[00940]

Hi all,
 
since it's a bit difficult to get the original parts for the dynalo, I've redrawn a pcb, using 2SJ74/2SK170 at input and bc550c/bc560c for the rest.
 
The pcb is inspired by the one provided by Kevin Gilmore in the original article, with the addition of some local decoupling caps. It's single sided (red tracks are jumpers) and avoid small tracks (most are 56mil), to ease home-etching. Dimensions are 6*10 cm.
 
So, what do you think ? What could be improved ?
 

 

 

[stixx]

Hi,
when you are looking for "original" devices still, try here -->http://www.kessler-electronic.de/Halbleiter/Transistoren/2_SA/2SA1015_i7_440_0.htm
I don't know though whether they are any good...
 
 
PS. I am building a Dynahi with the original boards at the moment. Basically just need to case it up. I guess there is now way of adding your mentioned
extra decoupling on that one??

 

[00940]

You're right, the original devices can be found. Reichelt still has those too. However, it's very difficult to find transistors made by Toshiba. Most on the market are now made by CDIL and they're really not as good. Furthermore, it's difficult to get transistors from the same hfe range. BC550C are more easily sourced from a good manufacturer such as ON semi.
 
For the Dynahi, is it this pcb: http://www.djgardner.com/headphone/gilmore/dynahi/Dynahi_top.JPG ? If it is, there is already some decoupling on board.

 

[Avro_Arrow]

 
The positive rail for the servo takes an awfully long route around the outside

of the board and no decoupling when it gets there...
Me, personally, I would flood the ground all up the middle rather than
the skinny little traces leading back from the decoupling caps.
 
Other than that, looks good for a one sider.

 

[00940]

The long + rail is an heritage of Gilmore's board. It's true that it is long. Decoupling rail to rail is possible, rail to gnd would be difficult.
 
Flooding the centre isn't a problem. I'd expect the differences to be pretty minimal but it will save etchant 

 
What could also be done now that I think of it is to add RC filtering for the input stage. Easily jumpered if not desired.
 
That'd give this:
 

 

[Avro_Arrow]

 
That looks pretty do-able now.

Should be easy for toner transfer or photo etching.
Good, thick traces and good space between them.
Lots of holes to drill though...I'd recommend a carbide
drill bit in a drill press (thats what I used when I did my
own boards). I got mine at Drill Bit City.
 
Good luck with it!

 

[00940]

Thanks.
 
A drill press sure helps a lot. Until now, I've used cheap HSS drills and an heavy old press. It's rather slow but it drills straight. Works ok and I don't do that many pcb so they last me a bit. I'm seriously considering those carbide bits from Bungard though: http://www.reichelt.de/PCB-Drills/BOHRER-0-8/3/index.html?&ACTION=3&LA=2&ARTICLE=33810&GROUPID=582&artnr=BOHRER+0%2C8  Almost everything will fit in a 0.8mm hole and I can use a cheap HSS set for the rest.

 

[Avro_Arrow]

Those look like nice bits.
I used to burn through the HSS bits quickly.
The Carbide bits seem to last forever and stay
sharp longer. The only caveat is they are very
stiff and snap easily if used in a hand drill.

 

[stixx]

 For the Dynahi, is it this pcb: http://www.djgardner.com/headphone/gilmore/dynahi/Dynahi_top.JPG ? If it is, there is already some decoupling on board.

 
Yes, that's the one...
Thank you!

 

[00940]

While I'm at it...  what do you consider an optimal grounding scheme for the return ground from the headphones jack?
 
- two wires back to the A points (twisted with the respective output wire) ?
- a single wire back to B ?
- other ?
 
 

 

 

[Avro_Arrow]

option #1, two wires, but return them to the power supply.

 

[Avro_Arrow]

  option #1, two wires, but return them to the power supply.

The answer above is what happens when you are walking by the computer, see a message you want to answer
and start typing without fully engaging the brain...
 
Here is a better answer.
 
Think of it like this.
An amplifier is just a device the controls the amount of voltage/current that flows from the power supply rails,
through your headphones and back to the power supply zero volt point. There is no real reason why the headphone
ground needs to return to the amplifier at all when it's real destination is the power supply.
 
Short answer:
One thick wire from the jack to the power supply ground point.

 

[00940]

  The answer above is what happens when you are walking by the computer, see a message you want to answer
and start typing without fully engaging the brain...

 
I know the feeling too well. Looking back at my post, I cannot find the rationale to put the "a" points where they are

 
 

 An amplifier is just a device the controls the amount of voltage/current that flows from the power supply rails, through your headphones and back to the power supply zero volt point. There is no real reason why the headphone
ground needs to return to the amplifier at all when it's real destination is the power supply.

 
Ok, we agree on the overall understanding. But shouldn't we add the aim to keep wires carrying a current and its return close-by ? A second aim would be to avoid mixing wires with load return, signal ground and decoupling ground. It would seem to bring us to this scheme:
 

 
- Input/feedback ground goes to the PS by its own wire
- The return wire follows the output lead and then the supply wires.
- Decoupling caps have a dedicated return to the PS.
 
I'm lazy and didn't draw the wiring of the second board... because it bring us back to the original question: one or two return wire ? If one, we cannot keep all loops as close as they are.
 
But I might be too influenced by speakers amp design, where currents are much higher and loops control matter more.

 

[Avro_Arrow]

In low frequency audio, the signal and it's return do not try and follow
each other like at higher frequency. Minimizing the loop length is
more desirable than having the signal wire and it's return in close
proximity. The left and right signal wire can go straight to the
jack and the return wire can go straight to the power supply.

 

[00940]

I finally etched and drilled those pcb. Etching went smoothly (thx Bungard) but I broke a pair of drills in the process of course.

 
 
I've also done a pcb for a dual "Pooge" regulator (left pcb) to team the boards with.
 

 
The Pooge board (red is gnd). The negative version is exactly the same as the positive, just replacing bc550 by bc560 and flipping the led/caps (or two positive regs can be used with two windings):
 

 
 
 
Once the amps done and tested, I'll upload the boards (slightly modified, I've spotted a possible improvement or two).