The Multi Amp aka Dynalo Mk2

[GrindingThud]

Thanks! I think I'm going to attempt a 20V (21V) build and try to use the same PS from the KSA5.

The boards are small enough I think it will package well.

Which resistors get changed out?

on +-15V power supplies it will do about 30vpp balanced

on +/-20v power supplies it will do about 40vpp balanced

 

on +/-15v power supplies its 200ma per channel.

would need to adjust the bias resistors to reduce the current

for 20v.

 

works out to at least 2.5 watts rms pure class A into 50 ohms

[kevin gilmore]

http://gilmore.chem.northwestern.edu/kgdynalobalssproduction.pdf

 

r38 and r39

 

on my version its 255 ohms i think

 

200 ohms should be about right for 20V

[GrindingThud]

Thanks!

http://gilmore.chem.northwestern.edu/kgdynalobalssproduction.pdf

 

r38 and r39

 

on my version its 255 ohms i think

 

200 ohms should be about right for 20V

[GrindingThud]

I was thinking three output jacks, one balanced and two SE.

The board will always output balanced regardless of the input. Not really intended for SE but you could just leave the - leg not connected for SE.

[Pars]

On my Rotel RCD991, in implementing the discrete I/V stage, I had thought about the SE to BAL conversion, since it is single-ended to the output, where it is converted to balanced via an opamp. Per Filburt's recommendation, I had switched this out to a heavily bypassed AD8599 or something like that. I read with interest the input stage on this amp, doing SE->BAL conversion using the THAT340. Do you think that doing a small PCB to implement this would be an improvement over the opamp?

 

I wanted to preserve a) the output muting relays, and 'b) the balanced output. Even though I don't currently use them, I would like to keep them implemented for resale value and more importantly to preserve the output quality in SE mode.

 

The cetoole I/V stage I have built up has dc servos. Would the THAT340 implementation be pretty much 0V DC offset, or does that depend upon it being inside the servo loop?

Edited August 20, 2014 by Pars

[kevin gilmore]

The way the servo's is currently done results in +/-1.5mv of dc output.

If what you want to do is the se to bal conversion you want the

specifically designed thing i did instead of this. And build it with fets

to keep a high impedance.

[Pars]

What I want to do is do the SE->BAL conversion with the minimum sonic impact on the SE signal (and the BAL obviously, but I don't use balanced currently). What is this specifically designed thing BTW? I've probably seen it at some point, but I'm getting older...

[mypasswordis]

[kevin gilmore]

There are a number of parts to the complete preamp/everything solution.

 

One is this

http://gilmore.chem.northwestern.edu/baseboardbal3.jpg

(low res for now)

 

Which is the balanced input switch board. For unbalanced you use

the neutrik rca to xlr converters.

 

Next is the unbalanced/balanced to balanced converter

http://gilmore.chem.northwestern.edu/ubaltobal2.jpg

 

although there is a new rev of this board.  you build it with the

jfets because the that340 has significant bias currents.

Now here is the thing about this part, its completely NO feedback

ultra low distortion (<.001%) extremely fast, and gain is controllable

by one resistor. (very similar to ayre kx-r only lower distortion and otherwise

better)

 

Next is the smt step attenuator

http://gilmore.chem.northwestern.edu/attenuator5.jpg

256 steps, .5 db per step, two for balanced which includes

a balance control

 

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

Finally is the multiamp as the output board, and also unbal/bal headphone

board

 

All the pieces put together comprise a seriously kickass preamp/headphone

thing.

 

Verified recently with amb/linuxworks lcduino and standard code.

i made a change in the input board to get this to happen.

 

But i really want a production version with touchscreen, and wifi

so if anyone wants to step up and help...

 

use any or all of the pieces seperately as needed

 

its sweet i tell ya

Edited August 21, 2014 by kevin gilmore

[StrawHat]

Seems like a great project!

 

Do the input board and the attenuator work with the TI PCF8574 or only the NXP will do like the AMB version?

 

NXP discontinued the PDIP-16 version some time ago...

