Hi Russ
Alternatively, people could try my suggestions for themselves!
I understand your arguments, and it may well be that I'm deluding myself. I reckon that the vast majority of subjective asseessments of this sort fall in this category. But I've been doing this a long time, and, judging by the number of subjective failures I experience, I suffer from expectation bias less than most. The problem is- listening is a subjective experience, and that's what this stuff is for!
You know enormously more about electronic engineering than me that's for sure, (I'm a research scientist, but with a background in organic chemistry) which give you solid expectations. This is a sound basis for circuit design- we don't just want to throw components together randomly in the hope we eventually get a working circuit! But surely its good to try and tweek the boundaries a bit, and question assumptions. for instance, is low noise the only important determinant of performance for the npn BJT? And could some unsuspected time varying performance difference between the two halves of the balanced circuit upset the cancellation you describe? Sounds far fetched stately starkly like that, but not everyone seems to share your faith in the inevitable restorative effects of balanced circuitry (they may of course be talking rubbish-I have very little experience myself, and have no opinion either way).
A long winded way of saying that trying the mods, by making measurements and by listening, then responding, would be truly constructive criticism. Now, there's no way I could expect you to do this- for good theoretical reasons, you feel it's very likely going to a waste of your time. But your reply to my posting- and others like it is, IMHO a bit too dogmatic and discouraging. Advance the arguments making the effect unlikely, for sure- but then going on to dismiss my assessments utterly on theoretical grounds is surely unnecessary? Even though, IYO, my experiments are prompted by factual misconceptions in the first place? I might still be on to something. After all, all I'm trying to do is increase our collective listening pleasure
Hope I haven't caused offence, but I felt the need respond
Keep up the excellent work
Paul
Alternatively, people could try my suggestions for themselves!
I understand your arguments, and it may well be that I'm deluding myself. I reckon that the vast majority of subjective asseessments of this sort fall in this category. But I've been doing this a long time, and, judging by the number of subjective failures I experience, I suffer from expectation bias less than most. The problem is- listening is a subjective experience, and that's what this stuff is for!
You know enormously more about electronic engineering than me that's for sure, (I'm a research scientist, but with a background in organic chemistry) which give you solid expectations. This is a sound basis for circuit design- we don't just want to throw components together randomly in the hope we eventually get a working circuit! But surely its good to try and tweek the boundaries a bit, and question assumptions. for instance, is low noise the only important determinant of performance for the npn BJT? And could some unsuspected time varying performance difference between the two halves of the balanced circuit upset the cancellation you describe? Sounds far fetched stately starkly like that, but not everyone seems to share your faith in the inevitable restorative effects of balanced circuitry (they may of course be talking rubbish-I have very little experience myself, and have no opinion either way).
A long winded way of saying that trying the mods, by making measurements and by listening, then responding, would be truly constructive criticism. Now, there's no way I could expect you to do this- for good theoretical reasons, you feel it's very likely going to a waste of your time. But your reply to my posting- and others like it is, IMHO a bit too dogmatic and discouraging. Advance the arguments making the effect unlikely, for sure- but then going on to dismiss my assessments utterly on theoretical grounds is surely unnecessary? Even though, IYO, my experiments are prompted by factual misconceptions in the first place? I might still be on to something. After all, all I'm trying to do is increase our collective listening pleasure
Hope I haven't caused offence, but I felt the need respond
Keep up the excellent work
Paul
Hi Paul,
No offense was ever taken. I hope you don't find my advice offensive either.
I do have a lot of experience, especially with this circuit. I am doing my best to pass some of it along. It is OK to push the boundaries as you say. Your modding does not offend me in the least. I am always glad to see people thinking about such things.
Any mod you present should be vetted before it can be endorsed. You are posting on Twisted Pear's forum after all.
In this case I am really just trying to tell you in as helpful way I can that your CCS modification adds nothing to the circuit but noise from the additional active components.
I am not saying this to be rude or discouraging. I am only saying it to give you some constructive feedback on your idea.
