The Ultimate DIY Part 2 ? The KGITSOJC

[luvdunhill]

What about the cmos fet swittches that ML uses? Could be made a lot smaller than relays. Other options would be the metal can switches or even reed relays.

[Kerry]

Also, if you use SMD parts, you'll have more space because you loose a lot of through holes and can better use both sides of the board. If you did this it might fit better. Also, you can stack a couple of boards (one for left and and one for right), thought this is not as optimal.

[kevin gilmore]

Not going to do the cmos switches thing. Measured them too many times, low level distortions

i don't like. Nelson had a better idea, but those parts seem impossible to find too.

Board is going to have to be re-designed. The 4PDT relays for switching the balanced

inputs are discontinued. Attenuator board now available in the attenuator board thread.

[luvdunhill]

... What parts were those that were recommended?

[luvdunhill]

This is rather interesting:

http://www.ne.jp/asahi/evo/amp/Hybrid_IC/HyIc.htm

[kevin gilmore]

At this point i will use any complementary j-fets that i can find

regardless of package. But no match for the n-channel above.

This is getting goofy.

[kevin gilmore]

I'm tired of waiting on linear systems, and have found a suitable part

for the front end. Bipolar, 2 x pnp and 2 x npn all on one substrate.

Reasonably priced and less than 2 nv/roothz noise.

input impedance won't be as high, but should otherwise work great.

sample parts on the way.

[Horio]

Oooo, nice! You mind telling us what part you are thinking of using, or is that a secret?

[kevin gilmore]

http://www.thatcorp.com/300-series_Matched_Transistor_Array_ICs.shtml

its the only thing that is not stupidly priced, and is ultra low noise, and in

stock at mouser. Otherwise a pair of the analog devices parts is something

like $40 per channel.

if i'm clever enough i can make the new board take both fets and this chip.

[kevin gilmore]

So the thing has morphed a bit. And i like it, there is a board rev 0 that needs a fair amount more work.

Its currently 6.5 x 3.6 inches, but i will be able to shrink it a bit.

The original blowtorch was a bit sensitive to load impedance, and did not like driving 50 ohm loads (100 ohm balanced)

So i fixed that. The original was unbalanced input only too. I also fixed that.

The result is a balanced dynalo style setup.

2 choices of feedback, super symmetry or the john curl feedback. The two sound a bit different.

Turns out super symmetry depending on how you view the "utility" of the patent is actually to feb 4, 2014

2 choices of power supply, +/-15 or +/-20 the later has slightly lower distortion.

2 choices of inputs, bipolar or fet. The bipolar results in a lower input impedance, but otherwise identical performance.

bipolar part available from mouser. Singles versions of the fets available wherever you can find them.

Wider range servo to accomodate the bipolar input parts.

For the unbalanced to balanced version, you have a choice of .5x or 1x (unbalanced/balanced) or 1x/2x

For the output amp version you have a choice of 5x or 10x

output amp drives 32 ohm headphones in balanced mode if that is what you want to do.

mates perfectly with the step attenuator board.

So there will be a 100% mouser sourced version.

Edited October 30, 2011 by kevin gilmore

[deepak]

Very nice, glad to see this is back on track

[Shaman]

Nice to see some action again in this thread! I have a pair of "digital attenuator" boards waiting for their matching KGITSOJC boards.

[Horio]

Great to hear you've had some time to work on this project again Kevin. I'm really looking forward to this one, and I could definitely use a excellent preamp.

Are you still considering incorporating a phono stage into the preamp as well? Either way, the more balanced inputs the better!

[kevin gilmore]

I was not planning on doing the phono stage.

Seems a bit silly as i have not played an lp in 16 years.

The input board will be all balanced inputs.

If you need an unbalanced input you can use the rca to xlr adapters.

[kevin gilmore]

http://gilmore.chem....du/multiamp.jpg

6.1 x 3.4 inches, 233 holes

buildable either super symmetry, which works out great for a gain of one, unbalanced and balanced to balanced converter

buildable with john curl jc2 style feedback which works great for gain of 5 or 10 for the output amplifier

buildable with the thatcorp quad bipolar input chip, or singles fets thermally glued together.

also drives headphones NO PROBLEM.

