• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Legato Tweakers Thread...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I figured I would start a thread for those interested in tuning/tweaking the parts selection on the Legato to their preferences. No disrespect to Russ and Brian's work here, as their parts selection is very good, but of course one of the pleasures of kit DIY is being able to select different parts to suit individual tastes. Here is what I am thinking about so far:
Output coupling caps: 4.7 uF, Zen polypropylene (going into 11 kohms per phase minimum) leads from these become output wiring.
Power decoupling caps: maybe upgrade to Elna Silmic II
Signal path resistors: perhaps Vishay Z-Foils ($ though...), or PRP ( my fav affordable resistors)
What about the analog filter stage caps? Any ideas for improvments here?
Would love to hear other's thoughts on tweaks.
 
OK, some...

folks have their Legatos now, lets hear some news. Here is what I am up to:
Balanced only, going to run from a single Placid BP. After having a look at the schematic I decided to eliminate as many steel leaded parts as I could, and do some other parts changes:
R5-8 will be "naked" Z-foils, the other .25 Watt resistors will be PRP. Caps, C9,10,13-15 will be Elna Silmic II (yes, one must get creative to fit them). C19-22 will be Amtrans. After searching for output coupling caps that will fit neatly in my chassis design, and (hopefully) be quite transparent, I have selected Mundorf Zn, 2.7 UF 100 V film/foil caps. The Zns are not cheap, but they cost nothing compared to a good interconnect. I am jumping the board cap mounts, and will wire the Zns to the terminal outputs and then directly to the XLRs.
I still have not decided if I am going to use the buffer stage or not. Russ advises not to with my partnering gear, and I know from a numbers perspective he is right, but I also like the idea of super low output impedance. Anyone have any feedback on whether or not the buffer stage is a good idea?
 
wonder how this would go with +/-9vdc, i'm running battery power in a portable buff II and my battery pack is about 23v, with the LDOs im using i'm down to a stable 8.8vdc per phase. at the top of charge its 23v p2p, but at the bottom its 18.3 (6 x 3v (per cell) is drained, but I cut off at 3.1v)

yeah the greenies will be the first to go, all small films will be evox/rifa and i'll probably run some VCAP CUTF as bypass caps in parallel to nichicon KZ or silmic II. at the moment i'm using a OPA1632 based opamp stage (not IVY) but would love to have a portable discrete IV, i'll use balanced only, but will need the buffer as i'm driving low impedance headphones directly. not that fussed about the AC coupling after getting so used to DC coupled sound. Maybe I could use a servo instead.

is there any benefit to using a pricier grade of BC550 for higher HFE in this circuit?
 
qusp,

It will work just fine with 9V rails. There are a couple changes you would want to make:

Use a BC560 instead of BS250. Change R5-R8 to 392R. Done! :)


HFE is not that critical in the application because of the CFP. Still it won't hurt anything at all to use higher HFE. Give it whirl. just keep in mind, there is a filter cap there that will swamp most of the effect of the higher HFE. It is much better to have lower noise than high HFE in this application.

I fully expected some would use external caps. Thats really the only reasonable way to use film caps that large. Otherwise you make a PCB that is taken up completely with a few caps, not a great solution.

Cheers!
Russ
 
Last edited:
thanks Russ, OK well, be expecting an order for PCBs from me shortly then. i've also been playing with burson discrete opamps (duals running 'balanced') but prefer a fully differential stage, the balanced opamp IV is really nice and due to the nature of good opamps with well matched components, DC coupling isnt such an issue.

nice to know I can leave the high HFE rated parts, because I can buy 100 of the one in the BOM and match them well, for less than 14 of the higher rated parts would cost, then I would have some left for other projects. I think I already have some 560 and 392R here, so thats handy.

the CuTF are only 0.1uf, so arent huge, i'll see how they go anyway, I may just end up bypassing the silmics with SMD PPS films under the board.
 
qusp,

If you wanna get tricky, i might throw in the idea of using an assymetrical power supply for your battery powered legato. The negative rail doesn't really do much other than set the biasing from what i can see.

If you went with a 4 cell + 2 cell setup (+12V / -6V) you should be able to get much better performance than a 9/9. You'd just need to change the biasing resistors to suit the lower voltage across them.

It wouldn't be that great if you still wanted to use the ballsie, or the 9/9 for something else though.

Maybe there's a good reason not to do this, but i can't see one off the top of my head.
 
Any opinions on Black Gate non-polar type N? I have never used them. The 16V 100uF version should fit nicely on the board.

If you are going to bother changing the output caps, go for some polypropylene films. As far as bipolar electrolytics go, the Nichicon Muse are very very good. Since they haven;t made Blackgates in years, you would likely end up with counterfeits anyway.
 
Ok well there is no cap at the output of the BAL/SE stage. So the impedance of your line stage won't really come into play. All you will care about is the impedance of the BAL/SE stage which is ~4.7K. 2uf will give you a corner frequency of 17hz. That would be the absolute lowest I would use. 4.7uf or 10uf would be better. You could even simply keep the 100uf Muse caps and just bypass them with 2uf film(even 1uf in this case). Then almost all the signal will be going through the film cap anyway because they will be in parallel and the signal will take the path of least impedance.
 
Last edited:
Member
Joined 2007
Paid Member
If you are going to bother changing the output caps, go for some polypropylene films. As far as bipolar electrolytics go, the Nichicon Muse are very very good. Since they haven;t made Blackgates in years, you would likely end up with counterfeits anyway.

