advertisement


The art of Sen

Battery pack
Assuming
16 x series connected AAA NiMH battery, 1.2V 600mAh = 19.2 volt at say 30mA gives 20 hours use.
We need four sets of these = 64 cells in total spread over four banks.
Solder lug terminals I assume and wrapped in heat shrink.
Each bank about 40x40x45mm

So assuming say seven guys in a GB 448 cells.

Just a quick look in RS shows this @ £1.43 each + vat
I would think 20% discount is a safe bet so £88.00 a set of 64

Not looked at the charging circuit yet.

Just to say I have ran a pair of PP3 in a turntable motor supply since Teddy came up with the TeddyReg so thats some time ago and its fitted with a three position charge / off / run switch and it is still fine. Charger was a nasty diy implementation of a LM317 with a second 317 as current limit.
Patricks charger is on DIYA post 612 and even simpler.
Trickle charge at 30mA = 30hours ??
Standard charge at 60mA = 15 hours
Fast charge at 600mA = 1.2 hours
We could add an emergency boost charge...

For the Sen I would be looking at an automated charger probably using the mute / lock led or my Arduino.
For Arduino its analog inputs could monitor charge level and display on the screen.
 
cubeasic said:
Taking this approximation as a basis, on matched set would require about 32 Fets, so about 32 USD + shipping and import fees. And we would get 15 sets out of 500, maybe a little bit more...
Click to expand...
With 500 pcs I would expect >90% matching efficiency (from my experience with 600 pcs) or around 25 sets of 16.
Are you sure they are original Toshiba parts? Stay alert as fake FETs are becoming quite common.
 
Maybe I'm excited again. The XCEN seems to fit our criteria . Patrick certainly seems prolific! Can't wait to see how all this pans out. Just not sure I''ve got room for all the charging and battery paraphernalia at least e won't have to sort out the bal se anymore :)

Stefan
 
I am in the middle of sketching a PCB for the buffalo, which is a little bit tricky, to be honest. One has to work around the three pin via-connector, and all the big caps have to be on the outside of the PCB because they are larger than the 16 mm between IV-Board and Buffalo.

If I am right, here are the relevant schematics for the Buffalo SEN. Would it be best to include further things?

About the XCEN, it looks like as if it would be just for BAL-SE conversion, or is it a complete IV-stage?
 
I understand the sketch I put out up on post 55 is correct, the one you referred to does not have the stoppers. I have R40 value wrong should be same as R4.

One Vref, not the pair I showed and The Vout can share the common Agnd line so its just three terminals for Vout.
That gives seven terminals of which three are the Buffalo header connection.

All that bother with the Vref conditioning is not required, just a pair of 1K0 resistors and a cap, probably both caps but the Vref is most probably best fitted right on the AVCC regulator.

XCen is just a BAL/SE stage, I would leave that for now as we have a no cost BAL/SE option from Paul, plenty other options should we become bored. I think we just let the Xcen run its course.

Great your doing the boards.

My view is we use a pair of boards, left hand the the right hand being a mirror but had not thought it through much.
We need to accommodate both the Stereo and Dual mono guys.
One board would be better but may be tricky.

Yes stick the header connector right under the Buffalo and then stick most every thing else hanging out the side.
We could do with the DAC output to Jfet traces being as short as possible but the Jfets may not fit under the Buffalo. Moving the Buffalo DAC output headers to the extreme side of the Buffalo may help.
I will be able to give the Jfet heat sink dimensions in a couple of weeks.

Don't think we need extras yet, I was thinking a a basic implementation, we can easy etch some more boards if we need something extra?
On the other hand once the Jfets and heat sinks are mounted they won't want to come out easily so better first time round, fit once.

See what you come up with and post up some schematics and layouts then we will think some more.

I have an thought at the back of my head about the Vout Agnd going right back to the DAC, making it a star point perhaps, whose origin is way back at the Placid just to be radically correct. There may be some great gains by cleaning the Agnd of all the transient noise that it is currently subject to. Needs some thought here as it could be a free upgrade.

Tony
 
Hallo Tony,

I used your first sketch as orientation and already included the 100R gate stoppers.
Also - with my limited knowlege - I made the Iin-connection quite short. I guess one of the transistor-quads has to be under the buffalo, but height-wise, that shouldn´t be a problem. The heatsink-area from above is about 20x20mm, estimated from photos.

The total area is already about 10mm wider and 5mm longer than the Legato/Ivy size, I consider this not much of a problem.

I do not really know what I should do with the 220uF Silmic 2 at the AVCC Voltage divider. If I want to keep the board size, this has to be under the buffalo. Apparently there are some 6,3 Volt types which are only 12,5 mm high, so this should work, or is the voltage rating too tight? And: do you consider the 100pF/10R Iin-filter important? I sure do have space for it.