As for the virtues of a balanced DAC and I/V... Well when the output of the DAC is summed current from the DAC is always constant. This is why the balanced I/V is preferred. It turns a process into as close to a zero sum game as practical.
I also have made it clear I have no wish to discourage you or anyone else for that matter. My only aim is to give you food for thought as you seek to better yourself and your gear.
Those points made. I also have obligation on the Twisted Pear forum, for our users, to explain why in this case more parts will really not get you *better* performance.
Do you enjoy it anyway? Yes? great!. Would I suggest the mod? No. Might you come up with something wicked cool at some point? Absolutely, and I would be the first to say so.
You should take my advice for what it is, an invitation to learn more. You shared your idea. I would hope you wanted some feedback on it. Now if for some reason you didn't want feedback from the designer then you really should not be posting ideas here.
Cheers!
Russ
Last edited:
Well...
I am sure that no resistors "need" to be replaced! I used 0.1% TX-2575s for the 1K guys at R5-8, as these are likely the most critical resistors in the circuit. What is nice about such a simple circuit is that there are so few parts in the signal path, that using a few expensive parts is easy to "justify". I also used TX-2575s for the two current sources, 4K75 (no buffer or BAL/SE in my Legato).
I used PRPs for the rest of the resistors, with R1-4 at 180R and R20,21,23,24 at 390R. These gave me 1.9V output at 0dB. I really like the PRPs for affordable quality resistors.
My B-II with this Legato really sounds great!
I am sure that no resistors "need" to be replaced! I used 0.1% TX-2575s for the 1K guys at R5-8, as these are likely the most critical resistors in the circuit. What is nice about such a simple circuit is that there are so few parts in the signal path, that using a few expensive parts is easy to "justify". I also used TX-2575s for the two current sources, 4K75 (no buffer or BAL/SE in my Legato).
I used PRPs for the rest of the resistors, with R1-4 at 180R and R20,21,23,24 at 390R. These gave me 1.9V output at 0dB. I really like the PRPs for affordable quality resistors.
My B-II with this Legato really sounds great!
Hi
I think the critical resistors are actually the 1W types- R1-4, R20/21, R23/24- as it is these that perform the I/V conversion. Might be worth trying Takman 1W metal film, or even 1W Shinkohs if you can get them. But both these are physically larger than the existing types, and the latter are very good
I've tried putting two extra BS250's (actually TP0610KL- see below) in parallel with each existing FET- wow! You guys are going to love Legato 2. Definitely a case of practice following theory. An excellent idea- thanks Russ
If anyone tries this, note "BS250" describes a Philips type, which isn't an exact equivalent of the BS250KL (from Vishay Siliconix) that Russ uses. TP0610KL is identical to the latter- it shares the same datasheet- except the pinning is reversed- just turn each device through 180 degrees.
Now, possibly more controversially.......
Despite Russ's comments, I thought the CCS I added was sonically beneficial. Now, as Russ pointed out, in an ideal balanced circuit it can't do anything but add noise. So perhaps things aren't perfect? Well, in practice they never are, but Russ' combination of sound design, simulation and measurement strongly suggest they're close. But still?
I use a SE amp and so use the on board Bal to SE converter. A text book op amp-based stage, well executed and very common. But not an ideal load nonetheless. (Andy Grove discusses this in an article on the Audio Note website). Now many- most?- probably feel it's good enough, but, again- what if?
So I tried isolating the CFP stage by using a buffer on its outputs. Not the one Russ provided, but a Pass B1 buffer using a pair of 2SK170BL FETs, one used as a current source, powered from the +15V and 0V rails. I directly connected these directly to the outputs, as the dc voltage here sits midway between the buffer's supply. I used no gate stopper, nor series R on the output, and retained the bipolar coupling caps. It's extremely easy to implement, and easy to reverse if you don't like it
I used the FETs because
I'm using power supplies which are running close to their max current output- not enough to impement Russ' buffers
I know the B1 buffer well and rate it highly
easy to do
Russ felt the buffer he provided wasn't beneficial anyway
Definitely a keeper to my ears! Now whether anyone else agrees, who knows. Whether the improvement only applies to SE users, ditto.
