HeadWize: DIY Workshop > New Cavalli Kumisa III plus Boards

Message [#40] posted on: 11-06-2006 12:28 PM CST (US).

Joined: Oct. 12, 2005
Locale: Germany
Total Posts: 260

@ dBel84
thanks for burning in your amp the hard way <big grin>

@runeight
as for the servo, why did you go as low as 100k with the feedback resistor ? Do you think I'll run into trouble if I changed the 10uF electrolytics to 4.7uF polyester box caps and increased the 100k resistor to 200k ?

@amb
may you add some alternative pads for a more flexible part selection ? I'd like to suggest 5 / 7.5 / 10 mm leadspacing pads for C7 and C16, additional 5mm pads for C20 and C21 and maybe additional 5mm pads for TO-220 rectifier diodes.

Dave_M

Message [#41] posted on: 11-06-2006 02:30 PM CST (US).

Joined: Jul. 1, 2006
Locale: England
Total Posts: 126

[Quoting dBel84]
awpagan - yes the servo goes back to input which is perhaps not optimal but still sounds incredible.

Hi, that is how it is with my prototype. But you also say you never tested my servo layout. Could you describe exactly where the servo output joins your circuit?

runeight, that 10uF NP cap is not going to be popular I think. 1uF would be better if possible then you can use a cap such as a WIMA box cap.

[Edited by Dave_M on 11-06-2006 at 02:38 PM.]

dBel84

Message [#42] posted on: 11-06-2006 02:43 PM CST (US).

Joined: May 5, 2006
Locale: Portland, OR
Total Posts: 425

Dave - I thought you had been working off ver2.0 , the one i built is 4.01 (attached)..dB

Attachment: C2229.gif

Dave_M

Message [#43] posted on: 11-06-2006 02:56 PM CST (US).

Joined: Jul. 1, 2006
Locale: England
Total Posts: 126

dBel84, that is the same as the one I have. Putting the servo back to the input was the only way I could get it to work.

runeight - going by past experience, I won't believe the servo will work until I see it working <smile> I think it is worth someone testing the new servo design before AMB has boards made.

runeight

Message [#44] posted on: 11-06-2006 08:01 PM CST (US).

Joined: Mar. 8, 2002
Locale: Austin, Texas
Total Posts: 1879

Hey Dave, nice to hear from you!!

I don't blame you for the hesitation on the servo. But, the sims are generally pretty good at telling this kind of behavior. Servo back to the input is a respectable, but not optimal solution.

The 10u/NP caps should not be too difficult to deal with. I tried to keep them small and inexpensive. 100u/NP caps would be definitely a problem, but my sense is that the 10u are pretty easy to get.

amb

Message [#45] posted on: 11-06-2006 09:28 PM CST (US).

Joined: Apr. 1, 2004
Locale: Sunnyvale, CA. USA
Total Posts: 4408

[Quoting runeight]
The 10u/NP caps should not be too difficult to deal with. I tried to keep them small and inexpensive. 100u/NP caps would be definitely a problem, but my sense is that the 10u are pretty easy to get.

Panasonic ECG series 10uF 50V (Digikey P1280-ND)
Nichicon UVP series 10uF 50V (Mouser 647-UVP1H100MED)

Both of these will fit the PCB layout.

[Edited by amb on 11-07-2006 at 01:14 AM.]

runeight

Message [#46] posted on: 11-07-2006 12:53 AM CST (US).

Joined: Mar. 8, 2002
Locale: Austin, Texas
Total Posts: 1879

[Quoting Steinchen]
@runeight
as for the servo, why did you go as low as 100k with the feedback resistor ? Do you think I'll run into trouble if I changed the 10uF electrolytics to 4.7uF polyester box caps and increased the 100k resistor to 200k ?

Steinchen, you can change increase the 100k to 300k to get better results, but the 10u has to stay with the circuit as recently drawn. However, if you really don't like the 10u/NP caps you can change out the servo components as follows:

Replace both 10u/NP with 1u
Replace 100k with 3M
Replace 10k with 100k

These new values will give roughly the same behavior as before.

Edit: Diagram attached.

[Edited by runeight on 11-07-2006 at 11:09 AM.]

