advertisement


Buffalo Summary to date please

cutting42

cutting42

Arrived at B4 Hacker Ergo
Hi All

So having diagnosed my Buffalo problem as a duff Trident 1v2 I got to thinking. What is the current state of the art with the Buffalo? How can I improve it?

I realise Tony's huge 3 part thread plus a million pages on DIY hold the answers but I really don't have the time or the wish to wade through all that lot.

Maybe a few of you who have tweaked Buffalos out there can advise on the best bang for buck configurations and tweaks that I can do to my Buffalo. I don't mind spending a few quid on it and don't tell him yet but I might ask Tony to build it for me :D

TIA
 
Ok I´ll start.

Improve the pre-regulation of the tridents.
When coming from a Placid HD, a Salas Shuntreg or Paul Hynes Series reg will improve on the TPA-regulators.
 
cutting42 said:
I don't mind spending a few quid on it and don't tell him yet but I might ask Tony to build it for me :D

TIA
Click to expand...


HA HA pay back!!

I can't see much wrong with the 'official' implementation TBH. It all works very nicely, upgrades to clock supplies etc like Tony has done do seem a little high maintenance and the SEN seems a bit noisy if musical and difficult to implement well.

Going for a dual mono build and getting rid of the buffers on the legato and powering each side independently with a decent shunt would seem to be the best bang for buck.

Of course now you have such good speakers you could go active like me, It's a lot more bits though but should reap rewards. I believe James would be able to export FIR filters from his LSP CAD design for the crossover so if you implement them with say a minisharc 3 DACs and 6 amps you'll have just about the best system money can buy!!!!

Stefan
 
Welcome back Gareth you have been away for to long.

Its hard for me to be definitive here as mine is still evolving.

First off you have to decide what to use for I/V conversion.

Legato or Ivy both need quite a bit of work, there are others of course.
Regarding SEN, mine is absolutely silent, I did have noise but that was just poor implementation, its a rather sublime converter much in line with nCores and Ergos, they all have impart little character you get whats recorded.
Hardest part of SEN is sourcing the FETs otherwise it's a deceptively simple circuit with very few components and no boutique parts required.
I just ordered SEN mains power supply prototype regulator boards.

What ever you do its all down to regulation, the Tridents can be improved upon by replacing with proper feedback regulators with out doubt.

All Buffalo feeds respond well to splitting supplies with separate transformers giving reduced ground plane noise. Stripping out the clock in this way is beneficial as not only do you get a better clock you take all the clock noise out of the Avcc circuits, probably number one one the regulator to do list.

Avcc feeds are a must they respond excellently.

If using spdif feeds then work is needed here particularly with regard to isolation.
I2S is far superior but again needs isolation and incredible attention to wiring detail which also applies to most of the DAC.

It is not just the feeds but the returns and ground planes, with a standard Buffalo most everything gets dumped into the same ground plane.

WaveIO is my recommendation for getting the data into the DAC, again it needs good regulation.

If using Arduino then again isolation is needed, in this case for the I2C data bus.

Balanced operation just makes so much sense when your squeezing the most out of the Sabre and SEN > nCore in balanced operation is sublime.

Bang for buck is probably a pair of fleas on clock and Avcc along with Salas regulation, give the Salas the Walt Jung / John Luckins CCS tweak.
First the clock as this masks the Avcc.

It can all be done with Fleas but modifying Fleas is a PITA, they all need significant amounts of bug building and this can lead to failures if they are not handled with delicacy.

I expect to have a complete set of new regulator boards running first quarter next year, no bug build, one board all the options and all the tweaks included, this sounds the ideal way forward as it just needs small transformers and rectifier boards adding and I will have those also.

To give you an idea of what a half sorted Buffalo can do, I can easily hear the difference between a quantiser with 6 bits and 9 bits, the difference is profound it needs no consideration the effect is immediate and obvious. Try it with yours and see what you hear, I expect no change.

Tony
 
Tony,

When you say "All Buffalo feeds respond well to splitting supplies with separate transformers giving reduced ground plane noise. Stripping out the clock in this way is beneficial as not only do you get a better clock you take all the clock noise out of the Avcc circuits, probably number one one the regulator to do list.
"
do you mean that the best way forward is to separate the supplies/transformers to the four on-board regs (and presumably the board supply itself?
 
Agh, sorry...any chance of a link to the "Walt Jung / John Luckins CCS tweak."