I then tried a B1 powered from both rails, cap coupled on its input, input referenced to ground with 100k, dc coupled to the op amp on its output. Very difficult to implement, and didn't like at all, is the short story
Well, there you go. Better don my asbestos suit just in case. I advance my reasoning to try and demonstrate that I'm not just trying stuff randomly. But, all I care about in the end is the sound!
Best wishes
Paul N
I think the critical resistors are actually the 1W types- R1-4, R20/21, R23/24- as it is these that perform the I/V conversion. Might be worth trying Takman 1W metal film, or even 1W Shinkohs if you can get them. But both these are physically larger than the existing types, and the latter are very good
I've tried putting two extra BS250's (actually TP0610KL- see below) in parallel with each existing FET- wow! You guys are going to love Legato 2.
Definitely a case of practice following theory. An excellent idea- thanks RussIf anyone tries this, note "BS250" describes a Philips type, which isn't an exact equivalent of the BS250KL (from Vishay Siliconix) that Russ uses. TP0610KL is identical to the latter- it shares the same datasheet- except the pinning is reversed- just turn each device through 180 degrees.
Now, possibly more controversially
.......Despite Russ's comments, I thought the CCS I added was sonically beneficial. Now, as Russ pointed out, in an ideal balanced circuit it can't do anything but add noise. So perhaps things aren't perfect? Well, in practice they never are, but Russ' combination of sound design, simulation and measurement strongly suggest they're close. But still?
I use a SE amp and so use the on board Bal to SE converter. A text book op amp-based stage, well executed and very common. But not an ideal load nonetheless. (Andy Grove discusses this in an article on the Audio Note website). Now many- most?- probably feel it's good enough, but, again- what if?
So I tried isolating the CFP stage by using a buffer on its outputs. Not the one Russ provided, but a Pass B1 buffer using a pair of 2SK170BL FETs, one used as a current source, powered from the +15V and 0V rails. I directly connected these directly to the outputs, as the dc voltage here sits midway between the buffer's supply. I used no gate stopper, nor series R on the output, and retained the bipolar coupling caps. It's extremely easy to implement, and easy to reverse if you don't like it
I used the FETs because
I'm using power supplies which are running close to their max current output- not enough to impement Russ' buffers
I know the B1 buffer well and rate it highly
easy to do
Russ felt the buffer he provided wasn't beneficial anyway
Definitely a keeper to my ears! Now whether anyone else agrees, who knows. Whether the improvement only applies to SE users, ditto.
I then tried a B1 powered from both rails, cap coupled on its input, input referenced to ground with 100k, dc coupled to the op amp on its output. Very difficult to implement, and didn't like at all, is the short story
Well, there you go. Better don my asbestos suit just in case
. I advance my reasoning to try and demonstrate that I'm not just trying stuff randomly. But, all I care about in the end is the sound!Best wishes
Paul N
Nice one Paul and thanks for the update, the whole fun of this is sharing idea's for others to try out, beauty with diy if it doesn't suit simply revert things back
I've already got some BS250's ordered so looking forward to trying parallelling them up first. What sort of sound improvements did you notice after first trying the parallel fets?
I've already got some BS250's ordered so looking forward to trying parallelling them up first. What sort of sound improvements did you notice after first trying the parallel fets?
Hi Leo
Nice to hear from you! We must compare notes
At first I added just doubled up the FETs' using the BS250's from a second Legato I have. Got improved detail and naturalness. I find extra detail helps to identify a step in the right direction, but it's the extra naturalness that matters to me long term. I attend a lot of chamber concerts, and realistic timbe is ultimately everything to me
I then added a third FET to each pair (at first, I added the TP0610's the same way round, effectively "shorting" out the CFO with an internal protective diode until I noticed the different pinning). Not so sure I like this, which is, I admit, very odd. I'm still using the B1 buffers, the 2SC2547E's, and the FET CCS pair
I'll equip my second Legato board with some extra SIL sockets so I can play around with FET no. CCS, buffers etc. But not straight away- want to get some music in first!
