advertisement


The art of Sen

With those you can hardly go wrong:)
For perfectionism you could swap Q2/Q4 (Left C, Second Q) and Q2/Q4 (Right C, First Q) to meet Patricks Q1+Q2 = Q3+Q4 rule to perfection.
 
Thanks for the advice Nic

Also recieved your boards and heatsinks this afternoon.

Many thanks.
 
Hi Nic, Tony

For tips about fitting the FETs on the PCB see here and here.
Note that only 4 of the 12 FET legs actually have to be soldered to the PCB.
Click to expand...

TBH I've never understood what the hassle is, or why you can get away with soldering just four legs :) If it's relevant, I used 100R 0805 packaged SMD resistors, soldered directly under the board.

I just inserted the FETs slightly- all legs (need to preform a bit). Then fitted the heatsinks temporarily, and then pushed all the FETs and heatsinks down to equal height, a little less than 1cm above the board. The splaying of the legs holds them firm. Then solder. Then remove heatsinks, apply glue, and refit. Some of the glue will inevitably contaminate the leads, so make sure its a non-electrically conducting epoxy! I used this-

http://uk.farnell.com/fischer-elektronik/wlk-5/adhesive-thermal-conductive/dp/1211723

At the weekend, I refitted the necessary socketry to my DAC- in its new case- to allow a connection to a DSD bitstream (from SACD) I extracted from a cheap Denon player I modified (I don't seem to need TP Teleporters- I get away with a bout 0.5 m of CAT5). Blimey! Hitherto I'd found higher res formats underwhelming- not any more. SEN makes a world of difference

I also had a quick listen to a SE output taken from just the + phase out from SEN. Extraneous-noise free (unlike e.g. Legato used this way), and no obvious problems- it sounded excellent! Of course it's now 6 dB down re my BAL output, making a fair comparison difficult- but I will try.

I was looking back over an old TP DIYAudio thread, and found conformation- from a designer of the ESS9018, Dustin- that it would be OK to do this, and to just connect he - ve phase out of the DAC to ground-

http://www.diyaudio.com/forums/digi...-reference-dac-8-channel-154.html#post2224863

http://www.diyaudio.com/forums/digi...-reference-dac-8-channel-155.html#post2225524

Not sure how this would compare sonically to my ideal dummy load (a then unused SEN phase)

Best wishes

Paul N
 
Thanks for all the help and tips.
The gate stoppers are nothing special I guess, metal film 0805 ±100ppm/°C, 0.25 W or should I be looking for something exotic.
I have to buy them in so may as well get what is right.
 
Penny just dropped on taking just the Sabre DACa output single ended style, just one board would do a stereo pair.
 
Hi Tony

Pleased to help. Believe me, I hassled Patrick and Nic a hell of a lot before I got moving- you're getting there much quicker!

And I might want payback one day-I'm entirely clueless about programming/Arduino/coding.....

Paul
 
You will not find Arduino a problem Nic I am sure.
I only played with it for a few weeks and managed to get up to speed, got it working and improved the implementation. Just follow the tips in the thread to avoid hardware that I know does not work.

Something I will return to but for know it ticks the boxes.
Compared to DIY Hifi the Arduino is easy as there is not really a prerequisite for under pinning knowledge.
It is the lack underpinning knowledge that causes my stumbling in Hifi but even though, I get things working eventually 
More than happy top help anyone with Arduino of course.

Tony
 
orangeart said:
That sounds perfect if it works, and half the components and half hours of the batteries :)

Stefan
Click to expand...

Yes quite true but seems a little scary to leave either the Sabre DACb output floating or grounded and to leave half the kit unused goes against the grain :)

If the DACb was floating the current would pabably fall to within the range of a standard Fleas capabilty ;)
 
Bemused said:
If the DACb was floating the current would pabably fall to within the range of a standard Fleas capabilty ;)
Click to expand...
Are the Fleas series or shunt regs? What is the drop-out voltage?
I will be implementing batteries for all 5 buffalo supplies so I am looking for something a bit more "green" than my current shunt regs.
 
Fleas are series regulators, info on Martins site here

I have bodged them quite a bit, my latest here

I expect to try ALWSR for my AVCC supplies but have Flea variants in place of the Tridents.
Drop out is not too bad for higher supplies but you still need quite a head of supply even when down at 1v2 or 3v3

Circuit is fed with gyrator and 7812 regulator, could be dropped to 7810, maybe 7808 as absolute minimum. This creates the standing voltage for the AD797 error amp and With battery maybe the 78xx and gyrator could be omitted.

Standard Flea drop out is
1v5 across the capacitance multiplier
2v drop-out for the 7812
12V for the 7812 output plus 2v offset from the LED in the ground leg.
So 17v6 is minimum feed.

My CCS Flea variant put the 7812 reference leg on true ground so would be a little gain.

Tony
 
Being a bit of a wholly I decided to find out how the Sen actually works which involved finding out how a Jfet works, yes I did not know.
I found this great link
 
My original Jfets from post 100

Afraid I am not so smart with spead sheet presentation :)


qUADORIG.jpg


Rearanged folloing Nics advice
With those you can hardly go wrong
For perfectionism you could swap Q2/Q4 (Left C, Second Q) and Q2/Q4 (Right C, First Q) to meet Patricks Q1+Q2 = Q3+Q4 rule to perfection.
Click to expand...

qUADNOW.jpg
 
Measuring/matching FET Idss with 4 significant digits goes a bit beyond what is reproducible (although one can pretty close). You have well below 1% variation in all 4 quads (and between phases) and my guess is that DC-offset will be non-existent:)
 
Thanks Nic, being overly cautious probably.
Also though posting the results maybe of help to others.

Cant help but think that when they are glued in they aren't going to want to come out.

Had a dry run on fitting the fets to the sinks and board and it's no problem.

Just wondering either to solder all 12 legs to the board or bug wire with just four legs to the board. Bug wiring is a little harder but does leave the option of pulling the assembly from the board to fit to a future board.

Waiting for batteries, SMT capacitors, turned pin board stackers and glue now.

Tony
 
I made a start

Fitted the 100R gate stoppers under the board, I used 0805 size and its a little tight resulting in flux falling down the gate via so had to be blown out. It may be worth fitting a smaller size say 0603 but 0805 is about as small as I like to go.
I used RS 679-0807

I plan to bug wire the Jfet assemblies so only five of the twelve legs will need to be fitted to the board, however both the gate vias you will probably fill with solder need to be used.
Bug wiring reduces the leg count on the board and hence I think there will be more chance of pulling the quads should the need arise, for another board for instance. Remember the fets are glues into the heat sink.

I fitted a turned pin socket right in between the 100Rs for the Vref from the AVCC module.

The R1 / R2 dividing resistors are fitted on the output. 680K on the upper R1 and 470K on the lower R2.
I used Metal Film 1%
I am not quite sure I have the correct values as the seem to be intended to set the floating battery supply symmetry and I would expect the values to be identical.
I think there may be more going on with these resistors than meets my simple eye. I noticed that the output is shifted by about 3v3 from symmetry and our AVCC is 3v3.
I also used R1 as 680K not 681K as in the DIYA thread.

So far as I can tell thats the underside of the board done.

I have one set of fets glued up and will be cured tomorrow.

senreverse.jpg


003-8.jpg
 
You must log in or register to reply here.


advertisement


Back
Top