I'm trying to get my head around the BiB Salas builds at the moment and put a Mouser order in to complete some for Ian's FIFO project - yep, I bought into that Tony ;-)

I need super clean supplies for the clock and digital stages!
 
Pelham said:
Tony,

When you say "All Buffalo feeds respond well to splitting supplies with separate transformers giving reduced ground plane noise. Stripping out the clock in this way is beneficial as not only do you get a better clock you take all the clock noise out of the Avcc circuits, probably number one one the regulator to do list.
"
do you mean that the best way forward is to separate the supplies/transformers to the four on-board regs (and presumably the board supply itself?
Click to expand...

Yes the clock has its own dedicated transformer just like Martins / Andys original concept for the flea.
 
I've fiddled with Buffalo's since the dark ages of the ESS9008 24bit jobbie.

Tony's notes above are entirely valid.

Three areas to focus on:

1. Regulation/power. The DAC is moderately sensitive to power regulation and Trident V3 (current implementation) is a baseline, works well but is susceptible to quality of preregulation. Fancy Paul Hynes regs are sure to be great here, but the clock and analogue reference are highly susceptible to noise.

Using the Placid HD reg from Twisted Pear is a good starting point. I recently moved away from mains power to powering the Placid with a meaty Lipo and the gains are well worth it - much more relaxed and resolved music results. Big Lipos (5000 MAH) have internal resistance of a few milli-ohms and very low noise. Less operationally transparent of course (got to charge em up every 6 or 7 hours of running) but I'll live with that.

PH regs on AVCC and clock are probably areas for worthwhile gain, but you need to power the dac in the long interim between ordering from the overworked Mr Hynes and receiving your regs.

2. Input of digital signal. Despite the clever approach adopted by the ESS engineers to minimise jitter, the BUffalo benefits from a low jitter input of bits. Having tried a few routes, the WaveIO - as previously suggested by TOny - makes a very solid contribution on this front. A PFM flea to power the isolated I2S side of the circuit and a Placid HD or better to power the main XMOS side of the board will yield great results. (I will get around to trying battery power there also, maybe this weekend!)

3. I/V. Both the TPA circuits are good. The Legato benefits from a bit of judicious modding (filter caps primarily). I found a noticeable improvement moving from mains regulated power to raw Lipo power on the Legato. I use a 3 cell Lipo per rail.

Alternative I/v stages may yield improvements (Russ White has a new discrete stage which he has stated is an improvement on the available circuits, but for one reason or another has not released it). The SEN stage gets good press but I haven't got around to trying one yet.

I'm pretty happy with my current implementation though I am sure separating the dc feeds for the clock reg and AVCC would yield a little more goodness. It doesn't cover as much acreage as Tony's build, though, so I am clearly half-assed in my approach.

Mark
 
Pelham said:
Agh, sorry...any chance of a link to the "Walt Jung / John Luckins CCS tweak."

I'm trying to get my head around the BiB Salas builds at the moment and put a Mouser order in to complete some for Ian's FIFO project - yep, I bought into that Tony ;-)

I need super clean supplies for the clock and digital stages!
Click to expand...

My original Salas thread here

I can't find the cascode current source thread but the Walt Jung paper is here
 
My bestest configuration was the complete set of Paul Hynes regs (including his Trident replacement sets) to a dual-mono, single Legato Buff. II. USB only via Wave IO board, utilising the isolated I2S outputs & powered by a Flea, modified a la Mr Bemused.

It's a shame Paul's regulators are so difficult to get hold of these days; they made the most worthwhile improvement to sound quality, although I suspect the Salas regs will work just as well to power the Buff. board & Legato. I discovered a single Legato, bearing in mind I'm using single-ended outputs, sounded better than two.

I baled out when it came to the SEN, a bit beyond my capabilities. However, I'd really like to try transformers at some stage.

I eventually (shock,horror!) went over to the Dark Side & bought a Chord QBD76HDSD. Meanwhile, I made up two Buffalo DACS, single boards, one with four spdif inputs and volumite to be used as a digital preamp in a second system, the other as a spare USB-only DAC.
 
Suffolk Tony said:
It's a shame Paul's regulators are so difficult to get hold of these days; they made the most worthwhile improvement to sound quality, although I suspect the Salas regs will work just as well to power the Buff. board & Legato.
Click to expand...

Whilst I found the Salas very quiet it was no match for a high band width feedback regulator,
I am currently Salasless but will probabably return to using some Salas as pre-regulators.

Suffolk Tony said:
I discovered a single Legato, bearing in mind I'm using single-ended outputs, sounded better than two.
Click to expand...