Best wishes
Paul
Nice to hear from you! We must compare notes
At first I added just doubled up the FETs' using the BS250's from a second Legato I have. Got improved detail and naturalness. I find extra detail helps to identify a step in the right direction, but it's the extra naturalness that matters to me long term. I attend a lot of chamber concerts, and realistic timbe is ultimately everything to me
I then added a third FET to each pair (at first, I added the TP0610's the same way round, effectively "shorting" out the CFO with an internal protective diode until I noticed the different pinning
). Not so sure I like this, which is, I admit, very odd. I'm still using the B1 buffers, the 2SC2547E's, and the FET CCS pairI'll equip my second Legato board with some extra SIL sockets so I can play around with FET no. CCS, buffers etc. But not straight away- want to get some music in first!
Best wishes
Paul
Not exactly correct.
Let me explain myself better. I use the I/V purely balanced. So my thoughts and advice are often directed along those lines.
So when I advise people not to use the buffers, it only meant to be so if they have a highish impedance balanced load. *If* they are using the BAL/SE converter my experience is that they will get excellent results regardless of the absence or presence of the buffer. This does not mean they won't get a bit better results if they *do* use the buffer. Maybe they will. I am just not sure which they will prefer, so usually tend to suggest the simplest thing that will work.
The buffer I implemented is very well suited to the circuit as it is a class A emitter follower with a good amount of current.
So sure, I would not be surprised if some folks liked the Legato with the buffers in place better if they were only ever interested in the SE output.
To me that seems a dreadful waste of a particularly awesome balanced signal.
And I would need some objective convincing to (in terms of real results) to lean too heavily in favor of using the buffers when they are not strictly needed.
I am glad you found the same results as me when paralleling FETs. I trust people will enjoy the new Legato revision. But for those who have the current boards it is a very very easy thing to add a couple more FETs.
Cheers!
Russ
Last edited:
Hi barrows
He already did-
Legato pcb v 2.0.0 - Legato Discrete I/V Stage - Twisted Pear Audio Support
And sorry, I missed that you had already replaced R1-4 etc with 1W PRP's
Russ
which I was referring to when I said "Russ felt the buffer he provided wasn't beneficial anyway"- when using the provided Bal to SE converter
I'm not throwing away the rest of my system to accomodate one source component, so SE is it is! And it's not as if this is a minority position- outside of the US, the vast majority of amplifiers are SE
Thanks
Paul
He already did-
Legato pcb v 2.0.0 - Legato Discrete I/V Stage - Twisted Pear Audio Support
And sorry, I missed that you had already replaced R1-4 etc with 1W PRP's
Russ
which I was referring to when I said "Russ felt the buffer he provided wasn't beneficial anyway"- when using the provided Bal to SE converter
I'm not throwing away the rest of my system to accomodate one source component, so SE is it is! And it's not as if this is a minority position- outside of the US, the vast majority of amplifiers are SE
Thanks
Paul
Last edited:
Hi Leo
Nice to hear from you! We must compare notes
At first I added just doubled up the FETs' using the BS250's from a second Legato I have. Got improved detail and naturalness. I find extra detail helps to identify a step in the right direction, but it's the extra naturalness that matters to me long term. I attend a lot of chamber concerts, and realistic timbe is ultimately everything to me
I then added a third FET to each pair (at first, I added the TP0610's the same way round, effectively "shorting" out the CFO with an internal protective diode until I noticed the different pinning
I'll equip my second Legato board with some extra SIL sockets so I can play around with FET no. CCS, buffers etc. But not straight away- want to get some music in first!
Best wishes
Paul
Hi Paul,
Thanks, nice to hear from you too
Comparing notes would be great.How I have things at the moment for Legato is for the psu 3 x transformers , 3 x +/-v PH shunt regs . I altered their output from 15v to 12v.
The BP Muse coupling caps Russ and Brian provide with the Legato are excellent caps for lytics and probably fine for most, like some others I wanted to try alternatives, at first I tried these polyester (10uf 50v ) which are compact enough to be a good fit http://www.wima.com/EN/WIMA_MKS_2.pdf I'm also trying various other film caps, obviously downside with the MKP types is that they are physically much larger , I find the film caps to sound cleaner compared to lytics , just a personal thing of course.