One thing about designing the board: I only use vectorworks for designing, this is maybe stupid and old-fashioned, but at least I know how to use it. So, I could only provide PDFs or CAD-filetypes, like .DXF or .VWX. But I hope, for the first sketches, this should be acceptable.

Two more things:
what do you think is best to connect the AVCC voltage? I would just create one connector on the board to connect a wire from the AVCC shunt-supplies to the IV-board.

Second: For the dual mono guys: wouldn´t it be acceptable to use just two of the boards next to each other? Like with the "two-legato solution"? You had two SEN channels per phase then and could add up the outputs with adjusted gain. Stupid idea? Of course you would need 8 floating supplies then.
Or could two channels of the SEN be connected via jumpers to have acces to 8 FETs per phase?
 
Just a thought, for those going the battery/charger route because of the claimed superiority (I'm a believer), doesn't it make sense to also power the ultra sensitive analogue voltage supply of the Sabre with a battery too? Little extra effort if setting up relay controlled charging for the SEN/CEN.
 
And for that matter, why use "16 x series connected AAA NiMH battery, 1.2V 600mAh"? Surely, there are better, less costly, less involved, higher capacity battery solutions?
 
cubeasic said:
Hallo Tony,

I used your first sketch as orientation and already included the 100R gate stoppers.
Also - with my limited knowlege - I made the Iin-connection quite short. I guess one of the transistor-quads has to be under the buffalo, but height-wise, that shouldn´t be a problem. The heatsink-area from above is about 20x20mm, estimated from photos.
Click to expand...

I would not worry about limited knowledge, either someone will put us right or we make something crappy and have to do it again.

For me I would like the Buffalo output header traces to be nice and short.

Be great if the heat sink fits under and be nice to know the exact size, perhaps I will have a look over at DIYA next week. We will have to pick the shortest route between height of sink under board versus heat sink to side of board but board vertical spacing smaller.

cubeasic said:
The total area is already about 10mm wider and 5mm longer than the Legato/Ivy size, I consider this not much of a problem.
Click to expand...

I imagined quite a lot wider, only 10mm sounds great, perhaps others can comment on there situation. No problem for me as mine is still nailed to a plank.

cubeasic said:
I do not really know what I should do with the 220uF Silmic 2 at the AVCC Voltage divider. If I want to keep the board size, this has to be under the buffalo. Apparently there are some 6,3 Volt types which are only 12,5 mm high, so this should work, or is the voltage rating too tight? And: do you consider the 100pF/10R Iin-filter important? I sure do have space for it.
Click to expand...

1/2 AVCC so 6v3 should be fine, but again right now I tend to think the Vref conditioning is best bug wired onto the AVCC regulator along with the dividing resistors but its great if you have room for it on the main board to give us options.

I would have thought its C1 and C2 that are the problem.

I did not 100pF/10R Iin-filter important but it maybe :)
Not sure if the Buffalo already has a filter on it, better at source. Great if you can fit it in though.

cubeasic said:
One thing about designing the board: I only use vectorworks for designing, this is maybe stupid and old-fashioned, but at least I know how to use it. So, I could only provide PDFs or CAD-filetypes, like .DXF or .VWX. But I hope, for the first sketches, this should be acceptable.
Click to expand...

So vectorworks is like autocad? I can use Autocadd 2011, love it, does a lot better job dimensioning than Eagle, but eagle is great for routing.
Perhaps you could mail me the drawing in dxf.

I can mail you the Eagle if you wish, if worse comes to worse I can use your hard work to duplicate in Eagle.

Hell if push really did come to shove we could draw this out by hand with a marker pen onto the copper.

We only need a pdf for the etching.


cubeasic said:
Two more things:
what do you think is best to connect the AVCC voltage? I would just create one connector on the board to connect a wire from the AVCC shunt-supplies to the IV-board.
Click to expand...

Yes with my earlier thought about provision for connecting the one Agnd to source, maybe its the AVCC regulator but as I said it would be good to have the option to try taking it back to the Placid or LCLCLC if your system is more advanced. That ground could well be getting modulated by the entire Buffalo system.
Sort of how we wire up the HackerNAPs.

cubeasic said:
...
Second: For the dual mono guys: wouldn´t it be acceptable to use just two of the boards next to each other? Like with the "two-legato solution"? You had two SEN channels per phase then and could add up the outputs with adjusted gain. Stupid idea? Of course you would need 8 floating supplies then.
Or could two channels of the SEN be connected via jumpers to have acces to 8 FETs per phase?
Click to expand...

Not sure if anymore Jfets would work, there will be impedance and capacitance issues. This would definitly be a #II version.
Again eight supplies would be a #II version?

Two channels of the SEN be connected via jumpers to have access to 8 FETs per phase but the impedance would crash ?
Maybe that's OK for the Sabre but again it a #II version?

