The latest amp schematics are in post #92
The latest PCB is in post #158

Group buy has ended

For information on obtaining CK2III boards please go to AMB's Website

Photos of the final group buy board: #185

Thanks to amb for making these boards available.

Introduction

Several of you have written to me over the last month asking if I would return to headwize. I would like to thank all of you for your email and messages. Since I am a fan of this forum and appreciate what cmoy does to make it possible, I am going to take this opportunity to introduce the new Cavalli Kumisa III amplifier.

Much of this work was done by awpagan, DaveM, dbel84, and me on another thread. Dbel84 and I finished the work offline. I designed a PCB for the new amp and dbel84 did a prototype on a homemade, two-sided board. I will let him post with pictures and comments on the amp's audio.

AMB (www.amb.org), who has contributed to this forum in so many ways and who supplies boards and support for a number other amps, has agreed to supply and support the new CKIII boards. He has taken our original design and, with a few mods, translated it to Eagle format. He will post images and board availability information in a follow up post.

I have simulated this amp extensively and believe that it will rival or better most, maybe all, other amps in its class. It uses an all discrete design and should be relatively inexpensive to build.

The board has both channels and an integrated power supply section on a single board. It will need a transformer, volume control, jacks, enclosure, etc. to complete it.

Schematics are attached.

If you buy boards from amb I will support any questions/problems through this forum as best I can with the time that I have.

Edit: There was an error in the drawing of the servo. Schematic has been replaced.

[Edited by runeight on 01-05-2007 at 08:46 PM.]

There is no availability date on boards yet, and I would like to conduct a straw poll on how much interest is there for them. So, if you would like to build a CKIII amp, speak up here and be counted. That would help me determine the number of boards to order, at least initially. The more boards ordered, the more economical each would be.

Below is a preliminary PCB layout of the CKIII amp. There are still a couple of minor adjustments I would like to make, but it's pretty close to "final" form. I am also including a 3D rendering of the board with most parts populated (some are not exactly right, but that's the way it comes from Eagle3D), so you could get a better idea what the board looks like.

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. The recommended transformer is a dual 15V secondary unit rated at least 200mA per secondary (6VA total), or two separate single 15V secondary transformers (3VA each). A center-tapped transformer cannot be used with this layout. Note that this amp is not intended to be "portable" and has no provision for battery power. The output transistors do not need heat-sinking, but if you wish, you may bend the leads a little and bolt them to a case wall. Onboard heatsinks are used for the voltage regulator ICs.

I will do some independent spice simulations of this circuit in the near future and report back on the results.

[Edited by amb on 11-05-2006 at 06:47 PM.]

At what bias level is the output stage intended to be set ? Since the output transistors do not need heat-sinking I assume something like 20mA or 25mA ? (though I'll heatsink them anyway and try levels of around 50mA )

---

[Quoting Steinchen]

count me in!

At what bias level is the output stage intended to be set ? Since the output transistors do not need heat-sinking I assume something like 20mA or 25mA ? (though I'll heatsink them anyway and try levels of around 50mA )

---

The bias point is intended to be 30-40mA, but with the bias resistors as shown you can set it from about 15ma to 60mA. If you reduce the fixed bias resistor to 2k instead of 2k2 you can get to Benny's original 70mA, but dbel84 didn't really need to get that high to get good performance.

If you run at more than 50mA idle per channel you might consider increasing the VA of the transformer. A 400mA secondary would be better.

[Edited by runeight on 11-03-2006 at 09:19 AM.]

[Edited by dBel84 on 11-03-2006 at 03:05 PM.]

I guess I'll have to ask amb if he would start the marketplace thread since he will be supplying the boards.

[Edited by batman on 11-03-2006 at 03:09 PM.]

---

[Quoting amb]

I am not taking any orders at the moment. This is just an interest check.. do you still want it to go to the marketplace?

---

Hi amb, yes I think it should still go in the Marketplace b/c it's still about commerce and b/c you'll probably end up drawing names from the poll for sales.

The key features of the amp are that it is totally discrete and that it uses no NFB. All of the opamp based amps, even the ones that use discrete output buffers, employ feedback to control the gain of the opamp and to zero the offset of the buffer. The PPA amp (a really great headphone amp) puts the buffer in the feedback loop of the opamp.

This amp's topology is exceedingly simple with a pair of complementary input devices driving a current mirror that drives the output stage. Idle currents are kept high enough to get high speed and, therefore, really good transient response. Benny had thought the amp through pretty carefully on his original design.

The amp also uses the interesting feature of complementary voltage sources to pull the emitters of the input devices either up or down as appropriate. These voltage sources have come in really handy for fixing the servo.

To improve the amp, we made a few changes:

1. We replaced the Darlington output stage with CFPs. CFPs tend to have lower distortion than the Darlington arrangement. They also have fewer thermal problems because it's the driver devices that mostly control the bias current.
2. We used more commonly available bjts for the small signal devices (BC550/BC560)
3. We changed the servo design so that it uses the bases of the voltage sources as the control point for zeroing the offset. Doing this gives the servo more leverage over the offset while still keeping it out of the signal path. We also made the servo stable against low freq oscillation.

In my simulations the amp shows exceedingly low distortion when driving many tens of mA into 32R. This without any NFB. Reality will not be this good, but these low numbers indicate a very high performing amp.

And since the transient response is so good (see Benny's article), the high freqs and other transient audio behavior should be crystal clear.

But, the other important feature of this amp is that it is very simple. There are not a lot of components between the input jack and the headphones. This simplicity and elegant design make this, IMHO, a really cool, high performing amplifier that won't cost a whole lot of $$$. I guess this explains why I took an interest in it when I read the Kumisa thread and then went back to read Benny's article again.

I will be interested to hear comments from those of you who build one when you finally get boards. Which I see we cannot even talk about any further on this thread without serious punishment.

[Edited by runeight on 11-03-2006 at 07:00 PM.]

The graph goes from 1Hz to 1MHz. You can see that the response is within 1db from 1Hz to 1MHz. Benny's original measurements indicate that this response is not too far from actual behavior.

[Edited by runeight on 11-03-2006 at 05:15 PM.]

---

[Quoting cmoy]

Hi amb, yes I think it should still go in the Marketplace b/c it's still about commerce and b/c you'll probably end up drawing names from the poll for sales.

---

OK, the straw poll/interest check thread has been moved to the marketplace.

This thread will now primarily be for technical discussion, build support, Q/A and other comments related to the CKIII amplifier.

---

[Quote]

The recommended transformer is a dual 15V secondary unit rated at least 200mA per secondary (6VA total), or two separate single 15V secondary transformers (3VA each). A center-tapped transformer cannot be used with this layout.

---

---

[Quoting tomb]

Is there a difference between an xfmr with two secondaries and a CT xfmr wired for two separate secondaries?

---

I hope the illustrations make it all clear.

For the CKIII, you must use a dual secondary transformer or two separate single-secondary transformers because of the double bridge rectifiers. The two secondaries must be completely independent.

[Edited by amb on 11-04-2006 at 07:45 AM.]