I've also been trying out alternative values for the 15nf filter cap with interesting results (SQ wise) for filtering I personally like the Evox PFR's more than the Wima MKP types. Highest values they do is 10nf though so I've been trying that as well as lower values .
I'm looking forward to trying parelling the fets, soon as they arrive and I've tried them I'll post impressions . I've not tried the output buffer yet
Got to say though Legato is a very nice performer , its quite different to the IVYIII , its certainly worth trying both
Last edited:
Hi Leo
As it happens I have some 10u/50V Wima MKS2- I'll give them a try. Sometimes we all get too caught up in the whole bespoke capacitors for audio thing. The Nichicon Bipolars are superb electrolytic coupling caps; but even a (well made) polyester type like the Wima should best them
I also have IVY II and much- and increasingly- prefer the Legato
Paul
As it happens I have some 10u/50V Wima MKS2- I'll give them a try. Sometimes we all get too caught up in the whole bespoke capacitors for audio thing. The Nichicon Bipolars are superb electrolytic coupling caps; but even a (well made) polyester type like the Wima should best them
I also have IVY II and much- and increasingly- prefer the Legato
Paul
Hi Paul,
The main point is that *I* indeed would suggest that people try without the buffers first. The reason being I found no difference in performance between the two configurations. But, I never meant to imply that everyone should feel the same way.
I trust there are a few who may prefer the buffers be there. I personally do not. This is true for me regardless of whether the circuit is balanced or SE.
Now if someone were to show me objectively (measurably) that the SE output is enhanced by the buffers I would be very happy to change my advice.
I am just glad your enjoying it!
Cheers!
Russ
Last edited:
Russ,
Could you please describe the differences you heard when you paralleled the FETs? I am guessing it was significantly better given that you have designed new boards for it.
Also, is there much difference between having 2 FETs and 3 FETs in parallel?
Suresh
Hi,
I simulated legato2 with LTspice and i obtained a strange behavior:
with +12V/-12V of PS i get a second harmonic (of a 10KHz test signal) at -80dB, with +12V/-8V i get a second harmonic at -125dB (astonishing).
It seems that legato2 (and also legato - always from simulations) works better with asymmetrical PS.
Maybe i did some error on my simulation so i include the files to let you see and try.
Ciao
Andrea
I simulated legato2 with LTspice and i obtained a strange behavior:
with +12V/-12V of PS i get a second harmonic (of a 10KHz test signal) at -80dB, with +12V/-8V i get a second harmonic at -125dB (astonishing).
It seems that legato2 (and also legato - always from simulations) works better with asymmetrical PS.
Maybe i did some error on my simulation so i include the files to let you see and try.
Ciao
Andrea
Hi Andrea,
I was lucky enough to receive a Legato 2.0 board when I ordered from Twisted Pear. Been listening to it for a week and it sounds superb. Just did a FFT with 10kHz sine wave with Placid BP at +12/-12v and it looks ok to me
.See below.
Regards
Cornelius
<table style="width:194px;"><tr><td align="center" style="height:194px;background:url(http://picasaweb.google.com/s/c/transparent_album_background.gif) no-repeat left"><a href="http://picasaweb.google.com/vanwykca13/BuffIILegato20?feat=embedwebsite"><img src="http://lh4.ggpht.com/_87ClgeqViVA/TKzGFATYlEE/AAAAAAAAABA/gOx6-ozkhPA/s160-c/BuffIILegato20.jpg" width="160" height="160" style="margin:1px 0 0 4px;"></a></td></tr><tr><td style="text-align:center;font-family:arial,sans-serif;font-size:11px"><a href="http://picasaweb.google.com/vanwykca13/BuffIILegato20?feat=embedwebsite" style="color:#4D4D4D;font-weight:bold;text-decoration:none;">Buff II + Legato 2.0</a></td></tr></table>
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.