If that was the TPA method of using two Legato and then the additional balanced to single ended converters then yes I am in complete agreement. The TPA method had far to many active devices in the signal path but it did sell a lot of hardware perhaps.
Using balanced from the Legato may well be a different story nCores were not invented back then :D

Suffolk Tony said:
I baled out when it came to the SEN, a bit beyond my capabilities.
Click to expand...

Sen is real easy once you have a kit of parts and there are not many just the damned FETs.
Mains supply for SEN is on my current to do list, prototype regulator boards are ordered.

Suffolk Tony said:
However, I'd really like to try transformers at some stage.
Click to expand...

I seem to remember you were to speak to your local transformer man, what happened to that avenue?

Suffolk Tony said:
I eventually (shock,horror!) went over to the Dark Side & bought a Chord QBD76HDSD. Meanwhile, I made up two Buffalo DACS, single boards, one with four spdif inputs and volumite to be used as a digital preamp in a second system, the other as a spare USB-only DAC.
Click to expand...

Yes that blasted Chord has killed your bodging :D
 
MarkW said:
The SEN stage gets good press but I haven't got around to trying one yet.
Click to expand...

Blimey your going to have a few batteries on board.
Well worth a go if your already comfortable with batteries, I used batteries on mine as it was the fastest and cheapest way to move to SEN but did find they needed some additional filtering.
Next stem is to try SEN on a linear main supply with a "Flea alike"
 
MarkW said:
Alternative I/v stages may yield improvements (Russ White has a new discrete stage which he has stated is an improvement on the available circuits, but for one reason or another has not released it). The SEN stage gets good press but I haven't got around to trying one yet.
Click to expand...

Yes looking forward to seeing that one, Patrick (SEN) is also dabbling with something at the end of the DITA SEN thread.
 
Bemused said:
Whilst I found the Salas very quiet it was no match for a high band width feedback regulator,
I am currently Salasless but will probabably return to using some Salas as pre-regulators.



If that was the TPA method of using two Legato and then the additional balanced to single ended converters then yes I am in complete agreement. The TPA method had far to many active devices in the signal path but it did sell a lot of hardware perhaps.
Using balanced from the Legato may well be a different story nCores were not invented back then :D



Sen is real easy once you have a kit of parts and there are not many just the damned FETs.
Mains supply for SEN is on my current to do list, prototype regulator boards are ordered.



I seem to remember you were to speak to your local transformer man, what happened to that avenue?



Yes that blasted Chord has killed your bodging :D
Click to expand...

I did speak to someone at Sowter but I didn't really understand what he was on about! As it's not far from me I really should get along there & talk to them directly. It's in areas such as this that my lack of knowledge regarding some fundamental stuff becomes a bit embarrassing!

Yep, the Chord's messed me up I'm afraid...:rolleyes:
 
Thanks for all the advice so far. I have also had a chat with Paul Hynes and he will be building me some regs for the Stereo Buff initially. I felt that would get me a decent result. I also have a couple of fleas around and will try getting the clock isolated.

What about the IVY III or should I return to the Legato III, what needs doing to it? Supplies, cap rolling, it is pretty std currently and I am single ended so the SEN seems pointless.

Next priority is the input. I have a media PC with an optical interface currently, sounds like the Wave IO is the thing to have rather than the spdif. Same question, do I get back to Paul for a supply for this?
 
On the matter IVY or Legato, I can´t say much. I just had the Legato III.
It responds quite well to cap changes. I had WIMA MKP10 instead of the stock WIMA MKS2(?) at the end in positions C1 to C4.
Some tried Panasonic ECHU SMD which apparently also give very good results. These are also good for the caps C19 to C22.

The PSSR of the legato is low, so good regulation is also beneficial. I had very nice results with PH-regs, but again, Salas Shunts should also improve on the standard Placid.

I used SEN single ended with mains supply, and this gave an improvement over the legato.

My WaveIO supply comes From PH (set to 5V), because Tony (Suffolk) reported best results with these over LM317 and Placid (?).
 
Well, close to 18 months later thanks mainly to an EXTREMELY slow Paul Hynes, the Buff is up and running again. It is running his PR3i3 5.25v psu then a full set of DAC regs replacing the Tridents and the AVCC board.

It has been a while but I am extremely happy with the sound just not the wait to get here. Very impressive improvement over the original specs with Bass especially making a welcome return.

Lots of listening to to then get back to finishing my Ergos
 
You must log in or register to reply here.


advertisement


Back
Top