So a full preamp, would be 4 of these boards, plus 2 step attenuators (or balanced pots) and the input switching board.

the input board.

http://gilmore.chem.northwestern.edu/inputboard.jpg

Edited November 19, 2011 by kevin gilmore

[ujamerstand]

Looking great!

I think I've found the perfect controller for this project. A $35 computer from the Raspberry Pi foundation. Set to release some time in December. Runs linux, (yes, you can run fedora with LXDE) has an ethernet port, has 2x13 GPIO pins, some of which can be used to carry I2C and I2S signal. So it could be great as a music server if a dac is added. It'd be great to run a touch screen to it too.

[Kerry]

Looking really good

Thanks.

EDIT: Looks like a balanced dynalo. I've always wanted to do one this way. What transistors were you thinking of?

Edited November 20, 2011 by Kerry

[kevin gilmore]

yep, balanced dynalo, same setup as the balanced dynahi.

i've been listening to a lot of different configurations, including what has been

represented as an identical blowtorch clone with all the exact same parts and

values as the original. Blowtorch has almost 2db of unbalance between the

+ and - output, and it gets worse with low impedance load. Blowtorch also does

not drive headphones very well.

i did shrink the board to make it fit into a standard chassis.

http://gilmore.chem....inputboard2.jpg

Its now 12 x 4.5 inches and uses the same neutrik jacks as the T2.

Everything in stacks of 2 boards, hole thru the input board to

run the control from the bottom board.

still no p-channel fet samples as promised by linear systems

but i do have a clever way of using the singles fets on a 14 pin

dip header with one spreader bar with all 4 glued to it for

thermal stability.

The thatcorp parts are definitely very low noise.

Edited November 20, 2011 by kevin gilmore

[kevin gilmore]

so i've been playing with this for a couple of days. The unbalanced and balanced to balanced front end.

Its virtually impossible to hear the difference between this and my fully discrete version. To be fair, it does

not drive cables as well, and definitely has a higher output impedance, but it is about as simple as it gets.

Turns out the rowland criterion preamp uses virtually the same circuit.

The 50 ohm output resistors are definitely required if you want to drive 10+ feet of cable without

it oscillating.

The servo actually is required and i bumped the input impedance a bit to be more friendly to some

high impedance sources. (which evidently some others are not paying attention too)

http://gilmore.chem....u/ubaltobal.pdf

Something like a pair of lme49600's with your favorite choice of opamp in a closed loop configuration

as the output amplifiers after the volume control, and you have something super simple to build

and way good for cheap.

The discrete stuff is still better, especially the output amp, but the chips are damm close.

This is the technology texas instruments licenced from nelson pass.

Edited November 24, 2011 by kevin gilmore

[luvdunhill]

I used the same thing in my phono pre. I have to agree. I'm starting to think that phono pre is really one of the best places for these ICs. I still have some tweaking to do but I'm getting real close. Congrats though!

Have you looked that the THAT 1200/3/6?

Edit: one more question, I assume that this doesn't tolerate any DC on the input?

Edited November 24, 2011 by luvdunhill

[Shaman]

That's what I was going to suggest as well.

The THAT1200 series of balanced line receivers and the THAT1600 series of balanced line drivers seem very interesting.

I have all parts needed to breadboard a 1646 (based on the evaluation board BOM) which will be used as an unbalanced to balanced converter, but I still haven't got round to it.

Nice to see the progress on KGITSOJC!

[luvdunhill]

... Any chance you could stick this on your input selector board, before the volume control board, with perhaps the Cardas stacked RCA PCB mount connector? That would be a pretty slick deal.

Edited November 25, 2011 by luvdunhill

[kevin gilmore]

like this ??

http://gilmore.chem.northwestern.edu/inputboard3.jpg

I can still add the servo on the input, but probably not necessary.

Servo on the output takes care of all issues with DC. Why no one else

is doing servo's i just don't understand.

Still think that when you need unbalanced inputs, you use the neutrik converters.

This way you don't need extra relays and configuration for inputs, and you

can change on the fly.

The discrete output amplifier is still better, but not by much. The bang for the

buck on this one is way high.

[MASantos]

Is it possible to use it single ended with just one board?

[kevin gilmore]

Nope, not designed for that. I could however change out all the xlr jacks with stacked rca jacks

and then change the input amplifier. Then it would be a SE in only, SE out only thing.

If there really is interest in this, it would not be too hard as long as i can find a suitable

stacked rca jack.