Attachment: C2239.gif

amb

Message [#47] posted on: 11-07-2006 04:59 AM CST (US).

Joined: Apr. 1, 2004
Locale: Sunnyvale, CA. USA
Total Posts: 4408

Here is an updated PCB layout including the DC servo fix and Steinchen's suggestions.

[Edited by amb on 11-07-2006 at 07:17 AM.]

Attachment: C2235.png,C2236.jpg

Mister X

Message [#48] posted on: 11-07-2006 05:40 AM CST (US).

Joined: Jul. 23, 2003
Locale: Kalamazoo, Michigan
Total Posts: 606

More Suggestions:

OMG there is no LED! <big grin>
There is no apparent out ground.

Can you create more room around the ins and outs to allow for terminal blocks or molex headers?

Can you give us some larger through holes after the regulators for test points in the power supply section?
(these will come in handy when the inevitable troubleshooting threads appear)
It would also be nice if there were similar test point "holes" for adjusting the bias.
Along the same lines....
The leads on output resistors appear to be awfully close to exposed "tab" on the output transistors, given that proximity it seems it would be easy to short them with meter probe.

voodoochile

Member

Joined: Oct. 17, 2002
Locale: Concord, NH USA
Total Posts: 73

Message [#49] posted on: 11-07-2006 06:18 AM CST (US).

Nice looking boards, Ti; and glad to see you back, Alex. <smile>

dBel84

Message [#50] posted on: 11-07-2006 06:31 AM CST (US).

Joined: May 5, 2006
Locale: Portland, OR
Total Posts: 425

[Quoting Mister X]
The leads on output resistors appear to be awfully close to exposed "tab" on the output transistors, given that proximity it seems it would be easy to short them with meter probe.

I am glad you mentioned this, I have not yet shorted these contacts but it is a delicate thing to measure bias. My bigger concern was adding heatsinks and I eventually made my own from larger sinks which I cut down to be narrow enough to just fit the O?P transistors. Just another thing to be aware of - I did find some heatsinks from R/S that would work as is..dB

amb

Message [#51] posted on: 11-07-2006 06:55 AM CST (US).

Joined: Apr. 1, 2004
Locale: Sunnyvale, CA. USA
Total Posts: 4408

[Quoting Mister X]
OMG there is no LED!

The latest layout has an LED. <smile>

[Quote]
There is no apparent out ground.

It's intended that you connect the output jack ground directly to the star point. You could use a ring terminal for that, and screw the star point down to the case on a metal standoff, which would accomplish the grounding of the case too.

[Quote]
Can you create more room around the ins and outs to allow for terminal blocks or molex headers?

There isn't enough room for the Phoenix Contacts style of terminal blocks but there is enough room for Molex KK pin headers for the inputs and the transformer hookup, but the distance between the pins is 200 mils, so you could just snip the middle pin of a 3-pin header.

[Quote]
Can you give us some larger through holes after the regulators for test points in the power supply section?

There are already two vias labeled +15V and -15V near the center which could be used as test points. You can't actually solder a test point hook in there, but the via is a good "grab" point for a meter probe.

[Quote]
It would also be nice if there were similar test point "holes" for adjusting the bias.

Just use the output resistor leads. This board is too crowded for luxuries like that.

[Quote]
The leads on output resistors appear to be awfully close to exposed "tab" on the output transistors, given that proximity it seems it would be easy to short them with meter probe.

The output transistors are BD139/BD140 which are TO-225AA and the only metal surface is flush on the back side. There isn't a big metal tab like on the TO-220 devices, so there is no such risk.

[Edited by amb on 11-07-2006 at 06:59 AM.]

Nisbeth

Message [#52] posted on: 11-07-2006 11:14 AM CST (US).

Joined: Oct. 15, 2000
Locale: Denmark
Total Posts: 229

How big is the board?

/U.

dBel84

Message [#53] posted on: 11-07-2006 11:23 AM CST (US).

Joined: May 5, 2006
Locale: Portland, OR
Total Posts: 425

[Quoting amb]
The board layout is notable in its prominent star-ground near the center. The dimensions are 4.25" x 4.20" (10.8cm x 10.7cm), which is small enough to fit in a compact enclosure. The volume control pot is not board-mounted. You can mount the power transformer in the same case as the amp board (although I would recommend a good distance between the transformer and the board to avoid magnetic interference), or use an external box for the transformer.

from the first page of this thread..dB

cetoole

Message [#54] posted on: 11-07-2006 11:25 AM CST (US).

