you want is to use a discrete buffer in closed loop with the opamp. Since you use low impedance phones I think this is the way to go.

LM6171 has a warm sound signature and blends very well with the more "sterile" LM4562. I use this combination in a couple of amps with discrete buffers. The problem with LM6171 is the high input bias currents, but this is not an issue when used as ground channel. It's not as cranky as it's rumour. I would say the opposite, and it's cheap. Since it's got enough current for low impedance phones, you could get away without a buffer.

 

[morfic]

I Wasn't sure if the ground channel actually contributes to the overall sound or not, from what you are saying i gather it does, and thus i will do some mixing and matching.
I will certainly try the LM6171, it's added to my order already, added a LT1363 as well, since i also have a LT1364 i got with my hybrid amp, i have plenty of mix and match choices now.
Radioshack threw me a curveball of course, they no longer stock the pcb i wanted to use locally.
Since i will have to go mail order on the pcb, i am debating between the above and
this one.

Not sure if i need more "connections per pin" and could live with the single rail down the center. But since i need to go mail order may as well look around instead of jumping on the radioshack one (even if it IS the one on tangent's website)

Either way, i have enough extras in that order to build at least a second amp (if i forgo sockets on it), with the purpose of the extra parts being that i want to try different size input caps/power caps, different type bypass caps, various gains and once i have another power supply, different voltages.
Minimum rating is 35V with the psu i pick up later maxing out at 30V, so i guess it's all fun and games once i place the order

 

[morfic]

Nelson, do you have some discrete buffers to suggest aside from sijosae's, impossible to google and not hit threads that in the end are based around his.
Got boards (went with the audioxpress ones) and parts finally ordered, no buffers yet. I hope i can pick up resistors locally and should have enough resistors and caps (if needed) left over from the amp part order.

Thanks,

Daniel

 

[nelsonvandal]

Nelson, do you have some discrete buffers to suggest aside from sijosae's, impossible to google and not hit threads that in the end are based around his.

Sijosae was very productive. He has built amps with both diamond buffers and simple diode biased emitter followers. It's too bad it's the latter that most people associate with him. I prefer the diamond buffer. The DB has been around for a long time and is often used in monolithic opamps.

Some examples where diamond buffers are used in headphone amps:
http://tangentsoft.net/audio/ppa/amp2/
http://www.diamondstar.de/dDB_design.html
http://www.diyforums.org/
http://www.scribd.com/doc/6772415/Lisa-III-Schematics-and-Specifications

Other buffer options are single BJT emitter follower, double emitter follower, complementary feedback pair
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
http://ipodstudio.com/cms/media/lisa_ii_schematic_complete_1.pdf

MOSFET source follower
http://www.amb.org/audio/mmm/

The options are numerous. Any power amp output stage could be scaled down to make a buffer for an opamp in a headphone amp.

 

[morfic]

Thanks Nelson, looking at finished amp's schematics....triggered a big "DUH" moment :/

Really appreciate the eye opener, had been googling around on a couple topics, turns out some Nelsonvandal on head-fi.org gave some pretty good info and sounded the anti audiophile snake-oil horn, always good when you accidentally find out you seem to listen to the right guy
Building the cmoy is part fun, part wanting to learn, what sounds how and what sounds the same or notably different.

Good thing on most the discrete buffers, i stand a chance to find needed parts locally....i hope.

Hopefully i get the amp parts before the weekend, so i can stop "talking" and start doing

Daniel

 

[nelsonvandal]

I'm no wiz. I only try to be helpful. Don't take my words for granted, but I've listened to a lot of active ground circuits.

 

[morfic]

I'm no wiz. I only try to be helpful. Don't take my words for granted, but I've listened to a lot of active ground circuits.

No worries, in such a wide field, with such largely differing opinions, it just helps to start somewhere. Who knows down the road i may disagree with every single thing you said But w/o knowing anything myself i have no reason to do so and your help is much appreciated.

Digikey "outshipped" AudioXpress, so i get to look at parts w/o a board right now.

 

[morfic]

One question i meant to ask, tangent suggests to use the same opamp and buffer combination for ground channel as for L/R, and it seems i can mix and match opamps and buffers.
But are there useful configurations where there is only a buffer in the ground channel, but not on L/R. Or where there is a buffer on L/R but not the ground channel?

Or is there indeed no use in a configuration that is this asymetrical?

 

[jcx]

an active ground channel has to handle both L&R channel load currents, the worst case is 2X the current in either signal channel so the gnd buffer should have higher current sink/source capability than the signal channel

 

[morfic]

So in essence it could make sense to use a buffer only on the ground channel if the L/R channels can handle say 50mA each w/o a buffer to add a buffer to the ground channel alone, just to make sure it can handle the 100mA total.

Or if the L/R dual channel opamp can source 50mA per channel to combine it with an ground channel opamp that can source 100mA and not use a buffer on any of the 3 channels at all, until a deficiency in the amp is found that a buffer addresses.

Would either comment be correct to say?

If the ground channel is capable of less, will it simply become the limiting factor or are there negative effects beyond the imposed limit?

 

[morfic]

Are BUF634 really stackable, i've seen them used piggyback soldered to each other.

If i use one BUF634 per L and R, then i'd need 2 BUF634 on ground channel (or an equivalent capable of the same current, and i'm not aware of an IC of doubling it, LME49600 has same current with a higher short circuit current capability)

I'd like to try BUF634 (to help on the board i think i'd try TO-220) if i could just "double up" for the ground channel.

Discrete buffers would be a next step i think. (mainly a space issue on these boards (D4 by DH Labs))

Thanks,

Daniel