Thanks Brian! I'll definitely try the stock caps before spending $$$ on anything exotic. When I was running tubes my faves were Auricaps, but they're gone now as well. [btw, Michael Percy apparently still has the 16v 100uF black gate in stock... hopefully not fakes!]
 
If you are going to bother changing the output caps, go for some polypropylene films. As far as bipolar electrolytics go, the Nichicon Muse are very very good. Since they haven;t made Blackgates in years, you would likely end up with counterfeits anyway.

Even if they are genuine, they will be old, at best from 2006. (Black Gate (capacitor) - Wikipedia, the free encyclopedia)

What sounds best, fresh Nichi's or old BG's ? (don't answer!)
 
Member
Joined 2007
Paid Member
When to toss?

General question to those of you who build alot of stuff: When do you decide to throw away formerly expensive passive components? I have a number of formerly good caps from old, gently used projects. Re-use or throw away? (e.g. 8-12 year-old Auricaps, KimberKaps, MIT multicaps, etc.) They seem so totally sealed that it's hard to imagine much chemical or physical degradation could occur within the dielectric! ...but I know so little when it comes to hard core 'lectronix! :hphones:
 
General question to those of you who build alot of stuff: When do you decide to throw away formerly expensive passive components? I have a number of formerly good caps from old, gently used projects. Re-use or throw away? (e.g. 8-12 year-old Auricaps, KimberKaps, MIT multicaps, etc.) They seem so totally sealed that it's hard to imagine much chemical or physical degradation could occur within the dielectric! ...but I know so little when it comes to hard core 'lectronix! :hphones:

Film caps are generally fine over time. It is possible for them to be damaged with usage, however.

http://www.illinoiscapacitor.com/pdf/Papers/storing_capacitors_film.pdf

Electrolytic caps dry out over time, which is bad. They can often be reconditioned for use, however.

Electolytic Capacitor Application Guide :: DuraCap International Inc. .::. 519-539-4891 (end of document)
 
HI Russ

qusp,

HFE is not that critical in the application because of the CFP. Still it won't hurt anything at all to use higher HFE. Give it whirl. just keep in mind, there is a filter cap there that will swamp most of the effect of the higher HFE. It is much better to have lower noise than high HFE in this application.

Russ

I don't understand "a filter cap there that will swamp most of the effect of the higher HFE". Do you mean the 1n rather than the 15n cap, and why is this? I thought the 1n cap was just to lower gain at at high frequencies and so increase stability, and would have little effect on hfe in the audio band. What am I missing :). And wouldn't a higher hfe device further lower the input impedance of the CFP?

Be gentle- I mainly work with valves!

Thanks

Paul
 
The 1nf cap is a form of compensation. It basically reduces the gain at high frequencies as you said. The main point is that the transconductance is already so high at audio frequencies that increasing the beta on the BJT is going to have little effect. Still there is absolutely no harm in using high beta BJTs, just be sure they remain low noise types. That is the most important attribute here.

I hope that clarifies things a bit. :)
 
Hi

OK, I've tried out a few ideas, and have come up with a couple of modifications to the already excellent Legato, which, to my ears, are very worthwhile.

Firstly replace the BC550B's with 2SC2547E. This has very low noise- about as good as it gets- and about double the hfe. It's also reputedly very linear when used in common base mode. Whatever the reason, it does the business. With BC550B fitted piano from CD was the best I've heard (piano on CD has always sounded wrong to me). With the 2SC2547E fitted, it's better still!

I matched the transistors for hfe, though the batch I have were probably close enough for this to be unnecessary. Note the pinning of these devices is bce, not cbe, so some lead bending and insulation is needed. The biggest problem is that-like all good things it seems- this transistor is no longer made, though it's not too hard to find. 2SC2240 is a near equivalent, though whether it sounds as good I don't know. BTW I also tried replacing the BS250 Mosfets with the 2SC2547E's complimentary pair, the 2SA970. This worked, but- as I've found in similar applications in the past- the combination is unstable, and generates program-related frying noises and clicking. Avoid!

Perhaps even more effective is fitting the current mirrors with constant current sources. To achieve this I replaced R19 and R22 with a CCS composed of cascaded BF245C and BF245A JFETS (see attached). A value of 68R for the resistor gave the required current- 2.7mA, and little adjustment was subsequently required to reset the input voltage to 1.70V (one half AVCC). Leave the circuit to stabilise for a few minutes before final adjustment.

Well, there you go. Apologies if you don’t hear what I hear! :)

Paul
 

Attachments

  • Cascaded FET CCS.gif
    Cascaded FET CCS.gif
    2.7 KB · Views: 969
Perhaps even more effective is fitting the current mirrors with constant current sources. l

If this were a single ended circuit you would be totally correct that adding CCS would help quite a bit. In fact with an SE DAC it will help significantly.

But in this case, with a truly balanced source, both the upper and the lower half of the circuit are already a constant current source/sink by virtue of the symmetrical nature of the circuit. In other words, the circuit is always in relative stasis. One side of the circuit is always counterbalanced by the other, so that the entire circuit is always in balance. Any impedance effect at one half is countered by an equal and opposite effect at the other. I had actually tried simply adding a cascode(in the form of a ccs), but I actually found this to reduce dynamic range far too much. Essentially what you have done is add a complex cascode.

I think you have probably been reading I/V threads that concentrate on single ended DACs.

As for the transistor choice, it is certainly a good part, and worth using, but not likely to yield astonishing results. I only say this because the noise is already vanishingly low. Still there are dozens of great transistors that will work very well. That certainly would be one. I would choose one with the same pin-out as the stock part. :cool:

Good work with your experiments. I hope my constructive criticism does not dampen your enthusiasm.

Cheers!
Russ
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.