Joined: Sep. 14, 2005
Locale: Maryland, USA
Total Posts: 171

Couple suggestions/questions regarding the board layout. First, at 108mm wide, the board is just 8mm too wide to fit in a standard eurocard enclosure. It looks like it wouldnt be too hard to push it in a bit so that it fits. Second, why not move the star ground traces to the top of the board, since it would allow you to move several of the traces currently on the top of the board to the bottom, eliminating several vias.

vladco

Message [#55] posted on: 11-07-2006 12:15 PM CST (US).

Joined: N/A
Locale: N/A
Total Posts: N/A

Why we don’t use schottky diodes so we could get rid of extra caps? They are not that expensive.

AMB could you please put a few extra holes for lithics so we will have a flexibility with caps?
Vlad

Steinchen

Message [#56] posted on: 11-07-2006 05:51 PM CST (US).

Joined: Oct. 12, 2005
Locale: Germany
Total Posts: 260

vladco, why do you want to axe the snubber caps ? I'd even populate them with schottky diodes and if you don't like them just omit them, axing them wouldn't save much space anyway.

I support cetoole's suggestion to fit the board into eurocard size width, would make encasing much easier and cheaper.

btw, what about resistors across the inputs and ground ?

runeight

Message [#57] posted on: 11-07-2006 07:51 PM CST (US).

Joined: Mar. 8, 2002
Locale: Austin, Texas
Total Posts: 1879

cetoole, good idea on the board width. I suspect that there is enough room to take out 8mm and it could be that amb is already one step ahead of us.

vladco, I think that I agree with Steinchen that the caps don't take up much space and you can leave them out if you use fast rectifiers. Or in if you don't.

cetoole, good idea on the ground traces too. I'll take a look at this and chat with amb and dBel84.

Steinchen, I don't think we need resistors from input to ground. But, it would be a relatively easy addition as a "just in case" type of thing.

amb

Message [#58] posted on: 11-07-2006 08:33 PM CST (US).

Joined: Apr. 1, 2004
Locale: Sunnyvale, CA. USA
Total Posts: 4408

[Quoting runeight]
cetoole, good idea on the board width. I suspect that there is enough room to take out 8mm and it could be that amb is already one step ahead of us.

I am playing with this. It's not as easy as it seems. Reducing the board width by 0.25" doesn't seem like much, but fitting into a Hammond with its grooves on the sidewalls means that the parts must be moved quite a bit toward the center of the board. Things are already tight, and I don't want to make wholesale rearrangements or to compromise the sensible, direct traces we have now. So, no promises here, it might happen, but then it might not.

[Quote]
Steinchen, I don't think we need resistors from input to ground. But, it would be a relatively easy addition as a "just in case" type of thing.

Adding input resistors to ground is a good safety measure but this would definitely make it even harder to reduce board width.

amb

Message [#59] posted on: 11-08-2006 05:12 AM CST (US).

Joined: Apr. 1, 2004
Locale: Sunnyvale, CA. USA
Total Posts: 4408

OK, I tried various things to make the board narrower and in each case was not happy with it. So, I'm going to abandon that idea.

Instead of the Eurocard-sized Hammond 1455N series case, I suggest the 1455Q series. This is only about 0.75" wider than the 1455N, but will accommodate the board comfortably. The 1455Q160x could be used for a pcb-only solution (with outboard transformer), or the 1455Q220x could be used to house both the pcb and a toroidal power transformer.

If the 1455Q solution is acceptable, then the question is whether to keep the board size as-is and use standoffs to mount the board, or to widen the board a little to allow it to slide into the 1455Q's grooves?

Of course, this solution does not prevent anyone from using other cases. The mounting holes on the board will remain.

Btw, in my latest layout I added input resistors to ground, and changed the audio and transformer inputs pads to accept 100-mil Molex KK series headers. I'll defer posting it until I hear more opinions about the 1455Q enclosure.

[Edited by amb on 11-08-2006 at 05:13 AM.]

palchiu