For now the dual mono will have the layout architecture of dual but lose four converters.

But two boards for the dual mono sounds good, we would just half populate each of the boards.

Hope this is helping
 
Sparts155 said:
Just a thought, for those going the battery/charger route because of the claimed superiority (I'm a believer), doesn't it make sense to also power the ultra sensitive analogue voltage supply of the Sabre with a battery too? Little extra effort if setting up relay controlled charging for the SEN/CEN.
Click to expand...

Three cells 1V2 is 3V6 low charge depleted, maybe nearly 4V++ when charged, would the Sabre be happy.
But yes its a thought, though for another day.
 
flashgo said:
And for that matter, why use "16 x series connected AAA NiMH battery, 1.2V 600mAh"? Surely, there are better, less costly, less involved, higher capacity battery solutions?
Click to expand...

Capacity is OK i thought at perhaps twenty hours, NiCad are no go as are SLA ?

9 volt NiMH have no capacity but even at reduced capacity they are the same price.

AAA are tiny, about a quid each, very common, have good capacity and can have huge capacity if you want.

Please show something better:)
 
Redwine have got an interesting LFP battery solution. Even if they don't sell the bits, you'd think someone does.

I remember an old PFM thread about the best batteries for Hii-Fi. Damned if I can find it.

A simple switch so that either the battery charges or it powers your circuit would do the job.

[Edit: Some LFP bits here. Seems more expensive than NiMH, but there are nice charger PCBs available. I just don't relish the thought of 16 AAAs, any of which can drop dead when least expected.]
 
Bemused said:
One Vref, not the pair I showed and The Vout can share the common Agnd line so its just three terminals for Vout.
That gives seven terminals of which three are the Buffalo header connection.

All that bother with the Vref conditioning is not required, just a pair of 1K0 resistors and a cap, probably both caps but the Vref is most probably best fitted right on the AVCC regulator.
Click to expand...

I would suggest to derive Vref separately for each channel. I know that Patrick's schematics show only one derivation, but he was not aware that Buffalo has two AVCC supplies. Probably also best to load both AVCC supplies the same way.

Also I would suggest that you keep the Vref trim-pot in the circuit. Paul N and I did not need it, but our FETs were also extremely closely matched (all 16 within 1% Idss). I would build according to Patrick's circuit first as it is way more easy to take the pot away if not needed than adding it later. The pot serve the important purpose of zeroing out DC offset (if any) and as it only see a very low voltage (<2V) and low current (<2mA) it seems very unlikely to me that it can/will add any remotely relevant noise.
 
Nic I maybe confusing you with the Vref. we were discussing just ONE channel.
So one Vref.
The other chanel will be a duplicate of everything.
Thanks for the advice on the trimmer
 
1Some LFP bits here. Seems more expensive than NiMH, but there are nice charger PCBs available. I just don't relish the thought of 16 AAAs, any of which can drop dead when least expected.

They look good, not looked at battery tech for a while. Do the individual cells not fail in the same way as NiMh ones do?
 
orangeart said:
1Some LFP bits here. Seems more expensive than NiMH, but there are nice charger PCBs available. I just don't relish the thought of 16 AAAs, any of which can drop dead when least expected.

They look good, not looked at battery tech for a while. Do the individual cells not fail in the same way as NiMh ones do?
Click to expand...

Good point about the batteries, but to be honest you don’t hear of the battery boys being plagued by cell failure.
Could use conventional rechargeable un tagged AAA in a 4 gang holder, cost with holder would be about the same as tagged without.
That way should the project fail you would have a few conventional AAA cells to use elsewhere.
It still does not look unduly expensive compared to a Legato and Placid BP
 
NiMh batteries from a reputable maker are very reliable & its pretty unlikely one's going to suddenly fail, rather they'll all gradually lose their ability to hold charge. You could probably get away with just replacing the whole set every couple of years or so.

Anyway, are batteries strictly necessary here? Just seems a bit of a fag to me.
 
Suffolk Tony said:
NiMh batteries from a reputable maker are very reliable & its pretty unlikely one's going to suddenly fail, rather they'll all gradually lose their ability to hold charge.
Click to expand...

My experience of them is very different, I've worked as a photographer for a very long time and have to recharge my AA batts daily (well actually I cycle through 5 or 6 sets) they are all Ansmann professional batteries charged on a top of the line intelligent Ansmann charger. I replace each set when one in the set becomes faulty, about every 2 years on average.

I once spoke with a battery technician at Ansmann about this, he said it was normal behaviour and although the MTF of NiMh's was longer than that of NiCad, the failure of the former tended toward complete and sudden failiure rather than a dying slope.

Stefan
 
You must log in or register to reply here.


advertisement


Back
Top