[kevin gilmore]

edit:   all of my chips are ti made A versions

work fine

Edited August 21, 2014 by kevin gilmore

[mypasswordis]

If no one else wants to do it (someone please do!), I could start looking into the microcontroller stuff in a couple weeks.

 

Anyone want to sell me a pair of KG attenuator boards for testing, either populated or unpopulated? Also is 3.2" touchscreen big enough?

[nopants]

I think Kerry has a few attenuator boards left

[mypasswordis]

Thanks for the lead! I'll shoot him an email later tonight

[cetoole]

I have a bunch too that I could probably dig out when I get home if that doesn't work.

Edited August 22, 2014 by cetoole

[Kerry]

I've still got a few. Already in touch.

Edited August 22, 2014 by Kerry

[Pars]

Next is the unbalanced/balanced to balanced converter

http://gilmore.chem.northwestern.edu/ubaltobal2.jpg

 

although there is a new rev of this board.  you build it with the

jfets because the that340 has significant bias currents.

Now here is the thing about this part, its completely NO feedback

ultra low distortion (<.001%) extremely fast, and gain is controllable

by one resistor. (very similar to ayre kx-r only lower distortion and otherwise

better)

 

So you build it up with the 2 P-ch and 2 N-ch around the 14-pin chip (which I assume is where the THAT340 would go)? Also, this appears to be a mono board? I'll look thru this thread as I assume there are schematics somewhere. About what is the physical size of the PCB?

 

Thanks.

Edited August 22, 2014 by Pars

[kevin gilmore]

there is a picture of the ubaltobal board with the jfets.

will post board size later

[GrindingThud]

For headamp operation, is it recommended to put a relay delay on the output or is it stable enough during startup to not need that?

[kevin gilmore]

here is the ubaltobal board with the fets

http://gilmore.chem.northwestern.edu/dsc_0597.jpg

you can do the same thing with the fets to the multiamp and

then you can have a higher input impedance

and you can do the same thing to the ss dynahi.

Now that the fets are available thru diyaudio in small

quantities, this is an option again.

 

the boards are 5.47 x 3.02 inches

 

to get it much smaller everything would (and could be) surface mount

but the multiamp does do about 3 watts pure class A and the output

transistors dissipate a total of 4 watts, so if smt they would have to be

heatsinked

 

I tested turnon/turnoff thump and there just is not any

Edited August 22, 2014 by kevin gilmore

[GrindingThud]

I'm almost done with my Mouser BOM and have a couple questions. I notice in the schematic and the silkscreen file C2 and C5 are 5pf but in the photo you've used 10. I'm assuming the board was an earlier build and the latest is 5pf in the posted files.

I plan to use Cornell Dubilier micas there: http://www.mouser.com/Search/ProductDetail.aspx?R=CD17CD050DO3virtualkey59850000virtualkey598-CD17CD050DO3

I'm also looking for what type of caps to use for the .1uf and 1uf units surrounding the servo.....looks like MLCC or maybe a tantalum for the 1uf?

Thanks in advance.

this is the production version

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

Edited August 23, 2014 by GrindingThud

[Aive]

For the dynahi build I'm doing I'm basing BOM off the silkscreen as I think that detail takes precedent.

Also in re to the 0.1 uf caps I went with ceramic disc type based on other BOMs I've seen for KG amps. MLCC type is probably fine though as I've seen that specd before too.

Can't remember what I specd for the 1uf...

[kevin gilmore]

sometimes mouser is out of stock on some caps, so I switch to in stock items.

very little real difference between ceramic and mlcc

[mypasswordis]

I have a bunch too that I could probably dig out when I get home if that doesn't work.

 

Thanks anyway, Colin!

 

Kerry just hand delivered a pair of boards to me, what a super nice guy. Now I have attenuator boards and Multi Amp boards, and can breadboard the SE to bal converter. I have an idea of what microprocessor kit to use, and will probably stick to the 4D systems touchscreen that Max had some success with.

 

The funny thing about all this is, I have no need for a preamp since I don't have any speakers.

[kevin gilmore]

kerry's multiamp boards are a bit different than the production boards.